HP48GX vs. TI 86
rjr...@rocketmail.com
after I just started with the company, fresh out of BS structural engineering
program
the company will reimburse me and I am in the market to buy a new calc
not sure what to get either HP48GX (a direct replacement) or the TI86
I remember during an analysis class I had to borrow a TI from a friend to
do a really big matrix's, I think the HP had a limit to the matrix's size
Do you know what it is?
Which calc would you get the TI86 or the HP48GX
I'm sure your probably bias, but please give me you opinion, What is the main
difference between them?
please email me with answer "rjr...@rocketmail.com"
Thank you
ze...@magicnet.net
(rjr...@rocketmail.com) wrote:
Why are you considering the TI-86? The only one you should even think of
besides the HP-48GX is the upcoming TI-89. (Or of course the TI-92 with
the + module.) It can handle very large matrixs..and it can give you exact
symbolic results of the matrix operations, or you can choose floating
point of course. The TI-89 has 188K of user RAM, so you would have plenty
of memory. The HP-48GX should be able to handle very large matrixs also.
--
If only we knew that money is only an idea. There is no scarcity or loss connected to it. Nothing cost anything.
Ken Alverson
>do a really big matrix's, I think the HP had a limit to the matrix's size
>
>Do you know what it is?
The GX's array handling is limited only by its memory. The TI has a limit,
but I don't know it off hand (check their web site). You can certainly do a
lot more a lot quicker with the HP, and you are already experienced with
RPN, I think you will only feel limited by the TI-86 if you get it.
Ken
ze...@magicnet.net
<Ke...@youspamyoudie.tso.cin.ix.net> wrote:
> The GX's array handling is limited only by its memory. The TI has a limit,
> but I don't know it off hand (check their web site). You can certainly do a
> lot more a lot quicker with the HP, and you are already experienced with
> RPN, I think you will only feel limited by the TI-86 if you get it.
>
> Ken
The built in Simultaneous Equation solver in the TI-86 is limited to 30
equations with 30 unknowns. The same is true of the TI-85. the TI-86 has a
lot more memory than the TI-85, but still only does 30 equations. Go
figure...The HP-48GX, TI-89 and TI-92+ can do a lot more than 30
equations.
JEEjohn
>not sure what to get either HP48GX (a direct replacement) or the TI86
>
>I remember during an analysis class I had to borrow a TI from a friend to
>do a really big matrix's, I think the HP had a limit to the matrix's size
>
>Do you know what it is?
I've had 37x37 matrices in the TI-85 , while doing some matrix inversion
timings (note that it had to have memory to hold two matrices and then
some more). The TI-86 has 128K of memory, but I don't remember how it
addresses the upper 96K. If it can page the memory, then you may
be able to use a larger matrix. But I think it's like "archive" memory
that's not directly available.
I've had 90x90 matrices in the HP48GX w/ a 128K RAM card.
You should be able to use something about 75x75 in a HP48GX w/
no RAM card.
Another thing you might want to know: The TI-85 took about twice
as much time to invert the same size matrix that the HP48 did.
And the TI-92 (and hence the TI-92 Plus and TI-89) was **extremely**
slow! Inverting a 30x30 matrix took the HP48G 81 secs, the TI-85
174 secs, and the TI-92 1297 secs!.
So, I think you're still better off with the HP48GX.
John Edry
JEE...@aol.com
ze...@magicnet.net
jee...@aol.com (JEEjohn) wrote:
> Another thing you might want to know: The TI-85 took about twice
> as much time to invert the same size matrix that the HP48 did.
> And the TI-92 (and hence the TI-92 Plus and TI-89) was **extremely**
> slow! Inverting a 30x30 matrix took the HP48G 81 secs, the TI-85
> 174 secs, and the TI-92 1297 secs!.
>
> So, I think you're still better off with the HP48GX.
More than likely, you inverted the matrix on the TI-92 by just hiting the
enter key. That makes it give the answer in exact "symbolic" form. If you
first hit the green diamond before the enter key, you get the resulkt in
decimal form, and that is a lot faster than an exact symbolic form result.
JEEjohn
writes:
>In article <199808090048...@ladder03.news.aol.com>,
>jee...@aol.com (JEEjohn) wrote:
>
>> Another thing you might want to know: The TI-85 took about twice
>> as much time to invert the same size matrix that the HP48 did.
>> And the TI-92 (and hence the TI-92 Plus and TI-89) was **extremely**
>> slow! Inverting a 30x30 matrix took the HP48G 81 secs, the TI-85
>> 174 secs, and the TI-92 1297 secs!.
>>
>> So, I think you're still better off with the HP48GX.
>
>More than likely, you inverted the matrix on the TI-92 by just hiting the
>enter key. That makes it give the answer in exact "symbolic" form. If you
>first hit the green diamond before the enter key, you get the resulkt in
>decimal form, and that is a lot faster than an exact symbolic form result.
1) For those who don't know what the "green diamond - enter" is :
It runs the TI-92 in the approximate mode.
2) The results I posted **are** for the approximate mode!
3) I now have the TI-92 Plus module, and I just started running
some timings on it. My TI-92 Plus is slightly faster (about 7-8%)
than the original TI-92.
4) I ran symbolic mode on a 20x20 matrix inversion - 689 secs
The same matrix in approximate mode is 267 secs.
One comment is that the TI-92 Exact mode gave the result in fractions
that had 23-24 decimal integers.
5) The TI-92 Plus inverted a 30x30 matrix (in approximate mode)
in 1188 secs. Still very long! Symbolic mode will probably give
the result in something like 4,000 secs!
6) With some 200K of memory, I now should be able to get time results
for inverting 40x40 & 50x50 matrices.
End Result: the TI-92, TI-92 Plus (& hence the TI-89) should not be
used for matrix calculations, unless the matrices are very small, or
you're willing to wait a very very long time.
John Edry
JEE...@aol.com
Gregory de Lagrange Chancel
quite the same : TI 89 (expected in september) would be more appropriate.
2) Regarding matrix algebra, it is absolutely true that TIs are really slow,
even when you use the approximate answer mode. I use matrix calculations to do
econometrics on my calculator. I do it on both HP 48 GX (with a 128 K RAM card
in port 0) and TI 92 Plus (which is the same as TI 89 except for geometry)...
The TI is really slow (especially if you keep in mind that it is a fast
calculator for many other things - thank you Motorola for the 68000). On the
other hand, the matrix editor of the TI is really easier and faster to use than
HP one.
Jacek Marchel
>
> In article <199808090048...@ladder03.news.aol.com>,
> jee...@aol.com (JEEjohn) wrote:
>
> > Another thing you might want to know: The TI-85 took about twice
> > as much time to invert the same size matrix that the HP48 did.
> > And the TI-92 (and hence the TI-92 Plus and TI-89) was **extremely**
> > slow! Inverting a 30x30 matrix took the HP48G 81 secs, the TI-85
> > 174 secs, and the TI-92 1297 secs!.
> >
> > So, I think you're still better off with the HP48GX.
>
> More than likely, you inverted the matrix on the TI-92 by just hiting the
> enter key. That makes it give the answer in exact "symbolic" form. If you
> first hit the green diamond before the enter key, you get the resulkt in
> decimal form, and that is a lot faster than an exact symbolic form result.
>
That is not exactly true. TI92 even with diamond key is 5 times slower
in inverting the matrix (and TI92+ about 3 times) than HP48. That is
not necessarily the TI disadvantage. The reason for such a poor
performance is that TI allows symbolic objects as an matrix entries and
performs exact inversion of the matrix representing inverted matrix
in fractional or algebraic notation in exact mode and in decimal
notation in approx. mode where the simplification - loss of accuracy is
performed on the already inverted matrix. There are of course advantages
and disadvantages of such behavior. The advantages are clear. If the
matrix is close to but no singular, TI92 will give the exact proper
answer, where HP might fail or when the matrix is exactly singular,
TI will find that, where HP might fail again. Additionally,
when the matrix is involved in theoretical formulas, supplying
algebraic - symbolic matrix allow the user to develop final symbolic
formula, where HP without special software is completely unable to
achieve that. The disadvantages are also clear. TI92 timing is
practically unacceptable for engineering applications, where matrix
is part of the calculations, especially when the program is searching
for the optimum value using iterations in solver.
Such a program will outlast even the most patient user on the TI92.
It is a surprise, that the TI92 do not have approx. mode for the
intermediate calculations assuming , the entire matrix is numerical.
> --
> If only we knew that money is only an idea. There is no scarcity or loss connected to it. Nothing cost anything.
With regards.
ze...@magicnet.net
jee...@aol.com (JEEjohn) wrote:
> 1) For those who don't know what the "green diamond - enter" is :
> It runs the TI-92 in the approximate mode.
> 2) The results I posted **are** for the approximate mode!
> 3) I now have the TI-92 Plus module, and I just started running
> some timings on it. My TI-92 Plus is slightly faster (about 7-8%)
> than the original TI-92.
> 4) I ran symbolic mode on a 20x20 matrix inversion - 689 secs
> The same matrix in approximate mode is 267 secs.
> One comment is that the TI-92 Exact mode gave the result in fractions
> that had 23-24 decimal integers.
> 5) The TI-92 Plus inverted a 30x30 matrix (in approximate mode)
> in 1188 secs. Still very long! Symbolic mode will probably give
> the result in something like 4,000 secs!
> 6) With some 200K of memory, I now should be able to get time results
> for inverting 40x40 & 50x50 matrices.
>
> End Result: the TI-92, TI-92 Plus (& hence the TI-89) should not be
> used for matrix calculations, unless the matrices are very small, or
> you're willing to wait a very very long time.
I ran some tests also and I got the same results as you. The TI-92+ is
slow with matrix math. My HP-48 solves a system of 29 simultaneous
equations with 29 unknowns in around 30 seconds. My TI92+ took just over 9
minutes. By the way, why did I pick 29 equations? What does 29 equations
have to do with the history of computers? A litlle trivia question for you
folks.