Request for suggestions: algebraic equations in Free42

[Thomas Okken]

I am thinking about creating a paid version of Free42, with some of the enhancements I have been asked to do over the years, but didn't do because I felt they would take too much of my time to implement, like a big screen, big stack, units, and algebraic equations.

I am looking for inspiration about how to approach the algebraic equations in particular. I'm not looking to create another RPL machine, there are plenty of those already, and that really would be far too big a project for me to consider... but I know there are other, non-RPL, HP calculators with algebraic equations that people like, and maybe one of those would be a good starting point.

What machine would you recommend for me to look at? The 27S, 19BII, 35S, something else? Anything that I could run as an emulator to play with, or would be able to buy without spending an arm and a leg, and without having to embark on a restoration project, would obviously be preferred.

(I'm not making any promises yet as to whether this will ever see the light of day. Think of it as a feasibility study for now.)

[pinkman]

I only know 17bII and 35S, but IMHO you should merge both:
- 17bII style for the 42-like solver (use of soft keys for entries, alpha names, values)
- 35S style for the integration in programs (with name references in program, see below)

Improvements:
- on both calc you cannot reorder the eq list. There should be a function for hierarchical representation of the equations (only 1 level directory is enough), with names for the dir and for the equations
- HP35S allows to insert equations in programs, but not equations from the eq lib. Using names like in the 17bII should allow this function.

That was my 2 cents. You are not making any promises for creating such functions, but I promise I’ll buy this paid version of Free42 (“Pay42”? I’m joking).

Regards,

[rprosperi]

The Solver in the 27S/17B/17BII/17BII+/19B/19BII is not an 'Algebraic' solver in the classic sense, however I think this would be an excellent extension, and as many long-time members here have shown, it can be used for an extraordinary number of useful applications.

After exploring the responses, if you decide to pursue this, I'll send a machine with this "Pioneer Solver" (I'll send a non-buggy version, the 17BII+ versions are both buggy) that you can have for reference. In fact if you need to use it just to explore the idea, let me know (off list).

[mfleming]

I would certainly second Bob's suggestion to incorporate the Solver, but would extend that to the features offered by the 95/100/200LX Solver; loading & saving equation lists, and multi-line equation edit/formatting. The former feature is great for organizing equations into related groups, and the latter feature is essential as equations start to get really big. Oh, and make every function in the calculator available in an equation!

[SammysHP]

With all the new recent features that are not available in the HP 42s I wonder if it should still be called Free42. Also, while I think that every effort in creating a useful software should be appreciated in one way or the other (and if it's just by donating money), I heavily prefer open source software. So if this results in a paid closed source software, I'm not really interested in it.

But you could start by publishing a donation version of Free42 in the app stores (something in the $3-4 range) and see how many users are willing to pay for something that they can get for free. Sadly the stores keep a big part of the money. Adding a PayPal link in the app might work better, but is not allowed AFAIK (all in-app payments must be done through the app store).

[johanw]

(07-08-2020 09:16 PM)Thomas Okken Wrote:  like a big screen, big stack, units

That means backporting the DM42 code.

Quote:and algebraic equations.

from the discussion above I seem to not fully understand what this is supposed to do. At first read I tought it would be some kind of CAS system.

[SammysHP]

(07-09-2020 09:43 AM)johanw Wrote:  from the discussion above I seem to not fully understand what this is supposed to do. At first read I tought it would be some kind of CAS system.

I think the intention is to type algebraic equations in a natural way instead of writing a program expressing the same equation.

[Thomas Okken]

(07-09-2020 10:43 AM)SammysHP Wrote:  

(07-09-2020 09:43 AM)johanw Wrote:  from the discussion above I seem to not fully understand what this is supposed to do. At first read I tought it would be some kind of CAS system.

I think the intention is to type algebraic equations in a natural way instead of writing a program expressing the same equation.

Exactly. The aim is to be able to write, say, 'X^2-3*X+2' instead of LBL "FOO" MVAR "X" RCL "X" X^2 LASTX 3 * - 2 + END. But how exactly that should work is what I'm trying to figure out.

CAS is also a possibility, if I can find a suitable library to do the heavy lifting. But it's not the main objective at this point.

[johanw]

(07-09-2020 12:11 PM)Thomas Okken Wrote:  Exactly. The aim is to be able to write, say, 'X^2-3*X+2' instead of LBL "FOO" MVAR "X" RCL "X" X^2 LASTX 3 * - 2 + END. But how exactly that should work is what I'm trying to figure out.

OK, sorry but I missed that part of the calculator history so I'm not really knowledgable here. For my graduation I jumped immediately to CAS (Derive on MS-DOS), which resulted in such an ugly expression that I decided to solve it numerically by iteration.

Quote:CAS is a possibility, if I can find a suitable library to do the heavy lifting. But it's not the main objective at this point.

I know of https://github.com/ncalc/ncalc ,but that's in .NET and will probably work only in windows without too much glue code.

[SammysHP]

(07-09-2020 12:11 PM)Thomas Okken Wrote:  CAS is also a possibility, if I can find a suitable library to do the heavy lifting. But it's not the main objective at this point.

Xcas is used by the HP Prime and is available under GPL.

http://www-fourier.ujf-grenoble.fr/~parisse/giac.html

[Gene]

I still think merely dropping in the HP-17BII style equation solver would be very useful.

WORK=FORCE*DISTANCE

etc. would be useful without programming... of course, I can solve that formula without it. :-)

[johanw]

For that kind of solving I usually use this app: https://play.google.com/store/apps/detai...ab.android (oh well, its cracked paid version because the version I bought refuses to run on a device without a Google account).

But there is room for improvement there, it can solve a general 3rd order equation symbolic but fails for a 4rth order, while I remember having seen once a proof that 4rth and lower equtions are exact solvable (seen != understood. I was a 1st year student then).

[Vincent Weber]

Hi Thomas,

I concur with Thibault. I always thought that the 27S and the 42S are so complementary that they should be merged. So the complete 17BII/27S/19BII solver, with menu variables, L(), G(), S(), IF and SIGMA functions, should be implemented. I strongly suggest you read the 19B/27S technical application book, it really describes well the pitfalls of these functions. This is a real equation-based powerful programming language !

There are 2 improvements I can think of that are regrettably not present on the 27S:
1) make a equation a function (like DEF F(X,Y,..)=...) callable from another equation;
2) Indirect addressing. The ITEM function can return an element from a list from an index, but is not a L-value, so you cannot write to the list. This should be possible. Of course, you would need first to implement list-based statistics like on the 27S - or use 42S matrices instead ?

I also concur with Thibault that it would be awesome to use these equations in RPN programs, just like on the 32SII/33S/35S.

I think you could do this while staying true to yourself, i.e. keep the improvements discrete, without changing the normal operation of the 42S, like you did for LSTO and LRCL. This could be just another menu (EQU) that could be burried into the SOLVE menu (so that you don't have to provided a second shift key).

I'm really keen on the results ! A 27S/42S merger is a dream pocket machine !

Good luck and cheers,

Vincent

[BruceH]

(07-08-2020 09:16 PM)Thomas Okken Wrote:  What machine would you recommend for me to look at? The 27S, 19BII, 35S, something else?

Prior to getting a 42, I used an HP32S for my day-to-day machine. This is a lovely model and, even though it is RPN, it allows the use of an algebraic expression in the solver. This is incredibly handy, and much better than the 42S which requires you to write a small program to use the solver.

I would love to see a similar capability added to Free42.

[BruceH]

(07-09-2020 08:46 PM)Vincent Weber Wrote:  I concur with Thibault. I always thought that the 27S and the 42S are so complementary that they should be merged. So the complete 17BII/27S/19BII solver, with menu variables, L(), G(), S(), IF and SIGMA functions, should be implemented. I strongly suggest you read the 19B/27S technical application book, it really describes well the pitfalls of these functions. This is a real equation-based powerful programming language !

A stand-alone simulation of the 27S would be a challenge in itself. Fitting it alongside the 42S is double the effort again.

[Thomas Okken]

I'm not familiar with the 27S, or any of the other calcs mentioned earlier for that matter, but for what it's worth, I'm not planning on adding a full-blown algebraic mode. I just want to add algebraic expressions as a new data type, that can be executed using XEQ just like a program with MVAR instructions, and can be used from SOLVE and INTEG as well. It sounds like the 32S/32SII/33S/35S would be good examples for that kind of functionality?

I am still interested in learning about the other calculators as well, because even without an interactive algebraic mode, I'd like to see how existing calculators handle expression editing, storage, naming, evaluation, etc. Not to mention L() and G() and other esoterica.

And of course all the built-in functions should be callable from an expression, as well as other expressions, and keystroke programs. But now I'm getting ahead of myself.

[Logan]

Thomas,

This would be wonderful! Editing equations in RPN steps is a bit obscure at times so I really enjoyed the equation editor on both the 32s II and the 35s. I demonstrate them in these videos:
https://www.youtube.com/watch?v=c_2lrdDVS-A
https://www.youtube.com/watch?v=Ed2rt8fqX5s

For something on Free42, I could envision an "equation editor menu" which would give you quick access to parentheses, and left and right arrows for editing a specific location in the equation.

[Logan]

For additional reference, here is how the 17BII did it.

https://www.youtube.com/watch?v=l9ITvcC16Pk

[Marco Polo]

(07-08-2020 09:16 PM)Thomas Okken Wrote:  I am thinking about creating a paid version of Free42, with some of the enhancements I have been asked to do over the years, but didn't do because I felt they would take too much of my time to implement, like a big screen, big stack, units, and algebraic equations.

I am looking for inspiration about how to approach the algebraic equations in particular. I'm not looking to create another RPL machine, there are plenty of those already, and that really would be far too big a project for me to consider... but I know there are other, non-RPL, HP calculators with algebraic equations that people like, and maybe one of those would be a good starting point.

What machine would you recommend for me to look at? The 27S, 19BII, 35S, something else? Anything that I could run as an emulator to play with, or would be able to buy without spending an arm and a leg, and without having to embark on a restoration project, would obviously be preferred.

(I'm not making any promises yet as to whether this will ever see the light of day. Think of it as a feasibility study for now.)

A Free42/DM42 with units like 48s/g would be my immediate purchase.
Algebraic is a nice to have for me, but after units

Just my 2 cents

[johanw]

Something I would love to have, and would probably (I hope) be quite easy to implement, is an extension of the BASE menu to include some functions now found only on the 16C: 1 and 2 complement, signed / unsingned, and the option to work with bitmasks.