Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

pi to 24 significant digits; TI sin(x) bug HP doesn't have

18 views
Skip to first unread message

Rick Nungester

unread,
Oct 27, 2003, 8:58:46 PM10/27/03
to
Pi can be easily found to 24 decimal places on HP calculators, by making
use of sin(x) ~= x for small x (in radians). I am using an HP-32SII,
which provides 12 significant digits:

Enter pi. See 3.14159265359.
3.14159265358 has to be slightly less than pi.

sin(3.14159265358) = 9.79323846264e-12
pi ~= 3.14159265358 9 79323846264 (24 digits)

Trying the same technique on a TI-83+SE (10 significant digits, so I
would hope of easily getting 20) shows a problem with sin(x):

Enter pi. See 3.141592654. So far so good.
3.141592653 has to be slightly less than pi.

sin(3.141592653) = 5.898e-10. PROBLEM. The answer has dropped to 4
significant digits instead of 10, so the TI misses 6 important
significant digits for arguments close to pi. The HP calculator
correctly gives:

sin(3.141592653) ~= 5.89793238463e-10, or
pi ~= 3.141592653 5 89793238463

I understand why the TI does this, and it isn't optimum.

Rick

-----= Posted via Newsfeeds.Com, Uncensored Usenet News =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
-----== Over 100,000 Newsgroups - 19 Different Servers! =-----

XYZ

unread,
Oct 28, 2003, 3:23:41 AM10/28/03
to
> I understand why the TI does this, and it isn't optimum.


...And that's why we don't use TI's! (even though they just may be
constructed better now days, HP calc software rocks :-)))

Bhuvanesh

unread,
Oct 28, 2003, 8:59:24 AM10/28/03
to

TI calc software rocks too! :-)

Just chirping in for no reason,
Bhuvanesh.

Edward Sanford Sutton, III

unread,
Oct 29, 2003, 1:58:02 AM10/29/03
to
"Rick Nungester" <nung...@mail.com> wrote in message news:<3f9dce5e$1...@corp.newsgroups.com>...

Thanks for the cool idea. my ti83 was stolen, but my ti86 and ti89
can seem to check the results, but they like to round too much, which
makes the truth hard to see. =(
Something seems flakey somewhere in this to me though. if you need
sin(x)~=x to have a value very close to x, (taking note that
3.14159265358 is less than pi), is there a way to use 3.14159265359 to
find the extra pi digits since it is closer to pi (and also less than
pi), while not being equal to pi?
By the way, HP49G returned the same results as your HP-32SII.
Ed Sutton

Tom Lake

unread,
Oct 29, 2003, 5:07:18 AM10/29/03
to
> Thanks for the cool idea. my ti83 was stolen, but my ti86 and ti89
> can seem to check the results, but they like to round too much, which
> makes the truth hard to see. =(
> Something seems flakey somewhere in this to me though. if you need
> sin(x)~=x to have a value very close to x, (taking note that
> 3.14159265358 is less than pi), is there a way to use 3.14159265359 to
> find the extra pi digits since it is closer to pi (and also less than
> pi), while not being equal to pi?

The Pi key (2nd ^) gives the value of Pi to as many places as the calculator
can use anyway so what's point in computing it?

--
Tom Lake

Experience keeps a dear school but fools will learn in no other - Poor
Richard's Almanack


Arnaud Amiel

unread,
Oct 29, 2003, 6:31:13 AM10/29/03
to
mirr...@hotmail.com (Edward Sanford Sutton, III) wrote in message news:<76bad108.03102...@posting.google.com>...

>
> Thanks for the cool idea. my ti83 was stolen, but my ti86 and ti89
> can seem to check the results, but they like to round too much, which
> makes the truth hard to see. =(
> Something seems flakey somewhere in this to me though. if you need
> sin(x)~=x to have a value very close to x, (taking note that
> 3.14159265358 is less than pi), is there a way to use 3.14159265359 to
> find the extra pi digits since it is closer to pi (and also less than
> pi), while not being equal to pi?
> By the way, HP49G returned the same results as your HP-32SII.
> Ed Sutton


Of course, it works but 3.14159265359 is very stlightly more than pi so you get

sin(3.14159265359)=-2.0676154E-13 Which is by how much it is slightly bigger.

1+(-.2067154)=.7932846

Which are the numbers you are after.

Arnaud

Edward Sanford Sutton, III

unread,
Oct 29, 2003, 11:59:03 PM10/29/03
to
aam...@hotmail.com (Arnaud Amiel) wrote in message news:<31032eba.03102...@posting.google.com>...

Hmm, seems I managed to screw up my decimal place when checking my
last digits quite a few times yesterday then while looking at it. That
did answer my next question of if it was still usable though too so
thank you very much. Now I wonder what other fun things there are to
do with my calc. =)
Ed Sutton

0 new messages