On the other hand, entering a real matrix ((1; 1); (1; 1)) and a second (complex) matrix ((1; 1); (1; i)) and dividing the first by the second results in c_matrix_division.png

This is the resulting matrix as it should look (compare Tom's post). Which release did you use?

The way it should look depends on how you do it.

inv(b) * a gives [[1 1][0 0]]

a * inv(b) gives [[1 0][1 0]]

The HP 50g, HP Prime and Casio fx-CG500 do it the first way and it looks like the 43S does it the second way.

Well it seems all HP calculators have implemented B/A as INV(A)*B, back to the HP-15C for which the manual says:

The ÷ function is useful for solving matrix equations of the form AX = B, where A is the coefficient matrix, B is the constant matrix, and X is the solution matrix. The descriptor of the constant matrix B should be entered in the Y-register and the descriptor of the coefficient matrix A should be entered in the X-register Pressing ÷ then calculates the solution X=A^-1*B

Both are possible with WP43S
One by using 1/X the other by using [M]\(^{-1}\)
However, in both cases the operation must be transportable through the different operating systems used.

@Walter
What I'm wondering is that if I first use [M]\(^{-1}\) and then multiplication it works. Isn't that what the program should do?

Last edited by PierreMengisen on Sun Aug 22, 2021 12:34 pm, edited 4 times in total.

Pierre
[TI59 with PC100C; TI-84 Plus CE-T; HP41CV with HP IL loop & 2*82161A DCD & 82162 TP; HP15C; HP28S; DM41; DM41L; DM42; DM41X]

A new indication: with the version wp43s_jm.exe it works as agreed.
What is the difference between wp43s.exe and wp43s_jm.exe?
(the first one is 2'034'348 bytes while the second one is 2'075'070 bytes

There is no intentional change in the Win Sim version of wp43s_jm. I included that in my last release for the first time thinking it may be useful to someone to have a second reference. The benefit of wp43s_jm is only in the hardware version where it works without qspi change and comes with USER mode pre-populated to the 43S L1 layout.

I am starting to suspect differences by different compile setups as I have also listed a bug that on my own Mac sim compiler my standard 3x3 A matrix inversion crashes the sim while the same Mac compile by Mihail (and Windows and DM42) works.

I will investigate by compiling the exact same release myself and comparing behaviour.

Jaco Mostert
Elec Eng, South Africa C43 (WP34C) on DM42 sn. 03818 & 06199 for complex math, HP42S; HP32Sii, WP34S&C, HP28C, HP35s, EL-506P, EL-W506, PB700; owned FX702P & 11C; used HP67 & HP85; iOS: 42s Byron, Free42, WP31S/34S, HCalc.

Both are possible with WP43S
One by using 1/X the other by using M^-1
However, in both cases the operation must be transportable through the different operating systems used.

@Walter
What I'm wondering is that if I first use [M]^-1 and then multiplication it works. Isn't that what the program should do?

Yes, in principle. But it looks like your problem is going to vanish with next release.

BTW, please note that 1/x and [M]^(-1) are two completely different cups of tea on the 43S.