Looks interesting, too bad i already have an 89, but i might as well try it out. I'm glad people are developing apps. on to graphlink then i need something constructive...
this is MY first comment (i think) :)

Hey, you got first comment!

Boy, am I so depressed. I spent a whole hour trying to make an assembly program. (I'm learning assembly). It's for the 82. I was trying to display a sprite, and then move it with the arrow keys, but it won't move. :( Sigh.

#include "TI82.H"
#include "KEYS.INC"

.ORG START_ADDR
.DB "no_one_2000 is sad.",0

ROM_CALL(CLEARLCD)
LD HL,0
LD (CURSOR_POS),HL
LD HL,Mean
ROM_CALL(D_ZT_STR)
WaitBLAH:
CALL GET_KEY
CP G_ENTER
JR NZ,WaitBLAH
RET
Mean:
.DB "Assembly is mean.",0

.END

hey there are plenty of sprite tutorials on the web! Anyhow you really don't have a sprite anywhere in that code. A sprite is generally represented by binary numbers. What you have is soem text. A sprite could look something like this:

Sprite:
.db %11111111
.db %10000001
.db %10000001
.db %10000001
.db %10000001
.db %10000001
.db %10000001
.db %11111111


that is a box...the ones correspond to a pixel being on or black.

Hmm... that looks hard. I've been programming a lot in Basic, but i've been wanting to move on now for awhile. So, is assembly that much better than C? I knew C a couple years ago but forgot... C sure is easier than assembly though... is the speed gains from assembly worth it? Thanks.. I know it's off topic but i can't find forums anywhere here.

Now on topic: I think it's great also, that finally the TI-83+ math functions are improving. I'd have downloaded it all in a second if I didn't *just* get and 89... sigh. Maybe I still will.

Assembly can seem intimidating at first but it really is not that bad. In many regards it is better from a programmers point of view because he/she is allowed to tell the processor exactly what to do without a wrapper of any sort. I definately think you should take the time and find some tutorials....there are plenty on the web you just need to look.

Dah, that code was obviously meant to be a joke... ;-)

Yeah, I'm not really that good with Assembly. :( I know Javascript, Java (somewhat), and a few other things, but this is so different. I guess having only about 8 registers is a bit intimidating.

Oh, that's not my program. If you want to see my program, I'll e-mail it to you. It doesn't lock up or anything, it just won't move. You use CLEAR to exit.

There's really a lot of tutorials on the web? The only one I could find was karma.ticalc.org (which is pretty good).

TI-89 assembly:
http://basil.cs.uwp.edu/ Documentation/bsvc/68kasm/
http://www.technoplaza.net/ assembly/index.cgi?p=68kmain
http://www.technoplaza.net/assembly/

I don't know about any others... didn't look.

I don't have an 89. How about an 82? Anyone know any good sites for the 82?

Whoa! Last week that was true...

But I got one for Christmas! Oh yeah, oh yeah, oh, oh, oh yeah.

w00t! I'll give it a try. I downloaded the latest TiLP, does FLASH for the 83+ work yet? I guess I'll find out.

Nope...

w00t

Great job!

Nice that someone cares about the calculator's main purpose.

Simplify and differentiate will probably be most useful.

Somewhere on Internet is available source code of MuMath.

This means that some future versions of Symbolic can have also integration, and diferential equations.

Try www.MuPad.de and go to the developers' page. Last I heard, the source to MuPad (the best free CAS I have been able to find) was a free download, but once you have symbolic differentiation working, it should be a fairly trivial thing to add symbolic integration. All you really need to do, if he did it the way I think he did, is replace the differentiation rules with integration rules.

I do not think integration would be that easy. If normal problems encountered in a second level calculus class require much algebraic manipulation to be able to integrate something. I used recursion and binary trees for differentiation and simplification. I will take a look at the source but I can't seem to find the source on the web site itself. Do you have a link? Maybe I should email the developer address.

I'm actually developing symbolic differentiation for the 83 (I'm just very slow...). However, I chose a simple, non-recursive method, because I'm not sure whether or not the stack would be enough. Currently I'm sort of sucking with the parser...

there seem to be a few ways of doing this but they all rely on stacks, you could simulate a stack(maybe just add on the op or fp stacks). What i did was i relocated the stack into free ram. this let the stack grow much larger than the 400 allocated for hardware stack. First i opened up an edibuffer then set the stack to the last byte of the edit buffer. then my binary tree up and my stack grew down in memory. when these two got danerously close the memory error was thrown.

webMathematica's Integrator http://integrals.wolfram.com/
takes about 500 pages of Mathematica code and 600 pages of C code, so it won't be that easy...

That's why I'm doing differentiation, not integration. :)

I am sorry, I did not fully understand how you implemented the differentiation. I thought it was just a table driven kind of thing where you take the expression, find the outermost operator, and apply the rule regarding it. I think that this is how HP got that nifty step-by-step calculus stuff. It is a _lot_ easier to make this kind in RPN, though.

For the most part, integrals I encounter are about the same complexity of differentiating, just in reverse and a little wierder thanks to the complexities of that little "C" in them. In my room, right now, I have 2 little plastic sheets with the diff. and integ. rules on them, and these have worked for about 98% of the integrals I have needed to solve, and all of the derivatives.

For a starter, try to integrate sin(x^2) and x^x. And let's see what you come up with. :)

2*x*cos(x^2)
and
(ln(x)+1)*x^x

Wait, that'd be the differentiation.
Those are hard to do.

x^3 x^5 x^7
sin(x) = x - ----- + ----- - ----- + ... ...
3! 5! 7!

thus:

x^6 x^10 x^14
sin(x^2) = x^2 - ----- + ------ - ------ + ... ...
3! 5! 7!

and:

x^3 x^7 x^11 x^15
Integrate( sin(x^2)dx ) = ----- - ----- + ------ - ------ + ... ...
1!3 3!7 5!11 7!15

infinity | x^(4n-1) |
= SUM | (-1)^(n-1) * --------------- |
n = 1 | (2n-1)!(4n-1) |

...and i'm only a junior...

if you don't believe me try the following on your calcs (if you use an 86):

fnInt(sin(x^2), x,0,3.1415926535898/2) ----------------------- (1)
sum seq((-1)^(x-1)* (3.1415926535898/2) ^(4x-1)/(2n-1)!/(4n-1),x,1,10) ---- (2)

They should give almost the same number :)
Technically, you should substitute 0 for (3.1415926535898/2) in (2) and subtract it from (2) as well, but it is just 0.

(Don't know why the 89s and the 92+s don't give this value - i guess its still better to think about a problem first before plugging it into the calc)

As for the x^x I don't think there is an integral

Speaking of 86's, why doesn't somebody make a symbolic like program for the 86 (although I think there is a program called cas86 in the basic archives of ti86 math programs but i don't know how it works or what it does).

Is it even *possible* to do the x^x ? Well, sure, you could define a new special function, but that's not what I mean.

isn't that integration by parts?

As my former maths teacher said: "Differentiation is for children at the kindergarten. Integration is an art."

Unfortunately there are no "integration rules", it's a rather intuitive process, where lots of individual cases should be separately programmed, and such a program simply cannot be fit on those Z80 calcs - only if we restrict it to simple cases.

Oh yes, there is an algorithm (actually, several) for symbolic integration. For example, look up Rothstein-Trager or Risch integration. There is also a method called Horowitz reduction that helps with integration. All of this should be in any basic computer algebra text.

When I mentioned MuMath, it was written for CP/M, so it was designed for Z80 computers.
It was written by the same software house as Derive and TI89 ROM.

You can download it on link above

The problem is, of course, that TI-calculators, wonderful though they are, are entirely stupid.

Integration, as you said, is not really rule-based; the problem is a general intelligence one, and it is IMPOSSIBLE to code a good general intelligence (see: http://singinst.org/GISAI, or the whole damn site for that matter, as well as its 'partner' http://sysopmind.com/) on a computer that only do < 10^15 flops. Idiots-savants are not able to integrate like you or me, because their prowess is in an incredibly small domain, like a computer's.

It might be possible to hack up an integration feature, but it would be slower than molasses in the middle of an antarctic winter on Pluto, and would screw up a lot.

>> The problem is, of course, that TI-calculators, wonderful though they are, are entirely stupid.

Ack! Revising my kludge-mode writing...

TI-Calculators are dumb.
Integration is mostly an intellectual process which cannot be done by simple rules.
Intelligence requires a computer with power near that of a human brain.

For more information on AI, visit http://singinst.org/GISAI or sysopmind.com

<<Integration is mostly an intellectual process which cannot be done by simple rules.
Intelligence requires a computer with power near that of a human brain.>>

If that is correct, then the TI-89 is one such computer.

NO, because it cannot do integration in general, only some specific cases. Try those two ones I mentioned a bit above, and you'll see. A human can integrate ANYTHING in theory, given the knowledge and enough time. However, no matter how much time you give to a computer, it won't be able to do that.

Yes, if the human introduces new functions for every integral he can not solve.

And there are many of them. The most famous are

Integral (sin(x)/x) dx
Integral (e^(-x^2)) dx


Anyway, on my postgraduate studies, my thesis for artificial inteligence course was about algebra calculators like TI89.

The symbolic integration, chess, and the most of other "artificial inteligence" miracles is actually method "try several ways and select the most appropriate", with many optimisations.

The human thinking method is "by analogy". When I remark that I have to integrate function that is rational function of sines and cosines, my experience tells me that I need to introduce replacement sin x=t,
cos x dx=dt.

"Yes, if the human introduces new functions for every integral he can not solve."

And now we have named the most important feature of humans that distinguishes them from calculators. :)

Err, that is one of the most obsurd things ever uttered by someone who doesn't know what they are talking about. To put it in terms comparing it to the computer processor, the brain functions at about 3.8 THz, that's right *tera*hertz, and can store about 32TB, *tera*bytes, of info. Of course, it isn't really in binary or anything, so this isn't that acurate, but it does give a fairly cool comparison.

Any math operation can be broken up into a process that the calc can understand. What makes it seem like an art is not telling the program about the bits you use. FOr example, having a calculator play chess may at first seem complicated, and only handled case-by-case, until you realize that the way a computer handles data needs to be changed to make it integrate (yukyuk) with chess. You need to include checking, jumping, special moves, promotion, in addition to things such as bishops move diagonally and rooks move linearly. THings aren't special-case, just underutilized :)