TI-89 -- Not impressed
Christopher R. Carlen
I had considered buying a TI-89, thinking that the symbolics might be
useful, even though for many years I have been disinclined toward the
calculator CAS because I have been used to Mathematica, and had been
suspicious that the calculator CAS would be too feeble.
Well, the TI-89 confirmed my suspicions, by failing some simple tests
that really seemed like it should be able to do.
Particularly,
Integrate[ t*e^(-s*t), t, 0, infinity ] results in "undef", though a
proper integral with a variable upper limit seems to evaluate ok.
and Solve[ a*x^3+b*x^2+c*x+d=0, x ] fails. I didn't try solving a cubic
with numbers for the parameters, which I'd really have vomited if it
couldn't do. I would grant the thing the right to fail at solving the
general quartic, which Mathtematica can do easy, for reasons of memory
limitations. But the thing should be able to manage spitting out the
general cubic solution. Is this too much to ask of the TI-89? Can an
HP solve the general cubic?
Anyway, I am still very unimpressed, and maybe I just won't bother
buying one.
I haven't considered HP because I fear the thing will be way too slow
for my tastes, and the screen resolution is lowsy compared to TI. I
also wanted to run the mini-spreadsheet app on the TI-89 for some
portable data management efforts, and for which I figured the 160x100
screen might just be adequate. I still wish to do this. Maybe I'll
pick up a used one for $99, which might be palatable instead of $149 for
a weak calculator. That price is just too much for what it does. A PDA
might be better off, but I want to use the thing during school tests as
well.
Might I be persuaded differently about HPs? I really don't like the
"dead-endedness" of the HP calculator line. The world really needs a
new innovative and powerful calculator to be created. I'd do it if I
had any business sense, but that just isn't my thing.
I wonder how I could get my hands on an HP to fool with it? Are there
any PC emulators that fully implement the features of a 48 or 49? Or do
you need at least a ROM image, which I suspect I can't get without
owning the calculator?
Thanks for comments.
--
____________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
crc...@sandia.gov
Maxwell
> I wonder how I could get my hands on an HP to fool with it? Are there
> any PC emulators that fully implement the features of a 48 or 49? Or do
> you need at least a ROM image, which I suspect I can't get without
> owning the calculator?
Yes, you can get a HP48/49 emulator for PC and yes, you need a ROM image and
no, you do not need to own a calculator.
bye,
Max.
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
*** Usenet.com - The #1 Usenet Newsgroup Service on The Planet! ***
http://www.usenet.com
Unlimited Download - 19 Seperate Servers - 90,000 groups - Uncensored
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Chris Carlen
> Eres muy pretencioso tratando de resolver una ecuación cúbica
> simbolica.
> Fuera del software de matemática, no conozco ningún programa de PC
> que me de la solución simbólica de una ecuación cúbica
>
> Mas info en
> www.geocities.com/jaimezacalcs/hp-versus-ti.htm
Huh?
--
_____________________
Christopher R. Carlen
cr...@earthlink.net
Suse 7.3 Linux 2.4.10
J
> Eres muy pretencioso tratando de resolver una ecuación cúbica
> simbolica.
> Fuera del software de matemática, no conozco ningún programa de PC
> que me de la solución simbólica de una ecuación cúbica
Hi,
Could you write in english, please ?
Best Regards
J
Bhuvanesh
> Integrate[ t*e^(-s*t), t, 0, infinity ] results in "undef", though a
> proper integral with a variable upper limit seems to evaluate ok.
You need to do
Integrate(t*e^(-s*t), t, 0, infinity)|s>0
> and Solve[ a*x^3+b*x^2+c*x+d=0, x ] fails. I didn't try solving a cubic
> with numbers for the parameters, which I'd really have vomited if it
> couldn't do. I would grant the thing the right to fail at solving the
> general quartic, which Mathtematica can do easy, for reasons of memory
> limitations. But the thing should be able to manage spitting out the
> general cubic solution. Is this too much to ask of the TI-89? Can an
> HP solve the general cubic?
I doubt the HP49G can solve the general cubic either. Can someone check this?
--
Bhuvanesh
ed
You are very pretencioso trying to solve a simbolica cubical equation.
Outside the software of mathematical, I do not know any program PC that me
of the symbolic solution of a cubical equation.
"Chris Carlen" <cr...@BOGUS.earthlink.net> wrote in message
news:3D62F6D9...@BOGUS.earthlink.net...
Arnaud Amiel
A simple program could be written to do this and even Quartic, I had
done this on my HP48 + Erable. It was a bit slow, and frankly useless.
The biggest trouble was to simplify cubic roots.
Unfortunatelly, my RAM card seems to get regularly corrupted and I
lost the program and as I though it was useless, I didn't try to
rewrite it.
Arnaud
Nick Karagiaouroglou
Hi all!
No it can't. Except of course for numerical coefficients or symbolical
coeficients when the factorization is "easy" like for example the
expanded form of (x-a)*(x-b)*(x-c), for which the HP49G finds
symbolical solutions.
Greetings,
Nick.
Christopher R. Carlen
> No it can't. Except of course for numerical coefficients or symbolical
> coeficients when the factorization is "easy" like for example the
> expanded form of (x-a)*(x-b)*(x-c), for which the HP49G finds
> symbolical solutions.
>
> Greetings,
> Nick.
Well, I guess I have to agree with JaiMezaCalcs that it is a bit much to
expect of a small calculator. But the day when I can run Mathematica on
a pocket sized PC is rapidly approaching (just to expensive now), so
perhaps it isn't unreasonable to expect a next generation calculator to
do this.
Darn, we really need some more innovation in the calculator world. I
don't think TI is going to meet my expectations.
Steen Schmidt
this?
It can't out of the box. SymbToolz v2.0 can (was it ever released). SOLVE2
of that library does it easily (returns the general answer in approximately
0.8 seconds).
Regards
Steen
Christopher R. Carlen
Wow! Not that it is a very useful thing to do, but a good "acid-test".
I have found Symbtoolz 1.1 at hpcalc.org, but not 2.0. Might it ever be
released?
Thanks.
Steen Schmidt
> released?
Unfortunately not.
Regards
Steen
Steen Schmidt
>
> Integrate(t*e^(-s*t), t, 0, infinity)|s>0
The same restriction is necessary on the HP49G: 's>0' ASSUME.
TI89 returns '1/s^2' in a bit less than a second, while the HP spends around
47 seconds on the same thing.
Regards
Steen
Christopher R. Carlen
Why? How did you ever see the 2.0 version? Are you one of the HP
developers?
Bummer that the 2.0 isn't available. :-(
Bhuvanesh
Along those lines, the cubic() and quartic() routines from the tip
list can find exact solutions for cubics and quartics, respectively
(for the TI-89/92+/V200).
--
Bhuvanesh
Bhuvanesh
> Darn, we really need some more innovation in the calculator world. I
> don't think TI is going to meet my expectations.
Not even with third-party add-ons?
--
Bhuvanesh
Chris Carlen
>>Integrate(t*e^(-s*t), t, 0, infinity)|s>0
> The same restriction is necessary on the HP49G: 's>0' ASSUME.
> TI89 returns '1/s^2' in a bit less than a second, while the HP spends around
> 47 seconds on the same thing.
Eek. Why so slow? This is what concerns me about this calculator. I
have been looking through the manuals, and the commands are often
cryptic, the syntax is hard to figure out, there isn't much detail about
what is returned from a lot of operations, like solving a numeric
differential eq., where does the data go? Is it only plotted or can I
do other things to it numerically?
(The same question I wonder about with the TI, though.)
But the depth of functions that are useful to an EE like Laplace and
inverse Laplace transforms, very good. But do they take a minute to
complete all the time?
Chris Carlen
> Along those lines, the cubic() and quartic() routines from the tip
> list can find exact solutions for cubics and quartics, respectively
> (for the TI-89/92+/V200).
>
> --
> Bhuvanesh
What's the "tip list?"
Chris Carlen
>>Darn, we really need some more innovation in the calculator world. I
>>don't think TI is going to meet my expectations.
>
> Not even with third-party add-ons?
There are two problems: for the price, the TI-89, 92+, and 200
calculators should have at least either the same or better CAS and
numeric capabilities than the HP, or at least more CPU power, memory,
and display res. Admittedly, the TI displays are pretty good already.
But I have personally purchased all the chips to build a low-end Palm,
Motorola Dragonball 16MHz and 33MHz varieties, memory chips to have
increments of 1MB flash and 4MB RAMs, and the only thing I haven't got
yet is a display. For large quantities, these things can be obtained
cheap enough to build a 33MHz calculator with QVGA resolution for about
the price of a TI-92+.
But TI is wallowing at about 1/3 of this CPU speed, and memory could
certainly be increased 4-fold. These improvements would allow more
power to be packed into the CAS.
The second problem is that the HP is much more powerful mathematically,
yet my reading of the manual makes me a bit displeased as to how easy it
seems to learn it. The TI user interface seems much easier, but I could
be wrong here because I haven't much experience with either calculators,
only TI-83s.
Stephen
and its not obvious (maybe its not even possible) to move data between them.
Whats more Apps aren't easily written and generally cost, eg spreadsheet is
another $29us.
On the HP this doesn't even rate a mention as the whole HP system is you
have object and functions and all objects are available to all the
functions. New functions and libraries are easily created and become an
extension of the OS. So its easy to do multiple operations involving CAS,
symbolic, graphing etc on the same data.
This does mean that on the HP there are often dozens of different ways to do
something which can be confusing. For example yesterday I had a nasty
little problem that involved 3 rather simple equations combining to give a
rather complex loading diagram on a support structure. I was going to
manipulate it symbolically or numerically but before I started trying to
work out what I needed to do I plotted it to give me a feel for it. That
was enough to give me the locations where I needed results and then I
realized for the accuracy I needed I didn't need to solve it just get some
areas between some points. A couple of points and clicks and the result
were there on the stack, tagged as Areas, ready for some simple arithmetic
to scale them up to the final figures. Who cares if the TI could have given
me exact figure 10 times faster my 49 gave me figure that were within 1% or
2% quickly and easily. It mightn't allways be the fastest but damn the
HP49's flexible.
Stephen.N
"Chris Carlen" <cr...@BOGUS.earthlink.net> wrote in message
news:3D643E3F...@BOGUS.earthlink.net...
Dr. Klaus Graichen
Integrate(t*e^(-s*t), t, 0, infinity)|s>0,
it is possible only by means of the TI-89 and TI-92+, respectively.
Klaus Graichen
grai...@mvtat.tu-freiberg.de
"Steen Schmidt" <ssch...@nospam.dk> schrieb im Newsbeitrag
news:7ZR89.10126$ww6.1...@news010.worldonline.dk...
J.Manrique Lopez de la Fuente
aconsejable que uses el inglés para que la mayoría pueda entenderte.
Pelase, comp.sys.hp48 is an international group. I sugest you to use
English, to make your post understandable by everybody...
And about you comparasion.. You haven't ASSUME s>0 in the hp49, so it
is logical that it gives you "unsigned inf error".. as I told you in
es.comp.sistemas.hp48
Best regards,
J.Manrique
Calculators Users Group from Gijón (Spain)
#1077 HPCC Member
Jai...@hotmail.com (JaiMezaCalcs) wrote in message news:<d7793e28.02082...@posting.google.com>...
> ======================================================================
> Comparaciones calculo simbólico entre la TI89/92p/Voyage200 y la HP49
>
> Nota :
> integrate es el símbolo de la integrar(s)
> infinite es el símbolo infinito(00)
>
> TI89/92p/Voyage200
>
> integrate(t*e^(-s*t),t,0,infinite)|s>0 definicion de Laplace
> solution: 1/s^2
>
> HP49
>
> integrate(t,0,t*e^(-s*t),t)
> solution: “Unsigned inf error”
>
> Nota : el comando evaluar no es tan poderoso como en la TI89, en la hp49 no se puede
> integrate(t*e^(-s*t),t,0,infinit)|s>0 no acepta >
>
> TI89/92p/Voyage200
>
> solve(x^3-3*m^2*x+2*m^3=0,x)
> solution: x=-2*m or x=m
>
> TI89/92p/Voyage200
>
> factor(x^3-3*m^2*x+2*m^3)
> solution: (x-m)^2*(x+2*m) donde x=-2*m or x=m
>
> New TIvoyage200 in:
>
> www.geocities.com/jaimezacalcs/image/tivoyage200.gif
>
> www.geocities.com/jaimezacalcs/ti/pc/new-ti-voyage200.zip
>
> www.geocities.com/jaimezacalcs/hp-versus-ti.htm
>
> Imágenes mejoradas para EmuVirtualTI en:
> www.geocities.com/jaimezacalcs/ti/pc/skinti89-92plus.htm
>
> ======================================================================
Frank Travis
plug-in application cards) is Calcpro website www.calcpro.com. I
bought one of my HP 48GX calculators (and many plug-in applications
cards/manuals) from them. I have had good dealings with its manager,
Paul Nelson. Good luck
Christopher R. Carlen
> By the way: The brand-new TI-Voyage 200 cannot solve the integral
> Integrate(t*e^(-s*t), t, 0, infinity)|s>0,
> it is possible only by means of the TI-89 and TI-92+, respectively.
>
> Klaus Graichen
Are you serious? What is that calculator smoking? Can anyone else
verify this?
This is the kind of thing that worries me with TI, that the calculators
are being geared toward the purely textbook math curriculum of high
school, and basic college math. But not providing any advanced
capabilities to do serious engineering or scientific calculations
(serious for a handheld device, that is.)
Very disturbing. I suspect I'll get an HP-49g and later buy a pocket PC
type thing (provided that I can run Linux on it and not pay a MS tax),
then run something like Maxima or maybe even Mathematica.
Veli-Pekka Nousiainen
news:3D644572...@sandia.gov...
> Steen Schmidt wrote:
> >>I have found Symbtoolz 1.1 at hpcalc.org, but not 2.0. Might it ever be
> >>released?
> >
> >
> > Unfortunately not.
> >
> > Regards
> > Steen
>
> Why? How did you ever see the 2.0 version? Are you one of the HP
> developers?
No, but Steen is the *author* of the SymbToolz 1.1 & 2.0
and I guess that he was so pissed off by the HP ACO dissolve AND
distroying the future of the 3rd parties software development
that he still don't like to develope anything for HP calculators
BUT I surely would like to see a SymbTools 2.0
Please, Steen, say that I'm wrong :-(
> Bummer that the 2.0 isn't available. :-(
Ditto !
VPN
Veli-Pekka Nousiainen
X
> But TI is wallowing at about 1/3 of this CPU speed, and memory could
> certainly be increased 4-fold. These improvements would allow more
> power to be packed into the CAS.
unless the OS is written as a pure 32-bit system.
> The second problem is that the HP is much more powerful mathematically,
> yet my reading of the manual makes me a bit displeased as to how easy it
> seems to learn it. The TI user interface seems much easier, but I could
> be wrong here because I haven't much experience with either calculators,
> only TI-83s.
When you buy the 49G, get also the Urroz Books for the 49G.
http://www.engineering.usu.edu/cee/faculty/gurro/myBooks.htm
VPN
Christopher R. Carlen
>
> No, but Steen is the *author* of the SymbToolz 1.1 & 2.0
> and I guess that he was so pissed off by the HP ACO dissolve AND
> distroying the future of the 3rd parties software development
> that he still don't like to develope anything for HP calculators
> BUT I surely would like to see a SymbTools 2.0
>
> Please, Steen, say that I'm wrong :-(
Which leads me to often wonder why the whole code base for the 48/49
calculators can't be GPLed and then the calculators could live on
forever with continual improvement. In fact, in such a scenario HP
could decide any time in the future to get back into calculator
production, and find that they only have to implement a hardware
platform and do some porting, to build a product that would likely be
advanced far beyond the likes of TI.
In effect, they could profit from the free open-source software
development, and the development and user community could profit from
having a manufactured hardware platform to run their wares on.
What do you think? Sounds like a win-win situation to me. Who would
have the ability to make something like this happen? What are they
going to do with the internal code to the calculators, stick it in an
archive somewhere to go to waste?
I hope not, but sadly, this will probably be what happens.
Steen Schmidt
Well, the TIs do have a very good symbolic integration algorithm - it's
primarily based on table look-ups. Not that this makes it bad, but it
explains why the TIs are that fast on "text-book" integrations. The TIs are
made for school. When the integrations are more "real-world" like, the two
machines are definetely more equal - one are better than the other,
occasionally, but the other way around happens equally as often.
> This is what concerns me about this calculator.
When doing things like more advanced factorizations and so on, the HP49G
often beats the TI89 hands down - you'd also find the HP alot more
versatile, as it lets you operate on (almost) atomic objects - "almost",
because if you want to operate on a bit-by-bit basis (considering objects),
you need to resort to ML or at least SysRPL (the latter very rarely enough).
There are ML and SysRPL compilers built-in on the HP, and a very good
editor-enhancer exist - Emacs by Carsten Dominik. These object-manipulating
operations are not remotely possible on the TIs.
The SOLVE command on the TI89/92+/V200 is alot better than the HP one when
handling irrational equations, but it's the other way around when handling
polynomial fractions.
The fast answer about speed is that for "school" stuff the TI89 typically
feels alot faster, but for "real" use, the HP is miles more flexible.
> I have been looking through the manuals, and the commands are often
> cryptic, the syntax is hard to figure out, there isn't much detail about
> what is returned from a lot of operations, like solving a numeric
> differential eq., where does the data go? Is it only plotted or can I
> do other things to it numerically?
Yes, HP-documentation is sparse at best, but this group will help alot.
Knowing the HP48 series would help more though ;-)
> (The same question I wonder about with the TI, though.)
And that's often the case too, but look at
http://www-s.ti.com/cgi-bin/discuss/sdbmessage.cgi?databasetoopen=calculator
s for help on the TIs. Be warned though - conversation there tend not to be
quite as civilized as here, but you learn how to cope...
> But the depth of functions that are useful to an EE like Laplace and
> inverse Laplace transforms, very good. But do they take a minute to
> complete all the time?
No, 'X' LAP -> '1/X^2' is around 0.82 seconds on the HP49G. In the same way
a couple of examples:
'1/X*EXP(2*X)+8' ILAP -> '8*Delta(X)+Heaviside(X+2)' in 1.95 seconds.
'(8*X^2-2*X+1)/X^2' ILAP -> '8*Delta(X)+X-2' in 1.78 seconds.
'EXP(X)-5' LAP -> '-5/X+1/(X-1)' in 1.00 seconds.
'EXP(X+2)*X^2' LAP -> 'EXP(2)*(2/(X-1)^3)' in 1.21 seconds.
'(EXP(X+2)*X^2+(X-3)^2-EXP(X)*EXP(X))/8' LAP ->
'EXP(2)*(2/(X-1)^3/8)+(9/X+-6/X^2+2/X^3)/8+-1/(X-2)/8' in 2.99 seconds.
The last result simplifies to
'(4*X^3-12*X^2+7*X-2)/(4*X^4-8*X^3)+1/(4*X^3-12*X^2+12*X-4)*EXP(2)' with LIN
in 4.33 seconds. You could also choose to use EVAL, EXPAND, COLLECT, FACTOR,
PARTFRAC or a number of other rewrite commands, depending on which form
suits you better. A auto-simplification add-on also exist, if you prefer
such complications ;-)
Regards
Steen
Bhuvanesh
> Dr. Klaus Graichen wrote:
> > By the way: The brand-new TI-Voyage 200 cannot solve the integral
> > Integrate(t*e^(-s*t), t, 0, infinity)|s>0,
> > it is possible only by means of the TI-89 and TI-92+, respectively.
> >
> > Klaus Graichen
>
> Are you serious? What is that calculator smoking? Can anyone else
> verify this?
No, I cannot reproduce this with AMS v2.08.
> This is the kind of thing that worries me with TI, that the calculators
> are being geared toward the purely textbook math curriculum of high
> school, and basic college math. But not providing any advanced
> capabilities to do serious engineering or scientific calculations
> (serious for a handheld device, that is.)
But there are plenty of user routines to do such things. Even if
you're a die-hard RPN fan (there's an RPN program for the
TI-89/92+/V200), I would really recommend the TI-89.
--
Bhuvanesh
Marchel
> calculators should have at least either the same or better CAS
The CAS is practically equally good on both. Most of the time TI
is faster and in most cases it is HP that is unable to solve
something, not TI, altough opposite examples also exist.
> and numeric capabilities than the HP,
TI is more accurate in floating point.
> or at least more CPU power,
Motorola 68k is more powerfull than Saturn.
> memory,
Have you used all the memory of your HP49 or TI ?
TI could have better memory management.
Current is worse than HP.
> and display res.
Both, TI89 and TI92 have higher resolution than HP.
> Admittedly, the TI displays are pretty good already.
>
> But I have personally purchased all the chips to build a low-end Palm,
> Motorola Dragonball 16MHz and 33MHz varieties, memory chips to have
> increments of 1MB flash and 4MB RAMs, and the only thing I haven't got
> yet is a display. For large quantities, these things can be obtained
> cheap enough to build a 33MHz calculator with QVGA resolution for about
> the price of a TI-92+.
Including software ?
Have you seen prices for Palms recently ? I haven't seen one that cost
less than $120 (except models that are dropped and on sale) and that
only with primitive OS without any CAS type application.
The one that you might dream about is currently made by Sony
as Clie and cost abput $250 in a black & white display version.
> But TI is wallowing at about 1/3 of this CPU speed, and memory could
> certainly be increased 4-fold. These improvements would allow more
> power to be packed into the CAS.
By increasing 16 fold it would improve even better. That is not a reasonable
argument. High end calculators are already too expensive. Calculator in the
price range of $500 is not competitive to the laptops.
> The second problem is that the HP is much more powerful mathematically,
Any examples ? This statement is untrue. Each calc has it's strengths and
weaknesses.
> yet my reading of the manual makes me a bit displeased as to how easy it
> seems to learn it. The TI user interface seems much easier, but I could
> be wrong here because I haven't much experience with either calculators,
> only TI-83s.
You are wrong. One of the great advantages of HP is that it's interface
allows much easier and deeper integration of software, where TI seems
rather as a group of unrelated pieces of software that have hard time
to talk to each other.
> --
> _____________________
> Christopher R. Carlen
> cr...@earthlink.net
> Suse 7.3 Linux 2.4.10
Jack
Bhuvanesh
> Bhuvanesh wrote:
> > Along those lines, the cubic() and quartic() routines from the tip
> > list can find exact solutions for cubics and quartics, respectively
> > (for the TI-89/92+/V200).
> >
> > --
> > Bhuvanesh
>
> What's the "tip list?"
It's one of the best resources (perhaps the best, next to the manual)
for the TI-89/92+/V200:
http://www.angelfire.com/realm/ti_tiplist/
--
Bhuvanesh
Bhuvanesh
> But TI is wallowing at about 1/3 of this CPU speed, and memory could
> certainly be increased 4-fold. These improvements would allow more
> power to be packed into the CAS.
Of course it would be better if we had all this, but we have to work
with what we have. TI doesn't have any competitors (in the market they
are targeting), so they don't have much motivation to add these
things. They did increase the amount of FlashROM for the Voyage 200,
BTW.
> The second problem is that the HP is much more powerful mathematically,
Considering only built-in functionality (no user add-ons), yes.
--
Bhuvanesh
Chris Carlen
>> But I have personally purchased all the chips to build a low-end
>> Palm, Motorola Dragonball 16MHz and 33MHz varieties, memory chips
>> to have increments of 1MB flash and 4MB RAMs, and the only thing I
>> haven't got yet is a display. For large quantities, these things
>> can be obtained cheap enough to build a 33MHz calculator with QVGA
>> resolution for about the price of a TI-92+.
>
> Including software ?
No, my purpose is to build the core hardware platform of a 68000 class
computer from scratch, and program it from the ground up to something
perhaps as sophisticated as a basic monitor program which can load
programs into memory via a serial port, and jump to them. I'd like to
implement a simple calculator, perhaps in the future to develop further.
I already know how to do a basic expression parser, so if I can get
that far, it will satisfy my purpose, which is mostly self-educational.
>
> Have you seen prices for Palms recently ? I haven't seen one that
> cost less than $120 (except models that are dropped and on sale) and
> that only with primitive OS without any CAS type application. The one
> that you might dream about is currently made by Sony as Clie and cost
> abput $250 in a black & white display version.
>
>
>> But TI is wallowing at about 1/3 of this CPU speed, and memory
>> could certainly be increased 4-fold. These improvements would
>> allow more power to be packed into the CAS.
>
>
> By increasing 16 fold it would improve even better. That is not a
> reasonable argument. High end calculators are already too expensive.
> Calculator in the price range of $500 is not competitive to the
> laptops.
My point is simply that TI is about a factor of 2 to 3 weak in the CPU
and memory department. Consider the Palm m500, which they don't give
enough details about for me to be 100% certain, but I'd bet that it has
a 33MHz Dragonball, a 256x128 screen with backlight, and has 8MB of
memory. Factor of 3 all over the place. It's $200.
>> The second problem is that the HP is much more powerful
>> mathematically,
>
> Any examples ? This statement is untrue. Each calc has it's strengths
> and weaknesses.
Laplace transforms, discrete Fourier transforms, inverse Laplace, and
partial fraction expansions appear to be in the HP out of the box, and
these are the functions I will need regularly. Can TI do these things
out of the box? Can add on programs do them? I would like to get the
add ons if available.
Can either calculator for that matter, out of the box of with add ons,
do numerical minimization of functions of more than one variable? This
is another I'd like to do regularly.
Finally, which one solve systems of diff-eqs? It seems TI can and HP
not, so the advantage may lie with TI here. But I haven't tested to see
exactly what they can handle. I'd like to run small examples of the
systems that arise in electronic circuits with non-linear semiconductor
elements. Perhaps neither calculator can go this far.
>> yet my reading of the manual makes me a bit displeased as to how
>> easy it seems to learn it. The TI user interface seems much
>> easier, but I could be wrong here because I haven't much experience
>> with either calculators, only TI-83s.
>
> You are wrong. One of the great advantages of HP is that it's
> interface allows much easier and deeper integration of software,
> where TI seems rather as a group of unrelated pieces of software that
> have hard time to talk to each other.
This is an impression that I have been gathering as well. But the HP is
very different from the TI, and perhaps if I had never used any
calculator before they would seem equally obscure. Perhaps after
mastering the HP, one can do things in a way that would be much more
cumbersome than on the TI. But I was able to do things with the TI
within seconds, that I can't figure out how to do with the HP by just
looking at the keyboard.
Veli-Pekka Nousiainen
X
> Finally, which one solve systems of diff-eqs? It seems TI can and HP
> not, so the advantage may lie with TI here. But I haven't tested to see
> exactly what they can handle. I'd like to run small examples of the
> systems that arise in electronic circuits with non-linear semiconductor
> elements. Perhaps neither calculator can go this far.
I don't know about these things for sure,
but have you tried to use a vector of algebraic equations
as an input for LDEC/DESOLVE on the 49G ?!
Tell about the results if you get it to work...
VPN
Bhuvanesh
> > There are two problems: for the price, the TI-89, 92+, and 200
> > calculators should have at least either the same or better CAS
>
> The CAS is practically equally good on both. Most of the time TI
> is faster and in most cases it is HP that is unable to solve
> something, not TI, altough opposite examples also exist.
I wasn't thinking about it that way (I was thinking of breadth of
functionality rather than depth). I think you (Jack) are correct.
> > The second problem is that the HP is much more powerful mathematically,
>
> Any examples ? This statement is untrue. Each calc has it's strengths and
> weaknesses.
Maybe he's talking about the number of functions. There are more
functions on the HP49G than on the TI-89/92+/V200, although the HP49G
functionality can be easily duplicated with user programs on the TIs.
> You are wrong. One of the great advantages of HP is that it's interface
> allows much easier and deeper integration of software, where TI seems
> rather as a group of unrelated pieces of software that have hard time
> to talk to each other.
My impression from these years of reading posts on comp.sys.hp48 is
that it is the other way round (the TI-89/92+/V200 have better
software integration).
--
Bhuvanesh
Bhuvanesh
> My point is simply that TI is about a factor of 2 to 3 weak in the CPU
> and memory department. Consider the Palm m500, which they don't give
> enough details about for me to be 100% certain, but I'd bet that it has
> a 33MHz Dragonball, a 256x128 screen with backlight, and has 8MB of
> memory. Factor of 3 all over the place. It's $200.
Comparison with Palms is not a good idea. They are a completely
different product, for a completely different market.
> Laplace transforms, discrete Fourier transforms, inverse Laplace, and
> partial fraction expansions appear to be in the HP out of the box, and
> these are the functions I will need regularly. Can TI do these things
> out of the box?
Partial fraction expansions? Yes. There are add-ons for the rest:
http://tiger.towson.edu/~bbhatt1/ti/beta/MPL.htm
> Can either calculator for that matter, out of the box of with add ons,
> do numerical minimization of functions of more than one variable? This
> is another I'd like to do regularly.
With add-ons, the TIs can do this. The simplex method (and its
nonlinear variant) have been implemented, as has simulated annealing.
> Finally, which one solve systems of diff-eqs? It seems TI can and HP
> not, so the advantage may lie with TI here.
Correct. Lars' programs can also solve integro-differential equations.
--
Bhuvanesh
Marchel
> My point is simply that TI is about a factor of 2 to 3 weak in the CPU
> and memory department. Consider the Palm m500, which they don't give
> enough details about for me to be 100% certain, but I'd bet that it has
> a 33MHz Dragonball, a 256x128 screen with backlight, and has 8MB of
> memory. Factor of 3 all over the place. It's $200.
I agree with you that TI could give us a better hardware, but consider, that
HP49 was released after TI with even worse hardware and higher
price. There was once discussion on this forum about who can afford
expensive calculators. Keep in mind, that TI89 sells now for about $120
which is half it's original price, and HP49 doesn't sell at all anymore
at least in US. If you don't sell certain number of hardware you cannot
sustain low prices even with outdated hardware. HP was even forced
to abandon calculator business altogether.
> >> The second problem is that the HP is much more powerful
> >> mathematically,
> >
> > Any examples ? This statement is untrue. Each calc has it's strengths
> > and weaknesses.
> Laplace transforms, discrete Fourier transforms, inverse Laplace, and
> partial fraction expansions appear to be in the HP out of the box, and
> these are the functions I will need regularly. Can TI do these things
> out of the box? Can add on programs do them? I would like to get the
> add ons if available.
There are certain functions that HP has but there are also certain
functions, that HP lacks and TI has. For example HP does not have
entire geometry module out of the box. You and I don't need it
but there might be people who do need it. I for example found out, that
many algebraic expressions bog down HP to the point of being ridiculously
unusable, where TI still does the same algebraic conversion fast.
And I'm alking here about real life examples, not some exotic
tests designed to bog down that particular machine.
> Finally, which one solve systems of diff-eqs? It seems TI can and HP
> not, so the advantage may lie with TI here.
They both do.
> But I haven't tested to see
> exactly what they can handle. I'd like to run small examples of the
> systems that arise in electronic circuits with non-linear semiconductor
> elements. Perhaps neither calculator can go this far.
To the contrary, I found that both can handle nonlinear diff-equations
to the point of course. You can also solve it numerically if you are
just studying engineering problem with given numerical values.
Both calculators are capable to solve diff-equations numerically
altough there are some restrictions to the format of the equation
or set of equations if you are trying to solve higher degree equation.
> This is an impression that I have been gathering as well. But the HP is
> very different from the TI, and perhaps if I had never used any
> calculator before they would seem equally obscure.
The first thing to do, is to get used to and start appreciate RPN. If
you don't like it, you are probably better off with TI and will never
like HP. HP mistake was to introduce algebraic notation (and extreamally
poor hardware). Not only it did not saved the HP calculators, but also
enraged many dedicated HP users. The 49 RPN is significantly less
consistent than HP48 and probably less consistent than TI exactly
because it was trying to copy some of the TI interface ideas such as
algebraic input, or APPS. APPS give you nice interface but makes
object interchangebility less obvious than it was with HP48 interface.
> Perhaps after
> mastering the HP, one can do things in a way that would be much more
> cumbersome than on the TI. But I was able to do things with the TI
> within seconds, that I can't figure out how to do with the HP by just
> looking at the keyboard.
It is a matter of getting used to. Unfortunately HP menus
are usually less descriptive than TI but their
acess is more logically spreaded throughe the keyboard.
> _____________________
> Christopher R. Carlen
> cr...@earthlink.net
> Suse 7.3 Linux 2.4.10
Jack
Marchel
> > Any examples ? This statement is untrue. Each calc has it's strengths and
> > weaknesses.
>
> Maybe he's talking about the number of functions. There are more
> functions on the HP49G than on the TI-89/92+/V200, although the HP49G
> functionality can be easily duplicated with user programs on the TIs.
Number of functions does not neccesairly translates to "better" math.
Many of the functions on both calculators are obvious, rather simple macros
such as ABS(x) or simple combinations of other functions. They maybe
shorten the typing but not exactly add much more usability.
There is also a class of functions on both calcs that are very obscure
and usable only to the narrow circle of specialists. Those people clearly can
than state that one calc is significantly better than the other. For
example HP users can choose HP because it has powerful linear algebra
that exceeds TI, others might choose TI because of geometry application,
TI automatic expression simplification etc. I consider those functions
to be obscure. The general math CAS is in both calcs on the very similar
level, and in my opinion it is matter of personal choice rather than the true
math advantage.
> My impression from these years of reading posts on comp.sys.hp48 is
> that it is the other way round (the TI-89/92+/V200 have better
> software integration).
Consider for example, that the TI output of program goes to the program
screen and is unusable as a stack object for regular functions of the
calculator without special needs. On HP (when running in RPN
which is the only sensible way to run it) the good program take
input from stack and outputs data to the stack. It does two things:
makes output data naturally available for the other calc commands
and programs, make the program itself natural extension of the
calculator build in functions in the consistent manner.
Jack
Veli-Pekka Nousiainen
news:qZz99.154801$2p2.6...@bin4.nnrp.aus1.giganews.com...X
> Consider for example, that the TI output of program goes to the program
> screen and is unusable as a stack object for regular functions of the
> calculator without special needs. On HP (when running in RPN
> which is the only sensible way to run it) the good program take
> input from stack and outputs data to the stack. It does two things:
> makes output data naturally available for the other calc commands
> and programs, make the program itself natural extension of the
> calculator build in functions in the consistent manner.
Exactly why I got rid of the TI 89
I found it much easier to program the HP 49G
to do the tricks that I wanted.
I also like the Custom Menu capability of the EQW.
Does the EQW application on the TI 89 allow
for user program expansions, Bhuvanesh?!
VPN
Alan
What user add-ons exist for the 89 that make it comparable to the 49.
I own both calcs and I know of nothing from TI or even third party
sources that bring it up to par with the 49. The equation writer is
nice but you have to pay $15.00 for it when the 49's comes with the
calc. But what else exists that makes the 89 better mathematically?
Bhuvanesh
> > My impression from these years of reading posts on comp.sys.hp48 is
> > that it is the other way round (the TI-89/92+/V200 have better
> > software integration).
>
> Consider for example, that the TI output of program goes to the program
> screen and is unusable as a stack object for regular functions of the
> calculator without special needs. On HP (when running in RPN
> which is the only sensible way to run it) the good program take
> input from stack and outputs data to the stack. It does two things:
> makes output data naturally available for the other calc commands
> and programs, make the program itself natural extension of the
> calculator build in functions in the consistent manner.
This is why I generally use functions instead of programs. IMHO,
functions should be the default TI-Basic program type. Programs do
have their advantages too, in particular being able to use all
instructions and built-in functions.
--
Bhuvanesh
Bhuvanesh
> Does the EQW application on the TI 89 allow
> for user program expansions, Bhuvanesh?!
Of course.
--
Bhuvanesh
ed
ed
Veli-Pekka Nousiainen
X
> > Consider for example, that the TI output of program goes to the program
> > screen and is unusable as a stack object for regular functions of the
> > calculator without special needs. On HP (when running in RPN
> > which is the only sensible way to run it) the good program take
> > input from stack and outputs data to the stack. It does two things:
> > makes output data naturally available for the other calc commands
> > and programs, make the program itself natural extension of the
> > calculator build in functions in the consistent manner.
>
> This is why I generally use functions instead of programs. IMHO,
> functions should be the default TI-Basic program type. Programs do
> have their advantages too, in particular being able to use all
> instructions and built-in functions.
Oh - that limitation had the very first Symbolic Calculator
the HP 28C. You had to write functions in Algebraic form
<< -> argument'function(argument)+argument^2*...' >>
restricting the user to build-in functions to build up new functions
instead of using any commands available (even side-effects)
Later, when the HP 28S was introduced
any kind of program starting with "an argument syntax"
or local variables (lambda variables/temporary variables)
like this:
<< -> width length height << any commands >> >>
was a "legal" Algebraic Function.
In the introduction of the 48SX it went even further.
User could supply his/her own function with a derivative
starting the derivative name with der
<< -> x y << commands >> >> 'My.Fn' STO
<< -> x y dx dy << commands for derivative >> >> 'derMy.Fn' STO
AND
One could use APPLY to even add step-wise evaluation
similar to the build-in functions to their own functions.
Unfortunately it seems that this feature has been removed from
the HP 49G OS and I'm not sure about the derNAME anymore
since derivative now are d1NAME
Naturally the CAS in the 49G has even more power available,
but nevertheless there are only minor restrictions to functions
versus commands/programs in the HP 4x series.
VPN
Veli-Pekka Nousiainen
news:d2e5be41.0208...@posting.google.com...
X
> What user add-ons exist for the 89 that make it comparable to the 49.
> I own both calcs and I know of nothing from TI or even third party
> sources that bring it up to par with the 49. The equation writer is
> nice but you have to pay $15.00 for it when the 49's comes with the
> calc. But what else exists that makes the 89 better mathematically?
Faster/better SOLVE ?
Mathematically? - Nothing!
VPN
Marchel
>
> > Consider for example, that the TI output of program goes to the program
> > screen and is unusable as a stack object for regular functions of the
> > calculator without special needs. On HP (when running in RPN
> > which is the only sensible way to run it) the good program take
> > input from stack and outputs data to the stack. It does two things:
> > makes output data naturally available for the other calc commands
> > and programs, make the program itself natural extension of the
> > calculator build in functions in the consistent manner.
>
> This is why I generally use functions instead of programs. IMHO,
> functions should be the default TI-Basic program type. Programs do
> have their advantages too, in particular being able to use all
> instructions and built-in functions.
One of the limits of such aprroach is, that the function outputs only single
value. HP program can output as many values on the stack as the user want.
>
> --
> Bhuvanesh
Jack
Chris Carlen
>
> One of the limits of such aprroach is, that the function outputs only single
> value. HP program can output as many values on the stack as the user want.
>
Can a TI-89(92+) function return complex numbers, lists, matrices? If
so, then perhaps this isn't a problem.
Bhuvanesh
> What user add-ons exist for the 89 that make it comparable to the 49.
> I own both calcs and I know of nothing from TI or even third party
> sources that bring it up to par with the 49. The equation writer is
> nice but you have to pay $15.00 for it when the 49's comes with the
> calc. But what else exists that makes the 89 better mathematically?
It depends. What do you need? :-)
Here's the beginning of a master program list for the TI-89/92+/V200:
http://tiger.towson.edu/~bbhatt1/ti/beta/MPL.htm
Also, there is the free version of the EQW:
http://triton.towson.edu/users/bbhatt1/ti/eqw.htm
Check the recommended websites on my page:
http://triton.towson.edu/users/bbhatt1/ti/
I'll also take this chance to mention MathTools ;-)
http://triton.towson.edu/users/bbhatt1/ti/#func
Here's the HTML version of its documentation:
http://triton.towson.edu/users/bbhatt1/ti/MathTools.htm
--
Bhuvanesh
Bhuvanesh
> Faster/better SOLVE ?
For finding roots of polynomials, you mean? There's an add-on for
that. Besides that, the TI-89 solve() and cSolve() are quite nice.
--
Bhuvanesh
Veli-Pekka Nousiainen
I was defending the TI cSolve() this time!
eg. faster & better !!
;-)
VPN
"Bhuvanesh" <lalu_...@yahoo.com> wrote in message
news:662e00ed.02082...@posting.google.com...
Marchel
> Marchel wrote:
> >
> > One of the limits of such aprroach is, that the function outputs only single
> > value. HP program can output as many values on the stack as the user want.
> >
>
>
> Can a TI-89(92+) function return complex numbers, lists, matrices? If
> so, then perhaps this isn't a problem.
I disagree. Even, when it would be possible (which is not), this would be
rather poor workaround to the problem. Grouping outcome of program into
the list or matrix artificially puts them into the same physically related
class, so later you need more operations to split them usually accessible
only through the menus (unlike SWAP, DROP commands which are
primary keys on HP keyboard). If my program for example calculates
acceleration, speed and distance as three numbers it is doubtful it will
have any use for built in calculator functions when grouped in the list or
matrix simply because they represent different physical data that
most likely need different functionality to be applied to each member
of the list. It then force you to first breakup the list which is usually a
menu hidden command not immediately accessible through primary
keyboard key. And this is not a "made up" argument. On my HP49 I
have several programs - calculator extensions that use such behavior.
For example one of my programs calculates roots of the cubic equation
and always produces three solutions (even when two are complex or
identical etc.) on the stack as three separate numbers. Then using strictly
keyboard DROP and SWAP keys I can quickly isolate one that is needed
for other calculator functions or even keep other on the stack in the
meantime for further work. Trying to mimic such functionality on TI although
probably possible with tricks (through variables or in some other way) is
awkward and is not an natural extension to the TI interface. Keep in mind,
that we are discussing here consistency of the interface. Any workaround,
if even possible only reinforces my argument, that TI interface is
significantly less consistent, than HP interface in RPN mode.
Jack
Veli-Pekka Nousiainen
X
> For example one of my programs calculates roots of the cubic equation
> and always produces three solutions (even when two are complex or
> identical etc.) on the stack as three separate numbers. Then using
strictly
> keyboard DROP and SWAP keys I can quickly isolate one that is needed
> for other calculator functions or even keep other on the stack in the
> meantime for further work. Trying to mimic such functionality on TI
although
> probably possible with tricks (through variables or in some other way) is
> awkward and is not an natural extension to the TI interface. Keep in mind,
> that we are discussing here consistency of the interface. Any workaround,
> if even possible only reinforces my argument, that TI interface is
> significantly less consistent, than HP interface in RPN mode.
>
> Jack
Right on, Jack !
I have sometimes used the STARTEQW reserved variable
in order to form a selection list when a HP command (SOLVE)
returns several answers in a list (in my HP 49G).
I then select the answer that I want and continue with the calculation.
This could be one way to circumvent the TI 89 shortcomings
if this is possible to do from the add-on EQW application.
Bhuvanesh?
VPN
Chris Carlen
> "Chris Carlen" <cr...@BOGUS.earthlink.net> wrote in message news:3D6841E5...@BOGUS.earthlink.net...
>
>>Marchel wrote:
>>
>>>One of the limits of such aprroach is, that the function outputs only single
>>>value. HP program can output as many values on the stack as the user want.
>>
>>Can a TI-89(92+) function return complex numbers, lists, matrices? If
>>so, then perhaps this isn't a problem.
I just tried it, and it seems that one can return complex variables,
lists, and matrices from a function.
>
> I disagree. Even, when it would be possible (which is not), this would be
> rather poor workaround to the problem. Grouping outcome of program into
> the list or matrix artificially puts them into the same physically related
> class, so later you need more operations to split them usually accessible
> only through the menus (unlike SWAP, DROP commands which are
> primary keys on HP keyboard). If my program for example calculates
> acceleration, speed and distance as three numbers it is doubtful it will
> have any use for built in calculator functions when grouped in the list or
> matrix simply because they represent different physical data that
> most likely need different functionality to be applied to each member
> of the list. It then force you to first breakup the list which is usually a
> menu hidden command not immediately accessible through primary
> keyboard key.
Actually, it seems one can simply do it like this:
{m*v^2/2,m*v}->kep(m,v)
for example, to define a function kep(m,v) that returns the kinetic
energy and momentum of an object of mass m moving with velocity v.
Then do:
kep(2.34,5.67)[1] if you only want the first result, the kinetic energy,
or [2] if you want the momentum. If you want both results hanging
around for future use, then stick the returned value in a variable, and
index the subelements later.
This is par for the course with non-stack based programming languages,
and interactive interpreted languages as well. There is nothing
counterintuitive or inconsistent about it. One just needs to know the
conventions used as to what part of the result is stored in what
subelement of the result object.
> And this is not a "made up" argument. On my HP49 I
> have several programs - calculator extensions that use such behavior.
> For example one of my programs calculates roots of the cubic equation
> and always produces three solutions (even when two are complex or
> identical etc.) on the stack as three separate numbers. Then using strictly
> keyboard DROP and SWAP keys I can quickly isolate one that is needed
> for other calculator functions or even keep other on the stack in the
> meantime for further work. Trying to mimic such functionality on TI although
> probably possible with tricks (through variables or in some other way) is
> awkward and is not an natural extension to the TI interface. Keep in mind,
> that we are discussing here consistency of the interface. Any workaround,
> if even possible only reinforces my argument, that TI interface is
> significantly less consistent, than HP interface in RPN mode.
I'm not sure about less consistent, but less preferrable maybe, to one
who prefers a stack machine. The argument as to whether a stack vs.
load/store type paradigm is superior, I won't even go there because I
don't care. Just use what you like! Hopefully, we will continue to
have choices available. If you HP folks get stuck with no new
calculators one day, then I will sincerely regret that situation.
Good day!
Marchel
> I'm not sure about less consistent, but less preferrable maybe, to one
> who prefers a stack machine. The argument as to whether a stack vs.
> load/store type paradigm is superior, I won't even go there because I
> don't care. Just use what you like! Hopefully, we will continue to
> have choices available. If you HP folks get stuck with no new
> calculators one day, then I will sincerely regret that situation.
>
> Good day!
It is less consistent. I only gave you the simples example where HP is
automatically more consistent than TI with outputing everything into the
main calculator stack. You seem to forget that on HP you can use
many interactive features within program including graphs and interactive
user graphic input that creates entries on the main calculator stack.
You can put programs itself on the stack which opens myriad of
programming possibilities using programs itself as a product of
function for further executing. Try to do it so simply with TI without
tricks. Good luck.
Jack
Bhuvanesh
> It is less consistent. I only gave you the simples example where HP is
> automatically more consistent than TI with outputing everything into the
> main calculator stack.
You can very easily get a list of the solutions using exp>list(), or
by using zeros()/cZeros() instead of solve()/cSolve(), and can then
get a particular root using list subscripts. I don't see how that's
less consistent.
> You can put programs itself on the stack which opens myriad of
> programming possibilities using programs itself as a product of
> function for further executing. Try to do it so simply with TI without
> tricks. Good luck.
Define myprgm(args)=Prgm:...:EndPrgm ??? Yes, you can define programs
and functions from within programs.
--
Bhuvanesh
Veli-Pekka Nousiainen
> Define myprgm(args)=Prgm:...:EndPrgm ??? Yes, you can define programs
> and functions from within programs.
Remarkable! Maybe I should have kept the TI 89 I had.
Can it use list as programs. You know one can manipulate the lists
with POS POT PUTI GET GETI SUB REPL SIZE HEAD TAIL
DOLIST DOSUBS NSUB ENDSUB STREAM REVLIST SORT SEQ
and that applies to programs as list in the HP 4x series
Just supply { } instead of << >> and remember to use EVAL to run.
RPL is surely not just a stack-based language like Forth
but has also strong roots in LISP as well.
Anything similar in the TI?
Maybe I'll buy a V200 after all !!
:-D
VPN
Timité Hassan
> > There are two problems: for the price, the TI-89, 92+, and 200
> > calculators should have at least either the same or better CAS
>
> The CAS is practically equally good on both. Most of the time TI
> is faster and in most cases it is HP that is unable to solve
> something, not TI, altough opposite examples also exist.
Well,even TI users admit that out of the box the HP CAS is better.
I find it personnaly better because it is not alterated by TI-OS
limitations,it often has better algorithms and it doesn't have the
often nuisible autosimplification of the TI-CAS(examples:try to work
with small matrix whom elements are complex expressions with square
roots or try to expand rationnal fractions whom roots of the
denominator are complex numbers or expressions).
> > and numeric capabilities than the HP,
>
> TI is more accurate in floating point.
Yes but it is often much slower.
Do you know that it is possible to write TI-Basic program barelly
slower than the built-in functions for numeric computations ?
I have written a TI-Basic function for doing matrix inversions and it
is only 50 % slower than the built-in function for 15*15 random
matrix.
It is twice faster for diagonal matrix !!!
Assuming the legendary slowness of the TI-Basic,it is a shame.
> > or at least more CPU power,
>
> Motorola 68k is more powerfull than Saturn.
Yep,but TI would need to use at least a 68040 to beat the HP49 in
speed for everything because the TI engineers are unable to use a
decent compiler,the right algorithms and to correctly optimise the
software code.
> > memory,
>
> Have you used all the memory of your HP49 or TI ?
I have run out of memory with my TI92+ and there is only TI-Basic
programs and normal variables(no C/ASM programs,no Flash apps) in my
memory.
Additionnaly the RAM memory is highly insufficient even for TI-Basic
programming.
> TI could have better memory management.
> Current is worse than HP.
It is an euphemism.
> > But TI is wallowing at about 1/3 of this CPU speed, and memory could
> > certainly be increased 4-fold. These improvements would allow more
> > power to be packed into the CAS.
>
> By increasing 16 fold it would improve even better. That is not a reasonable
> argument. High end calculators are already too expensive. Calculator in the
> price range of $500 is not competitive to the laptops.
Well with the number of calculators TI sold,it could produce easily a
TI68k with a faster CPU,at least 1 MB of RAM and much more Flash ROM
than the Voyage 200 and at even a lower price.
Don't forget that all TI calculators are highly overpriced and use
almost obsolete hardware(less archaic than HP but,you get the idea).
> > The second problem is that the HP is much more powerful mathematically,
>
> Any examples ? This statement is untrue. Each calc has it's strengths and
> weaknesses.
It happens that out of the box the HP software has much less
weaknesses and much more strengths than TI software.
The OS core is much more powerful including memory management.
The object oriented design of the HP OS is way superior to the archaic
design of the TI OS.
The HP User Interface is much more flexible and customisable.
The HP software use most of the time better algorithms and is much
more optimised,etc...
And overall you can do much more with the HP than with the TI.
Only TI engineers and TI worshippers can believe the opposite.
> > yet my reading of the manual makes me a bit displeased as to how easy it
> > seems to learn it. The TI user interface seems much easier, but I could
> > be wrong here because I haven't much experience with either calculators,
> > only TI-83s.
>
> You are wrong. One of the great advantages of HP is that it's interface
> allows much easier and deeper integration of software, where TI seems
> rather as a group of unrelated pieces of software that have hard time
> to talk to each other.
Here we agree.
> > --
> > _____________________
> > Christopher R. Carlen
> > cr...@earthlink.net
> > Suse 7.3 Linux 2.4.10
>
> Jack
Timité Hassan
>
> > Darn, we really need some more innovation in the calculator world. I
> > don't think TI is going to meet my expectations.
>
> Not even with third-party add-ons?
Tell me Bhuvanesh:
Can the primitive memory management of the TI OS be fixed by third
party add-on ?
Can the autosimplification be disabled by add-ons ?
Can all the troubles TI give to third party developpers can be fixed
by add-ons ?
And you seem to forget that HP calculators have much more third party
add-ons than the TI calculators will ever have.
And at least HP is/was not giving third party developpers as many
troubles as TI is.
Timité Hassan
>
> > My point is simply that TI is about a factor of 2 to 3 weak in the CPU
> > and memory department. Consider the Palm m500, which they don't give
> > enough details about for me to be 100% certain, but I'd bet that it has
> > a 33MHz Dragonball, a 256x128 screen with backlight, and has 8MB of
> > memory. Factor of 3 all over the place. It's $200.
>
> Comparison with Palms is not a good idea. They are a completely
> different product, for a completely different market.
>
There are Palm OS PDA with at least 16 Mhz CPU and 2 MB of RAM with a
price as low as 99 euros(a bit less than US $ 99).
So how come the Voyage 200 sold at 243 euros can't have better or at
least equal hardware ?
And don't come with Bullshit about the softtware cost.
The TI software hasn't significantly evolved since the A.M.S 2.0x
release over 2 years ago and assuming the poor quality of TI flash
apps,TI doesn't spend significant amount of money for their
developpment and maintenance.
Timité Hassan
>
> Maybe he's talking about the number of functions. There are more
> functions on the HP49G than on the TI-89/92+/V200, although the HP49G
> functionality can be easily duplicated with user programs on the TIs.
Well,The HP49G cleary wins at this game as you can use directly System
RPL on the HP49 to write even more powerful programs while you have to
use C programming on PC to get similar results on the TI.
> > You are wrong. One of the great advantages of HP is that it's interface
> > allows much easier and deeper integration of software, where TI seems
> > rather as a group of unrelated pieces of software that have hard time
> > to talk to each other.
>
> My impression from these years of reading posts on comp.sys.hp48 is
> that it is the other way round (the TI-89/92+/V200 have better
> software integration).
You are dead wrong.
Here are some examples of software integration:
You can use the EQW or the Matrix writer to add an object to the
command line.
You can call EQW or Matrix writer to edit an object in the stack.
Or you can use built-in editors(EQW,Matrix writer,etc...) from users
programs.
Can you do the same on the TI ?
Timité Hassan
Well perahps,Jacek has used the wrong example.
The true problems with useing list to return several results is that
contrary to the HP lists,TI lists are not true lists in the sense that
they can't handle many variables types such as list or matrix.
If you must return results such as lists or matrix when you are
screwed.
Of course there are workarounds(such as put each result in a string
for examples) but it would have been much simpler if programs could
use the history to return and to get objects.
Timité Hassan
No the TI68k can't use list as program.
> Anything similar in the TI?
Some of the functions you have listed are built-in on the TI68k with
off course different name for example:
HEAD->listname[1]
SIZE->dim()
Some other can be emulated with Ti-basic functions:
for examples REVLIST->seq(list[k],k,dim(list),1,-1),etc...
> Maybe I'll buy a V200 after all !!
I advice you not as you would probably waste your money.
Because if you haven't found any real reason to buy a TI89 then you
have no real reason to buy a Voyage 200.
The only interesting new features are the clock and the time
functions.
Features availlable on the HP for ages.
TI has also probably fix some bugs(after 2 years of inactivity,it is a
minimum).
But overall nothing really stellar.
After that some people stiil don't understand why i have such a poor
opinion of TI.
> :-D
> VPN
Marchel
> Well,even TI users admit that out of the box the HP CAS is better.
> I find it personnaly better because it is not alterated by TI-OS
> limitations,it often has better algorithms and it doesn't have the
> often nuisible autosimplification of the TI-CAS(examples:try to work
> with small matrix whom elements are complex expressions with square
> roots or try to expand rationnal fractions whom roots of the
> denominator are complex numbers or expressions).
I don't. Out of the box it is TI CAS that is usually more faster and better.
Especially when it comes down to long equations. Long equations
which are normally the product of chain of calculations are rather norm than
the exception unless you are just very low level beginer in math (but then why
you even need advanced calculator). HP with long equations not only slows
down to the totally useless speed and also the display becomes amazingly
insufficient to communicate anything usefull especially when combined with
the fact, that HP by default not only is incapable to simplify expressions,
but also very often is incapable to do so with "EVAL" and similar commands.
The simple number of functions that HP suppose to have more over TI does
not convince me that HP is better. To the contrary, most of them are exotic
functions needed only rarely and most of them can be replaced with simple
macros on any other machine. There are some of the HP advanteges, like
those that you mentioned but what usefullness is of it, if it only practically
works on the short formulas and makes calculator die on the long ones.
By the way, in time when TI is done with chain of complex formulas and even
manages to simplfy them, HP usually is "still working" on the first one.
So much for the advnaced CAS.
> > > and numeric capabilities than the HP,
> >
> > TI is more accurate in floating point.
>
> Yes but it is often much slower.
That is in general not true. It is TI Basic that is very slow.
Most of the built in floating point calculations are usually
faster than HP. You can see that when ploting functions
that require complex floating point math to produce answer.
> Do you know that it is possible to write TI-Basic program barelly
> slower than the built-in functions for numeric computations ?
> I have written a TI-Basic function for doing matrix inversions and it
> is only 50 % slower than the built-in function for 15*15 random
> matrix.
> It is twice faster for diagonal matrix !!!
> Assuming the legendary slowness of the TI-Basic,it is a shame.
Matrix on TI is in fact very slow. For some reason TI implemented matrix
as always symbolic even if the entire matrix has real or complex numbers.
That is why when implemented in TI Basic as numerical matrix
computation you almost reach TI built in speed. But try symbolic
matrix on TI and on HP and you will find out that often TI outruns
HP in symbolic matrix despite the fact that TI manages to simplify
expressions and HP doesn't even try to do so. True, that numerical
matrix is faster on HP.
> > Motorola 68k is more powerfull than Saturn.
>
> Yep,but TI would need to use at least a 68040 to beat the HP49 in
> speed for everything because the TI engineers are unable to use a
> decent compiler,the right algorithms and to correctly optimise the
> software code.
Nonsense. With 68K TI generally outruns HP49 in most functions
despite the fact, that TI Basic is very slow and floting point is
implemented in BCD never intended to be mainstream M68K data
format.
> > Have you used all the memory of your HP49 or TI ?
>
> I have run out of memory with my TI92+ and there is only TI-Basic
> programs and normal variables(no C/ASM programs,no Flash apps) in my
> memory.
You are unique. In your case HP49 is a better machine.
> Additionnaly the RAM memory is highly insufficient even for TI-Basic
> programming.
What do you mean ? Do you need flash RAM or what ? ;-)
> Well with the number of calculators TI sold,it could produce easily a
> TI68k with a faster CPU,at least 1 MB of RAM and much more Flash ROM
> than the Voyage 200 and at even a lower price.
> Don't forget that all TI calculators are highly overpriced and use
> almost obsolete hardware(less archaic than HP but,you get the idea).
And how exactly do you know that ? Are you working for the internal TI
finance departament ? Actually I don't "get an idea". High end calculators are
not a mass market product as one would think. They are sold in a significantly
lower numbers than a typical engineering calculator. Short production series
often drive high cost.
> > > The second problem is that the HP is much more powerful mathematically,
> >
> > Any examples ? This statement is untrue. Each calc has it's strengths and
> > weaknesses.
>
> It happens that out of the box the HP software has much less
> weaknesses and much more strengths than TI software.
> The OS core is much more powerful including memory management.
OS core has nothing to do with "mathematically powerfull".
> The object oriented design of the HP OS is way superior to the archaic
> design of the TI OS.
"Object oriented" has nothing to do with "mathematically powerfull".
> The HP User Interface is much more flexible and customisable.
"User interface" has nothing to do with "mathematically powerfull".
> The HP software use most of the time better algorithms and is much
> more optimised,etc...
Not true. There are examples where TI can solve problems which HP
cant' and opposite. Especially in integrals it is TI which usually
does one when HP can't.
> And overall you can do much more with the HP than with the TI.
> Only TI engineers and TI worshippers can believe the opposite.
Not true.
Jack
T.F.
> Or you can use built-in editors(EQW,Matrix writer,etc...) from users
> programs.
> Can you do the same on the TI ?
The EQW Flash app can edit objects in the stack. Both the free version of
the EQW and the Flash app can be used from user programs. The Matrix writer
and most of the other built-in apps can't.
Bhuvanesh
Veli-Pekka Nousiainen
news:<P%Ea9.1902$D8.6...@reader1.news.jippii.net>...
> > "Bhuvanesh" <lalu_...@yahoo.com> wrote in message
> > news:662e00ed.0208...@posting.google.com...
> > > Define myprgm(args)=Prgm:...:EndPrgm ??? Yes, you can define programs
> > > and functions from within programs.
> >
> > Remarkable! Maybe I should have kept the TI 89 I had.
> > Can it use list as programs. You know one can manipulate the lists
> No the TI68k can't use list as program.
X
> > Maybe I'll buy a V200 after all !!
>
> I advice you not as you would probably waste your money.
ok, I wait for more convincing arguments from Bhuvanesh.
Your other posts showed the result stack problems of the TI
and other stupid design flaws.
While I'm often pissed off by some peculiarities of the 49G
I now see that there is even greater darkness...
Only if we could get units back to the EQW Environment !!
Parisse, Please !!!
VPN
Bhuvanesh
> > > Can it use list as programs. You know one can manipulate the lists
Of course you can manipulate lists on the TI-68k, if that's what you
mean.
> > No the TI68k can't use list as program.
A list as a program??
> > > Maybe I'll buy a V200 after all !!
> >
> > I advice you not as you would probably waste your money.
It kind of depends on what you'll use it for, but as a developer, I'm
quite pleased with the TI-68k series.
> ok, I wait for more convincing arguments from Bhuvanesh.
I haven't been following this thread (too busy at work).
I'll reply sometime soon. Better not reply until I understand
everything that's been said ;-)
--
Bhuvanesh
Bhuvanesh
> > Maybe he's talking about the number of functions. There are more
> > functions on the HP49G than on the TI-89/92+/V200, although the HP49G
> > functionality can be easily duplicated with user programs on the TIs.
>
> Well,The HP49G cleary wins at this game as you can use directly System
> RPL on the HP49 to write even more powerful programs while you have to
> use C programming on PC to get similar results on the TI.
I see nothing wrong with writing C programs on a PC. I do it quite
often, and not only for calculators.
--
Bhuvanesh
Veli-Pekka Nousiainen
> "Veli-Pekka Nousiainen" <DROP...@welho.com> wrote:
X
> I haven't been following this thread (too busy at work).
> I'll reply sometime soon. Better not reply until I understand
> everything that's been said ;-)
;-)
VPN
Marchel
>
> I see nothing wrong with writing C programs on a PC. I do it quite
> often, and not only for calculators.
There is one major problem. With HP49 you can write programs "on the fly" when you need them.
With RPL using objects and stack efficiently it is fairly easy to write rather complicated programs
for advanced problems. It is much more difficult to do so easily with TI Basic and impossible to
do so if you are happen to be away from the computer in C. The main reason to buy advanced
calculator is to be able to perform computations without carrying PC around. Of course, you can
always go back to the PC later and write C program, but the problem might not important anymore
or it might be too late to compute anything useful when you are back at the place where you have PC.
Don't you think, that it would be much more useful to have built in 'C' compiler into the
TI (especially full keyboard TI92 model ) and (hopefully) one, that uses binary floating point math
libraries ? You still can program on the PC if you wish, just like you can do on HP.
And just to close it, for the most of the small "on the fly" programs it is so much faster to write
them in RPL than in either 'C' or TI Basic. Especially 'C' with strong built in type checking
intended to prevent bugs in a large programs seems unnecessarily cumbersome for typical small
in size and complication calculator programs. Beside this, once compiled 'C' program seems
difficult to debug when run on the calculator.
>
> --
> Bhuvanesh
Jack
Parisse Bernard
Bhuvanesh a écrit :
>
> "Veli-Pekka Nousiainen" <DROP...@welho.com> wrote:
>
> > > > Can it use list as programs. You know one can manipulate the lists
>
> Of course you can manipulate lists on the TI-68k, if that's what you
> mean.
>
Some comments:
1/ for Veli-Pekka: about units and EQW, I can't do anything for this,
it's not my part of the code
2/ About the TI-OS. Since I'm currently implementing TI compatibility
for my xcas application (xcas should be able to run TI Basic
programs when it will be finished), I have a better understanding of the
TI OS.
On a mathematical point of view, an area where the code is much more
advanced on the TI is autosimplification. One can regret however that
it is not possible to switch it off (like with QUOTE on the 4xG series).
Some maths fields like linear algebra and arithmetic require additional
user programs.
On a programmation point of view, in my opinion the main weakness are:
* list can not contain list elements
* you can not pass functions as arguments (hence you must cheat with
strings)
* the distinction between function and programs with respect to the
IO screen seems very strange to me
* an unpart instruction should be there
I don't think that not being able to play with TI programs like
lists is a limitation
Otherwise writing programs in TI Basic is easy. It's easier
than RPN where you have to keep track of the stack (which
means that modifications of an existing program are always hard).
In fact, I have adopted TI Basic as one of the modes for programming
xcas
(except that the limitations above are not present:-)).
RPN is also there of course.
Bernard Parisse
Timité Hassan
Here we highly disagree and i don't think that we can agree one day.
The HP CAS give abilities to manipulate symbolic expressions at the
level which can only be almost match when you use the EQW Flash apps
on the TI.
Out the box the TI simply can't.
The autosimplification of the TI simply inputs such manipulations and
the unability to desactive it will always hold back the TI CAS.
And don't tell me that the TI is even remotely as powerful as the HP
for trigonometry,integer and polynomials arithmetics.
About equations that highly depend of the kind of equations.
And in my experience it is the TI which dramatically slow down when
one works with big expressions especially one with complex objects.
> > > > and numeric capabilities than the HP,
> > >
> > > TI is more accurate in floating point.
> >
> > Yes but it is often much slower.
>
> That is in general not true. It is TI Basic that is very slow.
> Most of the built in floating point calculations are usually
> faster than HP. You can see that when ploting functions
> that require complex floating point math to produce answer.
Well,when doing 3D plotting and 3D rotation the HP49 is significantly
faster.
In my experecience the HP is generally faster.
> > Do you know that it is possible to write TI-Basic program barelly
> > slower than the built-in functions for numeric computations ?
> > I have written a TI-Basic function for doing matrix inversions and it
> > is only 50 % slower than the built-in function for 15*15 random
> > matrix.
> > It is twice faster for diagonal matrix !!!
> > Assuming the legendary slowness of the TI-Basic,it is a shame.
>
> Matrix on TI is in fact very slow. For some reason TI implemented matrix
> as always symbolic even if the entire matrix has real or complex numbers.
> That is why when implemented in TI Basic as numerical matrix
> computation you almost reach TI built in speed. But try symbolic
> matrix on TI and on HP and you will find out that often TI outruns
> HP in symbolic matrix despite the fact that TI manages to simplify
> expressions and HP doesn't even try to do so. True, that numerical
> matrix is faster on HP.
I can't believe that you still think that TI software converts numeric
matrix to symbolic to do computations then converts it back to numeric
format before displaying results.
If TI software was able of such high performances for even symbolic
manipulations,i won't complain about TI software because even if it
would still suck for numeric matrix computations,it would be
incredibly powerful for symbolic computations.
The truth is that TI uses unapropriate algorithms for numeric matrix
computations.
Why the hell do you think that they use different algorithm for
numeric and symbolic QR and LU decomposition(read the TI92+ guidebook)
?
If you don't believe me,ask to Bhuvanesh or even to TI engineers if
you can.
The slowness of TI matrix computations is the combo of unapropriate
algorithms and of 16 digits BCD floating point numbers with a CPU
never design to handle BCD computations to begin with.
> > > Motorola 68k is more powerfull than Saturn.
> >
> > Yep,but TI would need to use at least a 68040 to beat the HP49 in
> > speed for everything because the TI engineers are unable to use a
> > decent compiler,the right algorithms and to correctly optimise the
> > software code.
>
> Nonsense. With 68K TI generally outruns HP49 in most functions
> despite the fact, that TI Basic is very slow and floting point is
> implemented in BCD never intended to be mainstream M68K data
> format.
This is what you think.
Btw have you ever use HP49 with the last ROM ?
> > Additionnaly the RAM memory is highly insufficient even for TI-Basic
> > programming.
>
> What do you mean ? Do you need flash RAM or what ? ;-)
No,i mean RAM.
Unless,i am wrong it is not possible to use Flash ROM as virtual
memory on the TI.
> > Well with the number of calculators TI sold,it could produce easily a
> > TI68k with a faster CPU,at least 1 MB of RAM and much more Flash ROM
> > than the Voyage 200 and at even a lower price.
> > Don't forget that all TI calculators are highly overpriced and use
> > almost obsolete hardware(less archaic than HP but,you get the idea).
>
> And how exactly do you know that ? Are you working for the internal TI
> finance departament ? Actually I don't "get an idea". High end calculators are
> not a mass market product as one would think. They are sold in a significantly
> lower numbers than a typical engineering calculator. Short production series
> often drive high cost.
No,i don't work for the TI finance departement but i know that TI does
significant profit over calculators sales(i have heard of dozens of
millions us $ of profits) and sell highly overpriced calculators.
In Europe:
The TI83+ is 15 euros more expensive than the more powerful HP40.
And the TI83+ SE is 10 % more expensive than the significanlty more
powerful Algebra FX 2.0.
The TI89 is almost as expensive as the HP49(13 euros of difference).
By the way the Palm m100 is sold at almost half the price of the TI89.
And you tell me that TI calculators are not overpriced ?
> > > > The second problem is that the HP is much more powerful mathematically,
> > >
> > > Any examples ? This statement is untrue. Each calc has it's strengths and
> > > weaknesses.
> >
> > It happens that out of the box the HP software has much less
> > weaknesses and much more strengths than TI software.
> > The OS core is much more powerful including memory management.
>
> OS core has nothing to do with "mathematically powerfull".
It does.
Do you really think that the unability of TI OS to use bigger than 32
KB objects or over 64 KB for expression stack don't limits the
mathematical power of the TI68k ?
If so explain me why TI89/TI92+ are unable to solve problem that
require less than their availlable RAM to solve.
> > The object oriented design of the HP OS is way superior to the archaic
> > design of the TI OS.
>
> "Object oriented" has nothing to do with "mathematically powerfull".
It does have an effect on computer algebra programming with TI-Basic.
This is why you must give an initial value to local variables before
useing them in TI-Basic programs and functions while you didn't to on
the TI92.
> > The HP User Interface is much more flexible and customisable.
>
> "User interface" has nothing to do with "mathematically powerfull".
Well,in my experience it is much easier and faster to view big
symbolic results on the HP.
And the built-in EQW and Matrix writer enables operations which are
impossible on the TI without the third party EQW.
Have you ever tried to edit big expressions results or Matrix with the
command line ?
A pain in ass,isn't it ?
> > The HP software use most of the time better algorithms and is much
> > more optimised,etc...
>
> Not true. There are examples where TI can solve problems which HP
> cant' and opposite. Especially in integrals it is TI which usually
> does one when HP can't.
I don't limit myself to integrals.
Try for example Taylor series on both calculators.
Let me give you one that the TI68k can't solve because of both wrong
algorithm and archaic memory management:
taylor(tan(tanh(x))-tanh(tan(x)),x,1,7)
Try also polynomials factorisation or matrix computations either
symbolic or numeric.
As you are at that compare both tools without bias for algebra and
calculus then perhaps you will see what i mean.
And about integrals why then the TI is unable to compute the
anti-derivative of functions such as:
f'(x)*exp(f(x)) where f(x) is for example a polynomial ?
> > And overall you can do much more with the HP than with the TI.
> > Only TI engineers and TI worshippers can believe the opposite.
>
> Not true.
Have you really exploited the HP49 or do you use it only as a HP48
replacement ?
> Jack
Veli-Pekka Nousiainen
news:80e7cd6f.02090...@posting.google.com...
> "Marchel" <mar...@comcast.net> wrote in message
news:<_ENc9.107207$On.45...@bin3.nnrp.aus1.giganews.com>...
> > I don't. Out of the box it is TI CAS that is usually more faster and
better.
and thus gives relatively fast school-book examples simplified.
The HP 49G uses a slower, more general RISCH algorithm to integrate.
X
> Out the box the TI simply can't.
and TI-EQW is pretty good.
> The autosimplification of the TI simply inputs such manipulations and
> the unability to desactive it will always hold back the TI CAS.
Unfortunately so. HP is better in many ways.
(Flexible stack-orientation - available even in Algebraic mode
EQW subexpression evaluation with expandable STARTEQW)
> And don't tell me that the TI is even remotely as powerful as the HP
> for trigonometry,integer and polynomials arithmetics.
> About equations that highly depend of the kind of equations.
> And in my experience it is the TI which dramatically slow down when
> one works with big expressions especially one with complex objects.
Both are too slow!
They need a low-power future Play-Station 3 chip or something...
> Well,when doing 3D plotting and 3D rotation the HP49 is significantly
> faster.
That's a clever trick by Cyrille, I guess, but the Zoom is cool !!!
> In my experecience the HP is generally faster.
I think it's slower and requires more from the user
and there is no auto-simplification
AND the CAS forces mode-changes
with a stupid error message if you refuse.
The only solution seems to be using a betaENTER program to
change the flags back, but Vectored Enter doesn't work anymore
in the Algebraic mode because of the new auto-quoting in Alg-mode.
Especially the Alg-mode users (in Finland, where I advice people
on the phone for HP/Radix without a charge [I've a got a free 49G])
need this "keep MY flag settings" feature. Units back to the EQW, please!
BUT
HP 49G is - as you explain - much more powerful and flexible.
X
> The slowness of TI matrix computations is the combo of unapropriate
> algorithms and of 16 digits BCD floating point numbers with a CPU
> never design to handle BCD computations to begin with.
The numerix matrix operations are the salt of the HP 49G !!!
X
> Btw have you ever use HP49 with the last ROM ?
I think that the fastest CAS was 1.14-7
If you have any possibilities to test - say between 1.16 vs. 1.18
(a wild guess) - I would be glad to here the answer.
I remember seeing an old and a new version of Ti vs. HP
where the old HP CAS was faster !!
I don't know what Parisse did between the versions
but maybe some re-arraing of the pre-parser and some ML coding
could make it faster.
Cyrille & Parisse together could do this. (wishful thinking :)
> The TI89 is almost as expensive as the HP49(13 euros of difference).
> By the way the Palm m100 is sold at almost half the price of the TI89.
> And you tell me that TI calculators are not overpriced ?
They are both overpriced, but try to buy a similar CAS environment
for your Palm. How much would it cost? Is it available at all?
Is it programmable or just a black-box solution?
You should not compare apples and stables :-)
Just apples and......lemons!! ;-)
> > OS core has nothing to do with "mathematically powerfull".
> It does.
Yep!
And even the HP OS/Saturn has too limited memory ( and CPU).
> > > The object oriented design of the HP OS is way superior to the archaic
> > > design of the TI OS.
> > "Object oriented" has nothing to do with "mathematically powerfull".
> It does have an effect on computer algebra programming with TI-Basic.
Yes!
I would still put emphasis on stack-oriented design of the HP OS.
> > > The HP User Interface is much more flexible and customisable.
> > "User interface" has nothing to do with "mathematically powerfull".
> Well,in my experience it is much easier and faster to view big
> symbolic results on the HP.
> And the built-in EQW and Matrix writer enables operations which are
> impossible on the TI without the third party EQW.
EQW has it's own matrix writer for a reason...
The in-flexibility of the TI OS/CAS/Programmimg
was the main reason for me to get rid of my TI 89,
which *was* faster/better on integrate/solve.
> I don't limit myself to integrals.
> Try for example Taylor series on both calculators.
To cut it shorter: there are many other things
that the TI 89/92/V200 can't handle as well as the HP 49G.
> > > And overall you can do much more with the HP than with the TI.
> > > Only TI engineers and TI worshippers can believe the opposite.
> > Not true.
> Have you really exploited the HP49 or do you use it only as a HP48
> replacement ?
I would like to ask the same question here,
but when I look into the mirror I found myself not good enough on
the TI 89 programming and OS quirks to give an un-biased statement.
As a former HP 48SX user I found the Reverse Polish Lisp system
better than anything else on a calculator when symbolics are involved.
For numerics I would use perhaps a HP 42S. (I only have a 41CX)
VPN
PS: Sorry for long comments, but I think that this forum is about opinions
and surely this was all calculator related with no stupid humor around.
I just had to defend both calculators - yet to find the 49G better for ME.
Veli-Pekka Nousiainen
Thank you for answering!
Is it possible to speed up the 49G RISCH via some look-up tables?
"Parisse Bernard" <par...@fourier.ujf-grenoble.fr> wrote in message
news:3D76128C...@fourier.ujf-grenoble.fr...
> Bhuvanesh a écrit :
> > "Veli-Pekka Nousiainen" <DROP...@welho.com> wrote:
> > > > > Can it use list as programs. You know one can manipulate the lists
> > Of course you can manipulate lists on the TI-68k, if that's what you
> > mean.
> Some comments:
> 1/ for Veli-Pekka: about units and EQW, I can't do anything for this,
> it's not my part of the code
Sorry! I was barking the wrong tree. Who was it ?!!
> 2/ About the TI-OS.
X
> On a programmation point of view, in my opinion the main weakness are:
> * list can not contain list elements
> * you can not pass functions as arguments (hence you must cheat with
> strings)
> * the distinction between function and programs with respect to the
> IO screen seems very strange to me
> * an unpart instruction should be there
> I don't think that not being able to play with TI programs like
> lists is a limitation
Not a serious one, since one can use string manipulations.
> Otherwise writing programs in TI Basic is easy. It's easier
> than RPN where you have to keep track of the stack (which
> means that modifications of an existing program are always hard).
Local & compiled local variables are available...
> In fact, I have adopted TI Basic as one of the modes for programming
> xcas
> (except that the limitations above are not present:-)).
> RPN is also there of course.
X
Marvelous!! On which PDA platforms it will run on?
The Linux derived Zaurus?? I'm waiting to see it finalized !!
Veli-Pekka
Bhuvanesh
> There is one major problem. With HP49 you can write programs "on the fly"
> when you need them. With RPL using objects and stack efficiently it is fairly
> easy to write rather complicated programs for advanced problems. It is much
> more difficult to do so easily with TI Basic.
Not really (you can easily write complicated programs for advanced
problems). The only problem with TI-Basic is that it is slow.
> And just to close it, for the most of the small "on the fly" programs it is
> so much faster to write them in RPL than in either 'C' or TI Basic.
My experience suggests that it's much easier to write programs "on the
fly" in TI-Basic than in either C or RPL. I know a couple of HP users
who say the same thing.
> Beside this, once compiled 'C' program seems difficult to debug when run on
> the calculator.
Normally, I debug my C programs on VTI before transferring them to the
calculator (because it's lower-level programming and I want to be able
to use my calculator for purposes other than just development :-) ).
--
Bhuvanesh
Bhuvanesh
> On a programmation point of view, in my opinion the main weakness are:
> * list can not contain list elements
Well, there are workarounds for this limitation. Basically, the TIOS
currently does not allow ragged arrays.
> * you can not pass functions as arguments (hence you must cheat with
> strings)
Functions as arguments? Do you mean passing them unevaluated? If not,
you can do, for example: Psi(0, Gamma(0.1)) to get 2.19924...
> * the distinction between function and programs with respect to the
> IO screen seems very strange to me
The basic difference between functions and programs is that functions
cannot have side-effects (they cannot change anything except locally).
> * an unpart instruction should be there
This is on the wish-list:
http://www.angelfire.com/realm/ti_tiplist/WishList.html
> I don't think that not being able to play with TI programs like
> lists is a limitation
> Otherwise writing programs in TI Basic is easy. It's easier
> than RPN where you have to keep track of the stack (which
> means that modifications of an existing program are always hard).
> In fact, I have adopted TI Basic as one of the modes for programming
> xcas
> (except that the limitations above are not present:-)).
Cool! :-)
--
Bhuvanesh
Marchel
> Here we highly disagree and i don't think that we can agree one day.
> The HP CAS give abilities to manipulate symbolic expressions at the
> level which can only be almost match when you use the EQW Flash apps
> on the TI.
I consider autosimplification as an advantage. Lack of it is for me
a serious HP fault as a calculator. Keep in mind, that I do not
consider advanced calculator as a math research tool. For this
I use Mathematica. Calculator is rather a practical tool for the
engineering work. I'm not interested in the complicated unsimplified
formulas that need significant effort and enormous time on HP49 to
reduce them to the readable form.
> Out the box the TI simply can't.
The problem with HP on the other hand is that in most practical cases
it can't simplify or simplification consume so much time, that the user
gives up the effort long before the equation has any "reasonable" form.
Thanks God for the break "ON" key. Otherwise I would throw my HP49
away long time ago.
> The autosimplification of the TI simply inputs such manipulations and
> the unability to desactive it will always hold back the TI CAS.
> And don't tell me that the TI is even remotely as powerful as the HP
> for trigonometry,integer and polynomials arithmetics.
For any practical purposes it is. I don't consider exotic functions rarely
used as a "big deal".
> About equations that highly depend of the kind of equations.
> And in my experience it is the TI which dramatically slow down when
> one works with big expressions especially one with complex objects.
With such equations HP simply dies.
> > That is in general not true. It is TI Basic that is very slow.
> > Most of the built in floating point calculations are usually
> > faster than HP. You can see that when ploting functions
> > that require complex floating point math to produce answer.
>
> Well,when doing 3D plotting and 3D rotation the HP49 is significantly
> faster.
I wasn't talking about 3D plot. 3D plot for all the practicall purposes
is a joke and a toy for kids on both calculators. Have you
ever used your calculators for any real work or you just play
with the new toy ?
> In my experecience the HP is generally faster.
In mine the opposite is true (keep in mind, that I use HP everyday
for other advantages).
> I can't believe that you still think that TI software converts numeric
> matrix to symbolic to do computations then converts it back to numeric
> format before displaying results.
Compare the timing on simple symbolic and real, small arrays. The
timing is very similar. Even if the same algorithm is used
the real array should be many times faster (dur to the automatic
reduction) than symbolic but it is not. Please, explain.
> If TI software was able of such high performances for even symbolic
> manipulations,i won't complain about TI software because even if it
> would still suck for numeric matrix computations,it would be
> incredibly powerful for symbolic computations.
And it is. It outruns HP in symbolic matrix math easily.
> > Nonsense. With 68K TI generally outruns HP49 in most functions
> > despite the fact, that TI Basic is very slow and floting point is
> > implemented in BCD never intended to be mainstream M68K data
> > format.
>
> This is what you think.
> Btw have you ever use HP49 with the last ROM ?
Of course. I use it everyday.
> > > Additionnaly the RAM memory is highly insufficient even for TI-Basic
> > > programming.
> >
> > What do you mean ? Do you need flash RAM or what ? ;-)
>
> No,i mean RAM.
Please explain, why RAM is insufficient for programming ? My PC runs
programs on RAM just fine.
> Unless,i am wrong it is not possible to use Flash ROM as virtual
> memory on the TI.
It would be a nonsenes to use flash RAM as a virtual memory. Flash RAM
is not only slower than regular RAM but also more expensive.
> No,i don't work for the TI finance departement but i know that TI does
> significant profit over calculators sales(i have heard of dozens of
> millions us $ of profits) and sell highly overpriced calculators.
Unfortunately the problem with rumors is that those are usually not true.
Corporations do not publish their profits made on the particular product
and it is impossible to know it outside of the financial departament of such
company.
> In Europe:
> The TI83+ is 15 euros more expensive than the more powerful HP40.
> And the TI83+ SE is 10 % more expensive than the significanlty more
> powerful Algebra FX 2.0.
Pontiac Grand Prix GTP is more powerfull than BMW 528 but is cost
significantly less. The power of the machine is only part of the equation.
> The TI89 is almost as expensive as the HP49(13 euros of difference).
> By the way the Palm m100 is sold at almost half the price of the TI89.
> And you tell me that TI calculators are not overpriced ?
Palm is sold without any advanced software and you are talking
only about models that are actually on sale, because Palm Inc
is getting rid of them. New Palm models cost more than advanced
calc. Competing windows CE machines usually cost twice as much as any
advanced calculator in the cheapest version. Beside this
plams are sold in much larger series than advanced calculators.
Mass production lowers the cost since Ford Model T.
> > OS core has nothing to do with "mathematically powerfull".
>
> It does.
> Do you really think that the unability of TI OS to use bigger than 32
> KB objects or over 64 KB for expression stack don't limits the
> mathematical power of the TI68k ?
I don't. It's exotic case, proof of using inappropriate tool for the job.
For such large expression you should use PC based software.
Even if you can generate such expression on the HP49
it will probably take 10 minutes or better for a single command
when you trying to do anything with it.
> If so explain me why TI89/TI92+ are unable to solve problem that
> require less than their availlable RAM to solve.
Why HP is unable to integrate many verys simple itegrals ?
> > "Object oriented" has nothing to do with "mathematically powerfull".
>
> It does have an effect on computer algebra programming with TI-Basic.
> This is why you must give an initial value to local variables before
> useing them in TI-Basic programs and functions while you didn't to on
> the TI92.
Usage of the uninitialized variables is a big "no-no" in any
advanced programming. Most of the modern compilers will issue
a warning of bad programming practice if they discover uninitialized
variable during compile time. Ask any programmer if you don't belive.
> > > The HP User Interface is much more flexible and customisable.
> >
> > "User interface" has nothing to do with "mathematically powerfull".
>
> Well,in my experience it is much easier and faster to view big
> symbolic results on the HP.
ROTFL. First you need to simplify them. After several minutes of
HP "gard work" maybe you will be able to see them in a readable form.
> And the built-in EQW and Matrix writer enables operations which are
> impossible on the TI without the third party EQW.
> Have you ever tried to edit big expressions results or Matrix with the
> command line ?
Actually that is the only practical way to simplify the large equations
within reasonable time on HP. Of course, one has to abandon "built
in" useless tools for simplification due to their total uselessness
and edit text directly using old "paper and pencil" approach and it's
own head.
> A pain in ass,isn't it ?
It is. I prefer, when calculator does it for me, like TI92 does.
Jack
Bernard Parisse
> Hi, Bernard!
>
> Thank you for answering!
> Is it possible to speed up the 49G RISCH via some look-up tables?
>
Most certainly, but I won't do it myself. Perhaps someone else will
when the CAS is released under LGPL at the end of the next year.
>>1/ for Veli-Pekka: about units and EQW, I can't do anything for this,
>>it's not my part of the code
>
> Sorry! I was barking the wrong tree. Who was it ?!!
>
EQW is mainly Gerald's work.
>>I don't think that not being able to play with TI programs like
>>lists is a limitation
>
> Not a serious one, since one can use string manipulations.
>
I meant making programs that modify other programs should
not be a common practice (i.e. only advanced programmer
should have to do that and not often).
>
>>Otherwise writing programs in TI Basic is easy. It's easier
>>than RPN where you have to keep track of the stack (which
>>means that modifications of an existing program are always hard).
>
> Local & compiled local variables are available...
>
Yes, but the slow CPU on which SRPL is implemented has
given birth to a lot of programs that use as little as
possible local variables (there was also a memory reason for
this on the 48). These programs which heavily use stack
manipulation are much harder to modify. And for maths
manipulation it is easier to use algebraic notation
(coding, checking, modifying).
> Marvelous!! On which PDA platforms it will run on?
> The Linux derived Zaurus?? I'm waiting to see it finalized !!
>
It runs on every platform where XWindows is installed (i.e. Linux
but also MacOSX once XDarwin is installed), as well
as on desktop Windows. I run it on an ipaq with Familiar Linux
installed, you can run it on the Zaurus (once you have installed
X11).
Bernard Parisse
>>* list can not contain list elements
>
>
> Well, there are workarounds for this limitation. Basically, the TIOS
> currently does not allow ragged arrays.
>
Of course one can always make workarounds but it makes programs tricky
harder to modify etc. I do not understand this limitation because
internally a TI list can contain any type of object, it's probably just
the parser, something that should be fixed.
>
> Functions as arguments? Do you mean passing them unevaluated? If not,
> you can do, for example: Psi(0, Gamma(0.1)) to get 2.19924...
>
I mean something like
{1,2,3} => l
Define f(x)=Func: ... EndFunc
map(f,l)
where map would apply f to all elements of the list.
f is evaluated and you will get argument error (even if you
define map(f,l) to return f(l)).
Just a quote instruction could fix that. TI-Basic is currently
not a fonctionnal language, that's really a shame because fixing
this is probably 1/4 h work for the OS programmers.
> The basic difference between functions and programs is that functions
> cannot have side-effects (they cannot change anything except locally).
>
Then a program should be usable like a function, i.e. return a value.
>>In fact, I have adopted TI Basic as one of the modes for programming
>>xcas
>>(except that the limitations above are not present:-)).
>
>
> Cool! :-)
BTW, do you know where the binary format of TI is documented? Or will
the libtifile give an API to read these files?
Veli-Pekka Nousiainen
> In mine the opposite is true (keep in mind, that I use HP everyday
> for other advantages).
Which are (the other advantages) ??
VPN
Timité Hassan
> "Timité Hassan" <timi...@yahoo.fr> wrote in message news:80e7cd6f.02090...@posting.google.com...
>
> > Here we highly disagree and i don't think that we can agree one day.
> > The HP CAS give abilities to manipulate symbolic expressions at the
> > level which can only be almost match when you use the EQW Flash apps
> > on the TI.
>
> I consider autosimplification as an advantage. Lack of it is for me
> a serious HP fault as a calculator. Keep in mind, that I do not
> consider advanced calculator as a math research tool. For this
> I use Mathematica. Calculator is rather a practical tool for the
> engineering work. I'm not interested in the complicated unsimplified
> formulas that need significant effort and enormous time on HP49 to
> reduce them to the readable form.
This is why,i think that we can't agree.
To counter what you say,the TI autosimplification can generates awful
results and takes much more time than even the worse cases of the HP.
I guess that you haven't work a lot with expressions involving complex
and square roots on the TI92+.
I have done it and i can tell you that the results are awful.
> > Out the box the TI simply can't.
>
> The problem with HP on the other hand is that in most practical cases
> it can't simplify or simplification consume so much time, that the user
> gives up the effort long before the equation has any "reasonable" form.
> Thanks God for the break "ON" key. Otherwise I would throw my HP49
> away long time ago.
It is true that the HP doesn't have built-in advanced
autosimplification but
i prefer doing simplification myself.
Sometimes,it is painful but at least you can control what you are
doing.
For example sometimes you need results with complex numbers at the
denominator(expansion of rationnal fractions for example) and it is
impossible to achieve it on the TI.
> > The autosimplification of the TI simply inputs such manipulations and
> > the unability to desactive it will always hold back the TI CAS.
> > And don't tell me that the TI is even remotely as powerful as the HP
> > for trigonometry,integer and polynomials arithmetics.
>
> For any practical purposes it is. I don't consider exotic functions rarely
> used as a "big deal".
To convert an expression with sincos to an expression with tan or
halftan is exotic,perhaps ?
What about having a complex expression not splitted between the real
and the imaginary part ?
It is also exotic ?
> > About equations that highly depend of the kind of equations.
> > And in my experience it is the TI which dramatically slow down when
> > one works with big expressions especially one with complex objects.
>
> With such equations HP simply dies.
I don't think so.
Just try to do some computations with 2*2 or 3*3 matrix whom elements
contains both square roots and complex numbers and tell me how fast is
the TI with them.
> I wasn't talking about 3D plot. 3D plot for all the practicall purposes
> is a joke and a toy for kids on both calculators. Have you
> ever used your calculators for any real work or you just play
> with the new toy ?
I never play with calculators.
If i wanted to play,i would use PC CAS.
3D plotting on calculators are interesting to quickly have an idea of
what a 3D plot look like.
For people without computers or without specific applications,it is
probably the only way to work with 3D plots.
Unless you think that they haven't the right to ?
> > In my experecience the HP is generally faster.
>
> In mine the opposite is true (keep in mind, that I use HP everyday
> for other advantages).
I was talking of numeric computations.
> > I can't believe that you still think that TI software converts numeric
> > matrix to symbolic to do computations then converts it back to numeric
> > format before displaying results.
>
> Compare the timing on simple symbolic and real, small arrays. The
> timing is very similar. Even if the same algorithm is used
> the real array should be many times faster (dur to the automatic
> reduction) than symbolic but it is not. Please, explain.
What arrays are you talking of ?
2*2 or 3*3 ?
Because with 10*10 numeric matrix:
the built-in function takes 23 s.
With the method you are talking of the TI92+ HW1 need at least 17 min
to convert the matrix to an exact one,compute the rest and computes a
numeric approximation of the results.
Even with 5*5 numeric matrix:
the built-in function takes 3 s
your method takes 33 s.
Btw i don't use matrix generates by randmat but a matrix generates by
randmat than mutiply by 101./109. Thus you get a matrix whom each
elements has at 13 digits after the dot.
The TI-Cas used the same algorithm for symbolic and numeric matrix
computations(besides for LU,QR decompositions and eigvc,eigvl).
But with numeric matrix the TI software use floating point arithmetic.
> > If TI software was able of such high performances for even symbolic
> > manipulations,i won't complain about TI software because even if it
> > would still suck for numeric matrix computations,it would be
> > incredibly powerful for symbolic computations.
>
> And it is. It outruns HP in symbolic matrix math easily.
No,kidding with an over 10 times faster CPU,it is even curious that
the HP49 was still able to outperform it for some symbolic
computations.
> Please explain, why RAM is insufficient for programming ? My PC runs
> programs on RAM just fine.
The 256 KB of RAM of the TI68k are insufficient for the kind of
programs
i used to write in TI-basic because some of these programs generates
object which can't be stored in the Flash ROM as they needed to be
modified.
Additionnaly some of the Flash apps use significant amount of RAM.
The Voyage 200 has almost only 128 KB of RAM availlable because of
some preloaded Flash apps.
> > No,i don't work for the TI finance departement but i know that TI does
> > significant profit over calculators sales(i have heard of dozens of
> > millions us $ of profits) and sell highly overpriced calculators.
>
> Unfortunately the problem with rumors is that those are usually not true.
> Corporations do not publish their profits made on the particular product
> and it is impossible to know it outside of the financial departament of such
> company.
I can't give you my sources(last times i did one of them has had some
troubles with HP) but they are very reliable.
> > > OS core has nothing to do with "mathematically powerfull".
> >
> > It does.
> > Do you really think that the unability of TI OS to use bigger than 32
> > KB objects or over 64 KB for expression stack don't limits the
> > mathematical power of the TI68k ?
>
> I don't. It's exotic case, proof of using inappropriate tool for the job.
> For such large expression you should use PC based software.
> Even if you can generate such expression on the HP49
> it will probably take 10 minutes or better for a single command
>when you trying to do anything with it.
I am not talking of exotic cases.
Because of autosimplification and/or wrong alogrithms some pretty
simple problems can generates incredibly big intermediate results and
thus a better memory management and more RAM become very important.
If you have already done computer algebra programming,you should know
what i am talking of.
> > If so explain me why TI89/TI92+ are unable to solve problem that
> > require less than their availlable RAM to solve.
>
> Why HP is unable to integrate many verys simple itegrals ?
Alogrithms problem,nothing to do with what we are talking of here.
> > > "Object oriented" has nothing to do with "mathematically powerfull".
> >
> > It does have an effect on computer algebra programming with TI-Basic.
> > This is why you must give an initial value to local variables before
> > useing them in TI-Basic programs and functions while you didn't to on
> > the TI92.
>
> Usage of the uninitialized variables is a big "no-no" in any
> advanced programming. Most of the modern compilers will issue
> a warning of bad programming practice if they discover uninitialized
> variable during compile time. Ask any programmer if you don't belive.
There you are comparing apples and oranges.
Computer algebra programming is quite different of other kind
programmings.
Do you use enavaluated results in something else than computer algebra
?
> > > > The HP User Interface is much more flexible and customisable.
> > >
> > > "User interface" has nothing to do with "mathematically powerfull".
> >
> > Well,in my experience it is much easier and faster to view big
> > symbolic results on the HP.
>
> ROTFL. First you need to simplify them. After several minutes of
> HP "gard work" maybe you will be able to see them in a readable form.
No comment.
> > And the built-in EQW and Matrix writer enables operations which are
> > impossible on the TI without the third party EQW.
> > Have you ever tried to edit big expressions results or Matrix with the
> > command line ?
>
> Actually that is the only practical way to simplify the large equations
> within reasonable time on HP. Of course, one has to abandon "built
> in" useless tools for simplification due to their total uselessness
> and edit text directly using old "paper and pencil" approach and it's
> own head.
My words would go beyond my thoughts so i prefer just say:
No comment.
> > A pain in ass,isn't it ?
>
> It is. I prefer, when calculator does it for me, like TI92 does.
Perhaps the TI92 is able to edit expressions and matrix without you
typeing on the keyboard ?
Sorry,i couldn't resist.
> Jack
Timité Hassan
> "Timité Hassan" <timi...@yahoo.fr> wrote in message
> news:80e7cd6f.02090...@posting.google.com...
> > "Marchel" <mar...@comcast.net> wrote in message
> news:<_ENc9.107207$On.45...@bin3.nnrp.aus1.giganews.com>...
> X
> > > I don't. Out of the box it is TI CAS that is usually more faster and
> better.
> If I understand correctly the TI Derive uses large look-up tables
> and thus gives relatively fast school-book examples simplified.
> The HP 49G uses a slower, more general RISCH algorithm to integrate.
The Derive software is more capable than the TI CAS for integrations.
I wouldn't be surprised if it also use an implementation of the Risch
algorithm.
Mathematic a uses both look-up tables and risch algorithm,i think.
> > Out the box the TI simply can't.
> Yes, but the out-of-the-box HP has some problems, too
> and TI-EQW is pretty good.
Yes,it is.
It is a shame it doesn't come built-in,it would highly increase the
power of the TI out of the box.
> > The autosimplification of the TI simply inputs such manipulations and
> > the unability to desactive it will always hold back the TI CAS.
> Unfortunately so. HP is better in many ways.
> (Flexible stack-orientation - available even in Algebraic mode
> EQW subexpression evaluation with expandable STARTEQW)
The worst with the TI is the impossibilit to get the result in a
different form than the autosimplified result.
> Both are too slow!
> They need a low-power future Play-Station 3 chip or something...
TI could use a low power 68040.
In case of the HP the only solution is to port the software or emulate
the tool on extremly powerful CPU.
> > Well,when doing 3D plotting and 3D rotation the HP49 is significantly
> > faster.
> That's a clever trick by Cyrille, I guess, but the Zoom is cool !!!
I agree;)
> > In my experecience the HP is generally faster.
> I think it's slower and requires more from the user
> and there is no auto-simplification
> AND the CAS forces mode-changes
> with a stupid error message if you refuse.
I was talking of numeric computations.
The TI has the potential to be faster but with TI engineers
laziness,it is hopeless.
> > The slowness of TI matrix computations is the combo of unapropriate
> > algorithms and of 16 digits BCD floating point numbers with a CPU
> > never design to handle BCD computations to begin with.
> The numerix matrix operations are the salt of the HP 49G !!!
> X
> > Btw have you ever use HP49 with the last ROM ?
> I think that the fastest CAS was 1.14-7
> If you have any possibilities to test - say between 1.16 vs. 1.18
> (a wild guess) - I would be glad to here the answer.
No,i haven't sorry.
> I remember seeing an old and a new version of Ti vs. HP
> where the old HP CAS was faster !!
> I don't know what Parisse did between the versions
> but maybe some re-arraing of the pre-parser and some ML coding
> could make it faster.
> Cyrille & Parisse together could do this. (wishful thinking :)
>
> > The TI89 is almost as expensive as the HP49(13 euros of difference).
> > By the way the Palm m100 is sold at almost half the price of the TI89.
> > And you tell me that TI calculators are not overpriced ?
> They are both overpriced, but try to buy a similar CAS environment
> for your Palm. How much would it cost? Is it available at all?
> Is it programmable or just a black-box solution?
> You should not compare apples and stables :-)
> Just apples and......lemons!! ;-)
It is true that there is no CAS for the Palm OS.
But 99 euros is quite cheap for the hardware.
> > > > The object oriented design of the HP OS is way superior to the archaic
> > > > design of the TI OS.
> > > "Object oriented" has nothing to do with "mathematically powerfull".
> > It does have an effect on computer algebra programming with TI-Basic.
> Yes!
> I would still put emphasis on stack-oriented design of the HP OS.
>
> > > > The HP User Interface is much more flexible and customisable.
> > > "User interface" has nothing to do with "mathematically powerfull".
> > Well,in my experience it is much easier and faster to view big
> > symbolic results on the HP.
> > And the built-in EQW and Matrix writer enables operations which are
> > impossible on the TI without the third party EQW.
> EQW has it's own matrix writer for a reason...
> The in-flexibility of the TI OS/CAS/Programmimg
> was the main reason for me to get rid of my TI 89,
> which *was* faster/better on integrate/solve.
>
> > I don't limit myself to integrals.
> > Try for example Taylor series on both calculators.
> To cut it shorter: there are many other things
> that the TI 89/92/V200 can't handle as well as the HP 49G.
I agree !
> > > > And overall you can do much more with the HP than with the TI.
> > > > Only TI engineers and TI worshippers can believe the opposite.
> > > Not true.
> > Have you really exploited the HP49 or do you use it only as a HP48
> > replacement ?
>
> I would like to ask the same question here,
> but when I look into the mirror I found myself not good enough on
> the TI 89 programming and OS quirks to give an un-biased statement.
> As a former HP 48SX user I found the Reverse Polish Lisp system
> better than anything else on a calculator when symbolics are involved.
> For numerics I would use perhaps a HP 42S. (I only have a 41CX)
I have been useing TI68k for over 7 years and i have a very deep
knowledge of those tools as i have used any model from the very first
TI92(released in october 1995 with beta ROM 1.0) to the Voyage 200.
So i perfectly know what i am talking.
I have started used HP48 a few months before getting my first TI92 so
i know quite well the HP4x calculators.
> VPN
> PS: Sorry for long comments, but I think that this forum is about opinions
> and surely this was all calculator related with no stupid humor around.
> I just had to defend both calculators - yet to find the 49G better for ME.
I think that i have wasted too much time trying to find excuses for TI
so now i don't hold my words anymore about what i think toward them
and their products.
Note that i have a very poor opinion of HP but it happens that i have
a lot of respect for HP ACO members and Bernard parisse thus i am
probably biased toward the HP49.
I will try to be fair,in the future.
rcobo
i put a pic running on the zaurus here:
http://www.geocities.com/zauborg/
it is GREAT! as functional as in the PC!
> It runs on every platform where XWindows is installed (i.e. Linux
> but also MacOSX once XDarwin is installed), as well
> as on desktop Windows. I run it on an ipaq with Familiar Linux
> installed, you can run it on the Zaurus (once you have installed
> X11).
the zaurus will be able to play XCAS from qtopia soon (with the
software qtopia-x)
regards
PS: Proffesor. in the zaurus, the right side of the display does not
show up all the way. Maybe the ipaq display is slightly wider. it does
not affect the functionality, but it eats the right last menu in half
(still, the menu works fine). i tried to move the window with no luck.
Marchel
> >(keep in mind, that I use HP everyday for other advantages).
> X
> Which are (the other advantages) ??
> VPN
Ease of quick "on the fly" programming.
Portability. My TI92 doesn't fit the pocket :(
Jack
Marchel
> "Marchel" <mar...@comcast.net> wrote in message news:<rHzd9.432958$2p2.17...@bin4.nnrp.aus1.giganews.com>...
> > "Timité Hassan" <timi...@yahoo.fr> wrote in message news:80e7cd6f.02090...@posting.google.com...
> This is why,i think that we can't agree.
> To counter what you say,the TI autosimplification can generates awful
> results and takes much more time than even the worse cases of the HP.
> I guess that you haven't work a lot with expressions involving complex
> and square roots on the TI92+.
> I have done it and i can tell you that the results are awful.
To the contrary. Try to solve the following problem.
Imagine for example that you have three points on the ellipse that
is described by the equation:
x^2/a^2 + y^2/b^2 = 1
Those points are for any given parameter "t" located as follows (rectangular coordinates):
P1: x1 = a*cos(t); y1 = b*sin(t)
P2: x2 = a*cos(t+2*pi/3); y2 = b*sin(t+2*pi/3)
P3: x3 = a*cos(t-2*pi/3); y3 = b*sin(t-2*pi/3)
where xn,yn are coordinates for each point, t is the angular parameter
and "a" and "b" are main axis of the ellipse. Assuming (skip the proof)
that for any given value of the parameter t there is exactly one point P
such that when connected with the points P1,P2 and P3 with lines, the
angles between each combination of the three lines is equal 2*pi/3. Now
find an equation of the curve that is the set of all such points "P" for all
possible values of angular parameter "t".
Such problem is relatively easy to solve TI89/92 and with autosimplification and
quickly leads to the solution. With HP49 good luck. I hope, you can do simple
analytical geometry up to conical sections curves.
By the way, this is a real problem in mechanical engineering and affects dynamical
imbalance of the front wheel drive autmotive halfshafts that use specific inner
CV joint.
For the fun of it also try to find out the angle that is between line created by points P-P1
ands axis "x". This might give you a clue, why such joint is a true kinematical
CV joint but not exactly dynamically balanced when exposed to some angle.
Universal (Cardan) joint does the opposite.
Jack
Parisse Bernard
> Imagine for example that you have three points on the ellipse that
> is described by the equation:
>
> x^2/a^2 + y^2/b^2 = 1
>
> Those points are for any given parameter "t" located as follows (rectangular coordinates):
>
> P1: x1 = a*cos(t); y1 = b*sin(t)
> P2: x2 = a*cos(t+2*pi/3); y2 = b*sin(t+2*pi/3)
> P3: x3 = a*cos(t-2*pi/3); y3 = b*sin(t-2*pi/3)
>
> where xn,yn are coordinates for each point, t is the angular parameter
> and "a" and "b" are main axis of the ellipse. Assuming (skip the proof)
> that for any given value of the parameter t there is exactly one point P
> such that when connected with the points P1,P2 and P3 with lines, the
> angles between each combination of the three lines is equal 2*pi/3. Now
> find an equation of the curve that is the set of all such points "P" for all
> possible values of angular parameter "t".
>
> Such problem is relatively easy to solve TI89/92 and with autosimplification and
> quickly leads to the solution. With HP49 good luck. I hope, you can do simple
> analytical geometry up to conical sections curves.
>
Could you give the method you are using on a Ti89/92?
I have solved the problem with a desktop software and it seems not
really
feasible with a calculator.
The method I used is naive. First set a:=1 which does not change the
problem
Then trig expand the coordinates of P2 and P3. Let M(x,y) be the
solution,
then we have to solve
d(M,P1)^2*d(M,P2)^2=4*scalar_product(vector(M,P1),vector(M,P2))
and the same equation with P2 replaced by P3
(taking squares remove all sqrt except sqrt(3) and we have a polynomial
system that
symbolic solvers can reduce using Groebner basis)
Hence the system, after adding and substracting the 2 equations to
remove the sqrt(3)
(6*(cos(t))^4*b^4+(-(12*(cos(t))^4))*b^2+6*(cos(t))^4+24*(cos(t))^3*b^2*x+(-(24*(cos(t))^3))*x+(-(3*(cos(t))^2))*b^4+24*(cos(t))^2*b^3*sin(t)*y+(-(18*(cos(t))^2))*b^2*x^2+(-(18*(cos(t))^2))*b^2*y^2+12*(cos(t))^2*b^2+(-(24*(cos(t))^2))*b*sin(t)*y+18*(cos(t))^2*x^2+18*(cos(t))^2*y^2-9*(cos(t))^2+(-(30*cos(t)))*b^2*x+12*cos(t)*x^3+12*cos(t)*x*y^2+18*cos(t)*x-3*b^4+(-(6*b^3))*sin(t)*y+21*b^2*x^2+9*b^2*y^2+3*b^2+12*b*x^2*sin(t)*y+12*b*sin(t)*y^3+(-(6*b))*sin(t)*y-12*x^4+(-(24*x^2))*y^2-9*x^2-12*y^4+3*y^2)/2
=0
-3*(cos(t))^3*sin(t)*b^4-(-(6*(cos(t))^3))*sin(t)*b^2-3*(cos(t))^3*sin(t)-(-(3*cos(t)))*sin(t)*b^4-9*cos(t)*sin(t)*b^2*x^2-9*cos(t)*sin(t)*b^2*y^2-3*cos(t)*sin(t)*b^2-(-(9*cos(t)))*sin(t)*x^2-(-(9*cos(t)))*sin(t)*y^2-(-(6*cos(t)))*b*x^2*y-(-(6*cos(t)))*b*y^3-6*cos(t)*b*y-(-(6*sin(t)))*b^2*x-6*sin(t)*x^3-6*sin(t)*x*y^2=0
It requires about 5 s. and 3M of RAM to solve with maple, therefore I
would consider
this type of problem as unsolvable with a calc with around 200K of RAM.
Unless
there is some trick to make the problem easier...
BTW, when comparing TI and HP CAS you often said that there are many
"exotic" functions on the 49. Could you be more precise?
Marchel
> Could you give the method you are using on a Ti89/92?
> I have solved the problem with a desktop software and it seems not
> really
> feasible with a calculator.
Not for TI89/92. You get an answer in about 10 minutes.
The steps entered onto the calculator are marked below by (#)
Here is the method:
First assume, that you have three (non coincident) points on the plane P1, P2, P3.
The points have rectangular coordinates:
P1: (XX1, YY1); P2: (XX2, YY2); P3: (XX3, YY3)
Through the point P1 put a line in the "angular direction" of s. The parametric equation of the line (parameter u) is:
XP1 = XX1 + u * cos(s)
YP1 = YY1 + u * sin(s)
Through the point P2 put the line in the "angular direction" s + 2 * PI / 3. The parametric equation of the second line
(parameter v) is:
XP2 = XX2 + v * cos(s + 2 * PI / 3)
YP2 = YY2 + v * sin(s + 2 * PI / 3)
Through the point P3 put the line in the "angular direction" s - 2 * PI / 3. The parametric equation of the third line
(parameter w) is:
XP3 = XX3 + w * cos(s + 2 * PI / 3)
YP3 = YY3 + w * sin(s + 2 * PI / 3)
It's easy to notice, that the lines are at 120 degrees (2*PI/3) to each other.
The goal is to find such direction s, that the three above lines intersect in one point.
Above equations are sufficient to find it. They have four unknowns: u,v,w and s.
To find s, one has to solve four following equations:
XP1 = XP2
YP1 = YP2
XP1 = XP3
YP1 = YP3
Switch on TI92+ and enter the equations as follows:
Step 01 # tExpand( XX1 + u * cos(s) - ( XX2 + v * cos(s + 2 * PI / 3) ) = 0 ) * 2 STO> e1
Step 02 # tExpand( YY1 + u * sin(s) - ( YY2 + v * sin(s + 2 * PI / 3) ) = 0 ) * 2 STO> e2
Step 03 # tExpand( XX1 + u * cos(s) - ( XX3 + w * cos(s - 2 * PI / 3) ) = 0 ) * 2 STO> e3
Step 04 # tExpand( YY1 + u * sin(s) - ( YY3 + w * sin(s - 2 * PI / 3) ) = 0 ) * 2 STO> e4
The tExpand and multiplication by 2 is to simplify a little bit the equations (is not neccessary
but usefull). Once we entered the equations, let's enter the coordinates of P1, P2, and P3 as
stated in the original poblem.
Step 05 # a * cos( t ) STO> XX1
Step 06 # b * sin( t ) STO> YY1
Step 07 # tExpand( a * cos( t + 2 * PI / 3 ) ) STO> XX2
Step 08 # tExpand( b * sin( t + 2 * PI / 3 ) ) STO> YY2
Step 09 # tExpand( a * cos( t - 2 * PI / 3 ) ) STO> XX3
Step 10 # tExpand( b * sin( t - 2 * PI / 3 ) ) STO> YY3
Now let's solve.
Step 11 # Solve( e1 , u )
Step 12 # Define [here bring just found solution for u from the history using keyboard]
Step 13 # Solve( e2 , v )
Step 14 # Define [here bring just found solution for v from the history using keyboard]
The 'u' that was found in step 11 was expressed in terms of v. Bring up u to remove v from
it's defining equation. Altough this step is not neccessary, it will speed up the further process.
Step 15 # u
Step 16 # Ans(1) STO> u
Step 17 # Solve( e3 , w )
Step 18 # Define [here bring just found solution for w from the history using keyboard]
Once the u,v,w was solved, bring up equation e4 (defined in step 4) to look at it.
Step 19 # e4 <ENTER>
The equation has a form:
6*(a+b)*(cos(s)*sin(t)-sin(s)*cos(t))/(cos(s)-sqrt(3)*sin(s)) = 0
As can be easily seen (thanks to TI92 simplification) the denominator can be removed.
This step is not neccessary, but speeds up the problem solving.
I skip the exception when cos(s) = sqrt(3)*sin(s) - it's an engineering problem, not
mathematical ;-)
Step 20 # Ans(1) * ( cos( s ) - sqrt( 3 ) * sin( s ) ) STO> e4
Now it's time to solve for direction s:
Step 21 # Solve( e4 , s )
The answer is in the form of:
s = mod( 2 * t - PI , 2 * PI ) / 2 + @n1 * PI - PI / 2 or a + b = 0
Assuming t to be within "reasonable engineering limits" we can skip modulus and assuming particular
solution @n1 = 1 the solution simplfies to the form:
s = t
The above is a proof, that the joint is a CV joint (the lines rotate by the same angle as the
points projected onto the ellipse. Let's use the above solution for s:
Step 22 # t STO> s
Because we already solved for u and s, the common intersection point can be found from the
very first two equations listed in this posting:
Step 23 # XX1 + u * cos( s ) STO> x
Step 24 # YY1 + u * sin( s ) STO> y
Thanks to the power of TI ;-) the x and y are nicely simplified and expressed:
x = (a-b)*(4*(sin(t))^2-1)*cos(t)/2
y = -(a-b)*sin(t)*(4*(cos(t))^2-1)/2
The last step is a "lucky" guess by observing the above equations (but one can assume it and test for it)
Step 25 # x ^ 2 + y ^ 2 <ENTER>
Ta - Da !!! :-)
IT'S A CIRCLE OF a radius of (a-b)/2
which indicates, that the ougoing end of shaft of that CV joint "orbits" a circle thereof creating vibrations.
I don't know how much RAM it actually used, but 200K was fully sufficient to solve it and the solving time
was short, mostly limited by my speed of typing. Trying to do so with HP49 is probably possible but I
gave up after several hours. I don't think so, that the above was "a trick". It's pretty straigtforward
use of analytical geometry using parametric equations for the line.
The above is a simple theory for the straight plunging tripod CV joint shown here:
http://cvcoupling.com/hisory_cvj.html
Jack
parisse
>
> x = (a-b)*(4*(sin(t))^2-1)*cos(t)/2
> y = -(a-b)*sin(t)*(4*(cos(t))^2-1)/2
>
> The last step is a "lucky" guess by observing the above equations (but one can assume it and test for it)
>
> Step 25 # x ^ 2 + y ^ 2 <ENTER>
>
> Ta - Da !!! :-)
>
> IT'S A CIRCLE OF a radius of (a-b)/2
>
> which indicates, that the ougoing end of shaft of that CV joint "orbits" a circle thereof creating vibrations.
>
> I don't know how much RAM it actually used, but 200K was fully sufficient to solve it and the solving time
> was short, mostly limited by my speed of typing. Trying to do so with HP49 is probably possible but I
> gave up after several hours. I don't think so, that the above was "a trick". It's pretty straigtforward
> use of analytical geometry using parametric equations for the line.
>
than my naive method. Which shows (again) than maths will ease
the solution. However I see no reason why you claim the 49 can
not solve the problem following the same steps. Your method does
not require any intensive computation. It's linear system solving
(solve step 1, 2, 3+replace in 4 and multiply by the determinant
of 1,2,3) and trigonometric rewriting. You can make step 25 without
guess since TLINerizing the equations will show you that the
solution describes a circle at 3* the speed of the original ellipsis.
It's probable that the 49 will be slower than the 89/92 during
the process because there is no polynomial gcd computation involved.
Could you answer to my question about what you call exotic functions
on the 49? Thanks
Bernard Parisse
>
> Step 01 # tExpand( XX1 + u * cos(s) - ( XX2 + v * cos(s + 2 * PI / 3) ) = 0 ) * 2 STO> e1
> Step 02 # tExpand( YY1 + u * sin(s) - ( YY2 + v * sin(s + 2 * PI / 3) ) = 0 ) * 2 STO> e2
> Step 03 # tExpand( XX1 + u * cos(s) - ( XX3 + w * cos(s - 2 * PI / 3) ) = 0 ) * 2 STO> e3
> Step 04 # tExpand( YY1 + u * sin(s) - ( YY3 + w * sin(s - 2 * PI / 3) ) = 0 ) * 2 STO> e4
>
more linear algebra and in RPN mode. Computing time is less than
3 minutes.
Take a=1, let B=b/a (does not change the problem by dilatation)
Enter with matrix editor the matrix of the system with respect to u,v,w
(4*4 matrix since there are 4 equations)
[[ 'COS(S)' 0 'COS(S-2*PI/3)' 'COS(T-2*PI/3)-COS(T)']
[ 'SIN(S)' 0 'SIN(S-2*PI/3)' 'B*(SIN(T-2*PI/3)-SIN(T))' ]
[ 'COS(S)' 'COS(S+2*PI/3)' 0 'COS(T+2*PI/3)-COS(T)' ]
[ 'SIN(S)' 'SIN(S+2*PI/3)' 0 'B*(SIN(T+2*PI/3)-SIN(T))' ]]
TEXPAND 'M' STO (4.5s)
DET (31s)
FACTOR (21s)
Looking at the factored expression, it is immediate (make a TLIN
for the factor containing T) that the only
solution to DET(M)=0 is S=T mod pi. We can avoid S=T+pi because
it corresponds to changing the sign of u,v,w, hence S=T.
M 'S=T' SUBST (3s)
RREF (28s)
TRIGCOS (32s)
This gives the reduced matrix, the last column is [u,v,w,0], for example
{1,4} GET 'U' STO
will save u
Then the x coordinate of the point is
U 1 + 'COS(T)' * TLIN (3s) -> (B-1)/2*COS(3*T)
the y coordinate
U B + 'SIN(T)' * TLIN (3s) -> (B-1)/2*SIN(3*T)
Hence (x,y) describes the circle of radius (1-B)/2, 3 times faster
than the points moves on the ellipsis (which is expected since T->T+2*PI/3
gives the same point). (After redilatation the radius of the circle
is (a-b)/2).
Marchel
news:3D79B391...@nospam.fourier.ujf-grenoble.fr...
> >
> Thanks for the detailled solution which needs much less computation
> than my naive method. Which shows (again) than maths will ease
> the solution. However I see no reason why you claim the 49 can
Probably it, can once you know the exact method. Keep in mind, that I
arrived to this method of solving the problem after playing with time
consuming different methods that led to nowhere. HP49 was simply
too slow to find out that this particular method leads to the solution.
HP49 is too slow and is missing simplification to do much of the search
for the right solution type of tasks. The good proof is you :-) One almost
need to be a math professor to develop proper solution for the HP49 and
that with the help of an engineer that found out the method on TI :-)
But seriously, I asked Timite to solve the problem. I wonder, how long it
would took him to find the right solution using capabilities and speed
of HP49. Now of course, we will never know. He had magnificent help
in you, who actually programmed HP49 and knew all it's strengths
and weaknesses, and on top of it, knew the method that leads to the
simple solution.
> Could you answer to my question about what you call exotic functions
> on the 49? Thanks
All 'MOD' comands (like ADDTMOD).
All 2S, 2C etc. commands (like ACOS2S)
Commands like CHINREM, DIVPC, EGCD, EPSX0.
Commands creating polynominals of special type or special numbers (like HERMITE or IBERNOULLI).
Commands that implement particular algoritm that duplicate regular math (like HORNER, HALFTAN).
Rarely used (in general) functions (like PSI, PTAYL)
Many matrix related commands that are used for matrix decomposition in a specific way etc, beyond
simply solving linear equations. Even eigenevalues are practically used in a narrow technical field
of vibrations resonance etc and most people do not have use for it at all.
Don't get me wrong. OIt is OK, if those functions are available. The fact is, that most of them are
used in a field of interest of a narrow group of experts in that field and are not needed and usually
not even known to the average user. If I would have a choice of those functions versus faster
machine with better display I would choose faster machine every time. That is why I don't consider
as a "big advantage" that HP49 has "more functions" than TI89/92. Counting raw number of funtions
does not yet translate to huge "quality" difference.
Jack
Marchel
This just proves the point that TI is faster :-)
tExpand(m) STO> n (2.5 s)
Det(n) (20 s)
Factor is not needed, the equation is already simplified to the form of:
3 * (b + 1) * (sqrt(3) * cos(s) + 3 * sin(s)) * (cos(s) * sin(t) - sin(s) * cos(t))
-------------------------------------------------------------------------------------
4 * (cos(s) + sqrt(3) * sin(s))
which is easy to see that is zero for any value of t only when s = t
to always zero the component: (cos(s) * sin(t) - sin(s) * cos(t))
t STO> s (0s)
rref(n) STO> p (15 s)
p[1,4] STO> u
x coordiante:
tCollect((u+1)*cos(t)) (1 s)
y coordinate
tCollect((u+b)*sin(t)) (1 s)
And the outcome is: x = (b-1)*cos(3*t)/2 and y = (b-1)*sin(3*t)/2
Overall TI did the task in 39.5 seconds where HP49 for the same (similar due
to the fact that autosimplification of TI eliminates some of them) sequence of
events needed 119.5 seconds. That makes TI89/92 3 times faster than HP49.
Jack
Bernard Parisse
>
Something I said in my previous post, the reason is that
there is no advanced algorithm involved (e.g. no
polynomial gcd, etc.), therefore the CPU speed is important.
> tExpand(m) STO> n (2.5 s)
> Det(n) (20 s)
>
> Factor is not needed, the equation is already simplified to the form of:
>
> 3 * (b + 1) * (sqrt(3) * cos(s) + 3 * sin(s)) * (cos(s) * sin(t) - sin(s) * cos(t))
> -------------------------------------------------------------------------------------
> 4 * (cos(s) + sqrt(3) * sin(s))
>
Rather strange answer for an autosimplifying calc, there is an
obvious factor which is not cancelled between numerator and
denominator. Moreover, a determinant has no denominator, except
if the input has, therefore we knew that only the 4 should be there...
> Overall TI did the task in 39.5 seconds where HP49 for the same (similar due
> to the fact that autosimplification of TI eliminates some of them) sequence of
> events needed 119.5 seconds. That makes TI89/92 3 times faster than HP49.
Yes, but for this problem. Of course xcas does it much faster on my ipaq.
Bernard Parisse
> arrived to this method of solving the problem after playing with time
> consuming different methods that led to nowhere. HP49 was simply
> too slow to find out that this particular method leads to the solution.
> HP49 is too slow and is missing simplification to do much of the search
> for the right solution type of tasks. The good proof is you :-) One almost
> need to be a math professor to develop proper solution for the HP49 and
> that with the help of an engineer that found out the method on TI :-)
>
My first try was a naive method because you said you solved with your TI
whose autosimplification made all the work. I'm further convinced you
could not solve the problem on a TI without the right mathematical
approach. Once you have the right math method, using the kind of linear
algebra instructions I used is standard (1st year university level).
It's automation of linear system solving.
> But seriously, I asked Timite to solve the problem. I wonder, how long it
> would took him to find the right solution using capabilities and speed
> of HP49. Now of course, we will never know. He had magnificent help
> in you, who actually programmed HP49 and knew all it's strengths
> and weaknesses, and on top of it, knew the method that leads to the
> simple solution.
>
There are a lot of mathematical problems which can not be solved
easily using naive method, and can be solved with injection of maths
ideas. That's why people will always have to learn maths despite
the fact that computer power increase. I'm sure Timite would have
solve it using the same kind of linalg instructions once you showed
your method.
The method I used on the 49 does not require any knowledge
of the way the 49 is programmed. Taking the determinant of a system
is standard, factoring an equation to find where it vanishes is standard
etc.
>
>>Could you answer to my question about what you call exotic functions
>>on the 49? Thanks
>
>
> All 'MOD' comands (like ADDTMOD).
> All 2S, 2C etc. commands (like ACOS2S)
> Commands like CHINREM, DIVPC, EGCD, EPSX0.
Well, then you are using some of them everyday without knowing it.
RSA cryptography (which is used in almost every cryptographic program
nowadays) use modular arithmetic (POWMOD). IEGCD is the instructions
you must use to find the secrete key from the public key (or conversely)
when you generate the keys from n=p*q. To find p and q, you need
ISPRIME. CHINREM may be used to compute gcd of polynomials using
modular methods.
> Commands creating polynominals of special type or special numbers (like HERMITE or IBERNOULLI).
> Commands that implement particular algoritm that duplicate regular math (like HORNER, HALFTAN).
> Rarely used (in general) functions (like PSI, PTAYL)
> Many matrix related commands that are used for matrix decomposition in a specific way etc, beyond
> simply solving linear equations. Even eigenevalues are practically used in a narrow technical field
> of vibrations resonance etc and most people do not have use for it at all.
>
I think you just show that these commands are exotic commands with
respect to you. My students are very happy to have all these
commands at hand. I'm sure every student in a university is happy
to have eigenvalues/eigenvectors or quadratic form reduction etc.
Real world is quantum mechanics and is therefore governed by
eigenvalues/eigenvectors
However I don't think these instructions are really useful for real
world computations but not because they are exotic, but because
the 49 CPU is too slow. Something I can not change unfortunately.
Marchel
> My first try was a naive method because you said you solved with your TI
> whose autosimplification made all the work. I'm further convinced you
> could not solve the problem on a TI without the right mathematical
> approach. Once you have the right math method, using the kind of linear
> algebra instructions I used is standard (1st year university level).
> It's automation of linear system solving.
That is an obvious truth. Rarely one can solve anything without the right
mathematical approach.
> I'm sure Timite would have solve it using the same kind of linalg
> instructions once you showed your method.
I'm not so sure. Too late to proove it.
> The method I used on the 49 does not require any knowledge
> of the way the 49 is programmed. Taking the determinant of a system
> is standard, factoring an equation to find where it vanishes is standard
> etc.
> > All 'MOD' comands (like ADDTMOD).
> > All 2S, 2C etc. commands (like ACOS2S)
> > Commands like CHINREM, DIVPC, EGCD, EPSX0.
>
> Well, then you are using some of them everyday without knowing it.
> RSA cryptography (which is used in almost every cryptographic program
> nowadays) use modular arithmetic (POWMOD). IEGCD is the instructions
> you must use to find the secrete key from the public key (or conversely)
> when you generate the keys from n=p*q. To find p and q, you need
> ISPRIME. CHINREM may be used to compute gcd of polynomials using
> modular methods.
I don't actually use cryptography. Also within the cryptography, one based on the
large primes is a narrow field of the cryptography, and a one that is only secure,
because we assume, that there is no algorithm out there that can find large
primes quickly. Did the Germans also assumed, that Enigma is mathematically
close to impossible to break ?
> I think you just show that these commands are exotic commands with
> respect to you. My students are very happy to have all these
> commands at hand. I'm sure every student in a university is happy
> to have eigenvalues/eigenvectors or quadratic form reduction etc.
> Real world is quantum mechanics and is therefore governed by
> eigenvalues/eigenvectors
> However I don't think these instructions are really useful for real
> world computations but not because they are exotic, but because
> the 49 CPU is too slow. Something I can not change unfortunately.
With the respect to the most of the TI or HP users. It is you (and your
students) who are happen to be a minority. If HP or TI would target
strictly math students they would have a problem selling probably one tenth
of what they are selling today. By the way, how your math students can
live without geometry application for HP49. Geometry is a mother of math
and only TI92 provides that :-) How many physicists are specializing in
quantum mechanics ?
Jack
parisse
I would be very astonished that you never use cryptography.
Maybe you just don't realize. On-line sales and bank
transactions are all crypted and they use public key systems,
most of them use RSA. Even if you don't use these type of
services yourself, you can not deny that cryptography based
on arithmetic is widely used in the world.
> Also within the cryptography, one based on the
> large primes is a narrow field of the cryptography, and a one that is only secure,
> because we assume, that there is no algorithm out there that can find large
> primes quickly. Did the Germans also assumed, that Enigma is mathematically
> close to impossible to break ?
>
You seem a little bit confused, the problem is not to find large
prime quickly, in fact there are fast algorithm to find (pseudo)-primes.
The problem is to factor integer with large prime factors.
A cryptographic system is never safe. It may be safe
for a period of time. Current algorithms on the market are very
far to be able to crack 1024 bit n=p*q for example. It might be
cracked in the future but who cares if it's 10 years in the future?
> With the respect to the most of the TI or HP users. It is you (and your
> students) who are happen to be a minority.
I think you are judging too much on the standards you know.
In my country, there are about 100 000 students per year
*in the highschool* who learn about extended GCD algorithm.
> If HP or TI would target
> strictly math students they would have a problem selling probably one tenth
> of what they are selling today. By the way, how your math students can
> live without geometry application for HP49.
There is 0h of geometry in the University (and only a few hours in
the last year of highschool). On the other hand, there is about
100h/year of linear algebra or more (for about 100 000
students/year)
> Geometry is a mother of math
> and only TI92 provides that :-)
Not completely true, I can do geometry on my ipaq. Anyway
this kind of geometry is important in schools under say 15
or 16, not after.
> How many physicists are specializing in
> quantum mechanics ?
>
This is not the point. The point is that like cryptography
some aspects of quantum physics affect our world. If it was
not teached to you, it does not mean that it is not important.
And eigenvalues/eigenvectors are at the heart of quantum
mechanics. And there are several other areas where egv are important
even outside physical application: e.g. statistics or biological
modelization of some phenomena.
Marchel
>
> I would be very astonished that you never use cryptography.
> Maybe you just don't realize. On-line sales and bank
> transactions are all crypted and they use public key systems,
> most of them use RSA. Even if you don't use these type of
> services yourself, you can not deny that cryptography based
> on arithmetic is widely used in the world.
You are right that I use it, but I don't need to know the theory.
As I mentioned, only small circle of specialists need to know
the details. The fact, that you use electronic, electrical or
mechanical devices does not mean that you know all details of
theory that drives them and also did not prevented you (or someone
other involved in HP49 design) from removing engineering library from
HP49 :-) Please, explain why ?
> You seem a little bit confused, the problem is not to find large
> prime quickly, in fact there are fast algorithm to find (pseudo)-primes.
> The problem is to factor integer with large prime factors.
I'm not confudsed. That is exactly what I meant.
> A cryptographic system is never safe.
This is false statement. Theoretically cryptographic system
with the secret key that is as long as the message itself and is never
reused is unbreakable. Also recently introduced quantum systems are
completely safe and on top of it they also are capable to discover
evasedropping.
> It may be safe
> for a period of time. Current algorithms on the market are very
> far to be able to crack 1024 bit n=p*q for example. It might be
> cracked in the future but who cares if it's 10 years in the future?
That is a common mistake of overconfidence, the Germans did. They
assumed, that with known to them algorithms it was impossible to
decypher Enigma messages within reasonable time. What they did
not consider, was, that Polish mathematicians created algorithm
that was not considered by them. Later French intelligence and after
them Alan Turing of British intelligence improved algorithm and
automated task, to the point, that they were able to decypher
German messages almost the same day. The proper sentence
should be "publicly known algorithms". That is the whole world
of difference.
> I think you are judging too much on the standards you know.
> In my country, there are about 100 000 students per year
> *in the highschool* who learn about extended GCD algorithm.
And used it later for ?
> > If HP or TI would target
> > strictly math students they would have a problem selling probably one tenth
> > of what they are selling today. By the way, how your math students can
> > live without geometry application for HP49.
>
> There is 0h of geometry in the University (and only a few hours in
> the last year of highschool). On the other hand, there is about
> 100h/year of linear algebra or more (for about 100 000
> students/year)
In your country maybe there is 0h geometry.
> > Geometry is a mother of math
> > and only TI92 provides that :-)
>
> Not completely true, I can do geometry on my ipaq. Anyway
> this kind of geometry is important in schools under say 15
> or 16, not after.
So who is teaching engineering drafting in French engineering schools ?
Do you use immigrants or rely on american CAD to do all the
projections and intersections for them ?
> > How many physicists are specializing in
> > quantum mechanics ?
> >
>
> This is not the point. The point is that like cryptography
> some aspects of quantum physics affect our world. If it was
> not teached to you, it does not mean that it is not important.
> And eigenvalues/eigenvectors are at the heart of quantum
> mechanics. And there are several other areas where egv are important
> even outside physical application: e.g. statistics or biological
> modelization of some phenomena.
Sure, but you also skipped when programming HP49 things like
calculus of variations - very usefull, partial differential equations
solving - the whole world is described with those, integral equations, etc.
Why have you choosen cryptography over partial differential equations ?
Maybe in your country they don't teach partial differential equations ?
Jack
parisse
> As I mentioned, only small circle of specialists need to know
> the details.
RSA cryptography is not complicated, we don't speak of elliptic
curve crypography. There are tons of people who understand the
theory.
> The fact, that you use electronic, electrical or
> mechanical devices does not mean that you know all details of
> theory that drives them and also did not prevented you (or someone
> other involved in HP49 design) from removing engineering library from
> HP49 :-) Please, explain why ?
>
I'm not responsible for the choice of HP not to keep the EQLIB
on the 49.
And I disagree, you don't know all details of electronic, etc. but
you know the principles like F=m*gamma or U=RI, etc. I believe the
same should apply to e.g. understanding how public key cryptography
work.
>
>>You seem a little bit confused, the problem is not to find large
>>prime quickly, in fact there are fast algorithm to find (pseudo)-primes.
>>The problem is to factor integer with large prime factors.
>
>
>
> I'm not confudsed. That is exactly what I meant.
>
Then reread your post, it's not what you said.
>
>>A cryptographic system is never safe.
>
>
> This is false statement. Theoretically cryptographic system
> with the secret key that is as long as the message itself and is never
> reused is unbreakable. Also recently introduced quantum systems are
> completely safe and on top of it they also are capable to discover
> evasedropping.
>
Secret key systems are not safe because you must exchange keys.
Quantum systems might be, but that shows another interest of
studying quantum mechanics!
> That is a common mistake of overconfidence, the Germans did. They
> assumed, that with known to them algorithms it was impossible to
> decypher Enigma messages within reasonable time. What they did
> not consider, was, that Polish mathematicians created algorithm
> that was not considered by them. Later French intelligence and after
> them Alan Turing of British intelligence improved algorithm and
> automated task, to the point, that they were able to decypher
> German messages almost the same day. The proper sentence
> should be "publicly known algorithms". That is the whole world
> of difference.
>
I don't care if some secret service can decrypt my transactions with
my bank. And RSA is a public algorithm. If someone uses an unknown
method to factor fast integers to decrypt some messages, it will
be very difficult for him to keep it secret..
>
>
>>I think you are judging too much on the standards you know.
>>In my country, there are about 100 000 students per year
>>*in the highschool* who learn about extended GCD algorithm.
>
>
> And used it later for ?
>
E.g. understand RSA. But extended gcd algorithm is a key
algorithm in arithmetic. Like RSA, I bet you are using it.
EGCD is the basic tool for partial fraction decomposition for
example hence is called every time you integrate a fraction.
>
> In your country maybe there is 0h geometry.
>
I don't know other country curriculum, but I would be
astonished if the kind of geometry package you find
on the TI92 is usefull at University levels.
>
> So who is teaching engineering drafting in French engineering schools ?
> Do you use immigrants or rely on american CAD to do all the
> projections and intersections for them ?
>
I really don't know, it's a too specialized area. It's most certainly
using software done by Indian (immigrants or not) for US corporations
or homemade soft. Believe it or not, we have a lot of well trained
engineers in France. This might change in the futur as fewer students
enter scientific studies.
>
> Sure, but you also skipped when programming HP49 things like
> calculus of variations - very usefull, partial differential equations
> solving - the whole world is described with those, integral equations, etc.
> Why have you choosen cryptography over partial differential equations ?
Because arithmetic has a sense on a calc as partial diff equation
require much more power.
> Maybe in your country they don't teach partial differential equations ?
>
Sure we teach PDE, but not at the same level as arithmetic.
OK, I think we are now too much out of topic, it would be a good
idea to stop this thread, that's what I'll do myself.
Marchel
> > You are right that I use it, but I don't need to know the theory.
> > As I mentioned, only small circle of specialists need to know
> > the details.
>
> RSA cryptography is not complicated, we don't speak of elliptic
> curve crypography. There are tons of people who understand the
> theory.
That doesn't change the need to have specific functions dedicated to it
on the calculator. They are not needed everyday and not even that difficult
to program when needed.
> I'm not responsible for the choice of HP not to keep the EQLIB
> on the 49.
> And I disagree, you don't know all details of electronic, etc. but
> you know the principles like F=m*gamma or U=RI, etc. I believe the
> same should apply to e.g. understanding how public key cryptography
> work.
The formulas you gave do not require engineering knowledge. They
are thought in my country in a 7th grade.
> > I'm not confused. That is exactly what I meant.
> >
>
> Then reread your post, it's not what you said.
I said, "this is what I meant", not "this is what I wrote".
I made incorrect statement, but I meant what you wrote.
> > This is false statement. Theoretically cryptographic system
> > with the secret key that is as long as the message itself and is never
> > reused is unbreakable. Also recently introduced quantum systems are
> > completely safe and on top of it they also are capable to discover
> > eavesdropping.
> >
>
> Secret key systems are not safe because you must exchange keys.
> Quantum systems might be, but that shows another interest of
> studying quantum mechanics!
Mathematically secret random key as long as message and
not reused is unbreakable. I said nothing about stealing key because
in any key cryptography keeping the key or part of the key secret is
a must. Protection of the key therefore is beyond the mathematical
problem of unbreakable code. Quantum system is unbreakable only
within known laws of physics which are not as strong as mathematical
proof because they might change in the future with new discoveries.
> I don't care if some secret service can decrypt my transactions with
> my bank. And RSA is a public algorithm. If someone uses an unknown
> method to factor fast integers to decrypt some messages, it will
> be very difficult for him to keep it secret..
Unless it is some intelligence service especially in the countries with
tyrannical governments. Of course, I don't think secret services are going to
decipher yours or mine bank transactions and reveal that way their
secret. But for that matter one could probably use Enigma with 10 wheels
and not many codebreakers would attempt to break it either.
> E.g. understand RSA. But extended gcd algorithm is a key
> algorithm in arithmetic. Like RSA, I bet you are using it.
> EGCD is the basic tool for partial fraction decomposition for
> example hence is called every time you integrate a fraction.
Internally yes. But I rarely use this function directly. That is why
consider this function as exotic. I did not said, that the function is
not needed internally. What I said is, that exposing this function to
the user does not add much functionality to the everyday user. I wonder,
if we could make a pool on this site of dedicated HP users and ask them
to list the functions they have never used directly (except when learning
the calc possibilities) and don't think that they will ever use them
in the future. I bet, that most of the MOD type functions will top the list.
That is my guess. Also many of the matrix decompositions will follow.
Again. I do understand the theory of each decomposition. Do I use
them ? No. And because it is so many of them, some day I probably
will forgot the exact syntax of the command anyway and will not use
it when needed immediately or even forgot, that the specific function
exist at all.
> I don't know other country curriculum, but I would be
> astonished if the kind of geometry package you find
> on the TI92 is usefull at University levels.
Probably not. I guess, intention was to use it on the high school level.
Most of the universities - especially technical, teach some of the drafting
projections.
> I really don't know, it's a too specialized area. It's most certainly
> using software done by Indian (immigrants or not) for US corporations
> or homemade soft. Believe it or not, we have a lot of well trained
> engineers in France. This might change in the future as fewer students
> enter scientific studies.
I do believe and I bet they are taught drafting geometry. Otherwise
they would have a hard time to read engineering drawings.
Most of the industrial strength CAD systems are still made in North
America and one in France. Of course in US there are many
programmers who are immigrants and not necessarily of
a European origin. Europeans in general do not have monopoly
for brain. Remeber Ramanujan ?
Jack