[wp31s] simple equation editor ?

[Stephan2]

Would a (simple) equation editor for the 31s be feasible (and desirable)?

What I have in mind is not programmability but rather entering straight forward
equations using rpn. Those equations would have a few variables. The user would
then be prompted for all but one values. The calculator then solves for this variable.

There would be no loops, branches or subroutines. Length and complexity could be
limited as appropriate. This should remain really simple and straight forward. For
everything else there is the 34s or excel. The aim should not be feature bloat but a
simple and intuitive feature in the spirit of the 31s.

Example: Volume of a cone, V = 1/3 * pi * r^2 * h

Suggested steps:
1) press f [more] EQNED
2) prompt: edit eqn#?
3) press 1 for first eqn out of 10 possible
4) key in
3
1/x
pi
*
R
x^2
*
H
*
V
-
[exit]
5) press f [more] EQNSLV
6) prompt: solve eqn#?
7) press 1
8) prompt: slv for: R H V?
9) press H
10) prompt: R?
11) enter value for R
12) repeat for V
13) calculator solves for H

Would this be possible?

[matthiaspaul]

This would require the calculator to support symbols. However, symbolic mathematics are the domain of RPL calculators rather than RPN calculators. While this might be a feature for the hypothetical "WP 43s", I don't think it would fit into the concept of the WP 31s.

Greetings,

Matthias

[Stephan2]

It would be a numerical approximation - basically a simplified version of the SLV
command available on the 34s and many other calcs.

The calculator just prompts for numbers and substitutes these for the variables. It
then solves using the algorithm used eg. in the 34s. The values chosen by the solve
algorithm then go into the remaining variable. The calc then iterates until the overall
result is close enough to zero.

The idea is to reduce complexity by avoiding full programmability in favour of a simpler
system. I believe some calcs implement this. Maybe the 33s or 35s. I don't have
neither for comparison.

[Marcus von Cube]

No one is asking for a symbolic solution, a numeric solver would do nicely. The problem is RAM. We have already restricted the number of registers to 10. This allows the implementation of full UNDO.

If we allow for equations with up to 26 variables we would have to increase the register space. And where to store the equations? We do not have program memory.

Conclusion: This seems to be out of scope for the present code base.

[Stephan2]

Couldn't the equations go into flash? Unfortunately I'm completely unfamiliar with the
memory layout of calculators. I don't think 26 variables would be necessary. Maybe
5-8 would be sufficient for many cases. And could maybe the UNDO be suspended
while the solver runs? I'm just asking naively.

[Paul Dale]

This might be possible.

Equations would need to go into flash, but we'd need to store them in RAM on entry. Both are possible.

The 31S supports keystroke programs internally (for XROM functions) but all the entry/edit support has been removed so these programs could be run -- a lot of the branching and looping infrastructure are still there.


It would be quite a bit of work and you'd end up with something looking a lot like the 34S minus some functions.



- Pauli

[Stephan2]

The intuitive usability should be a priority on the 31s, I really like it.
Here is a somewhat simplified proposal:

- simply use the existing registers instead of the proposed letter variables
- the equation above would then be
3 1/x
pi *
RCL0 x^2 *
RCL1 *
RCL2 -
[exit]
- the user provides values in the registers except for the one to solve for
- EQNSLV would then prompt for EQN# and R# to solve
- the solver then put its guesses into the register chosen and run the equation
- the result would be in X and the chosen register
- an equation editor could wait until a possible phase 2. Until then, there would
be useful preprogrammed routines provided in flash (TVM maybe?)

Necessary changes would be:
- new command EQNSLV in [more], prompts user for equation and register
- solver from 34s, modified to store guess in register (basically STO-> R#)
- ability to run command sequences from flash (no branches or looping necessary)
- no further new commands, changes to the front plate or anything else

I believe this could be an easy to use and powerful feature with many useful applications.

[BarryMead]

(02-15-2015 09:58 AM)Stephan2 Wrote:  - simply use the existing registers instead of the proposed letter variables
- the equation above would then be
3 1/x
pi *
RCL0 x^2 *
RCL1 *
RCL2 -
[exit]

Would adding this feature (similar to a small program) jeopardize the wp-31s's status as acceptable on college exams? On a recent post I heard that the calculator was deemed "Acceptable" on college exams because it has no "Program Memory" where students can tuck away formulas/answers to tricky exam questions. Any thoughts?

[Stephan2]

If necessary there could be a variant with the equation system disabled, possibly called
wp31se (scientific equation) vs. wp31s if they need to be recognisable from the outside.
There already are a number of firmware variants.

Ultimately it is an open system - anybody can run anything on it. Short of verifying the
firmware in depth, you can only trust the student or ban flashable calcs completely.

[Paul Dale]

Perhaps the 32sii algebraic equations might be more palatable than programmability.

Still, the key factor is finding someone who wants to implement this


- Pauli