Trying to fix trac #18488(solve() ignores conjugate)

[Utkarsh Sharma]

I recently came across https://trac.sagemath.org/ticket/18488 and I wanted to try fixing the issue. After going through past discussions and the solve() method in src/sage/symbolic/relation(which, I believe, is the solve() method being referred to here), I figured out that it uses maxima's solve() method, so the bug must be in maxima's solve() method.

So how do I go about solving this issue? one obvious solution would be to fix the maxima bug, but is there anything else that can be done? I think first we could check whether the bug is being encountered or not, and if it was encountered, We could provide an alternate solution, like solving for the real and imaginary parts separately, or maybe solving for the conjugate and then take the conjugate of both sides of the equation. Those are just my initial thoughts, I thought it would be best to start a conversation here before going any further.

[Thierry]

Hi,

On Sat, Jun 25, 2022 at 01:47:52PM -0700, Utkarsh Sharma wrote:
> I recently came across https://trac.sagemath.org/ticket/18488 and I wanted
> to try fixing the issue. After going through past discussions and the
> solve() method in src/sage/symbolic/relation(which, I believe, is the
> solve() method being referred to here), I figured out that it uses maxima's
> solve() method, so the bug must be in maxima's solve() method.
>
> So how do I go about solving this issue? one obvious solution would be to
> fix the maxima bug, but is there anything else that can be done?

The following procedure is common:

- reproduce the bug within maxima (you can get a maxima console with
"sage -maxima" command)
- if the bug shows an issue in the interface between Sage and maxima,
fix it within Sage (with a corresponding doctest)
- if the bug actually comes from maxima, then report the bug upstream
https://sourceforge.net/p/maxima/bugs/
- once the bug is fixed, either wait for the new maxima version, or add
the corresponding patch in SAGE_ROOT/build/pkgs/maxima/patches/ so
that the bug is fixed in Sage (depending on the emergency)
- a common mistake at this point is to close the Sage ticket. Don't do
that. Instead, add a doctest showing that the bug is fixed, so that we
get informed if it reappears.

Ciao,
Thierry

> I think
> first we could check whether the bug is being encountered or not, and if it
> was encountered, We could provide an alternate solution, like solving for
> the real and imaginary parts separately, or maybe solving for the conjugate
> and then take the conjugate of both sides of the equation. Those are just
> my initial thoughts, I thought it would be best to start a conversation
> here before going any further.
>
>

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/e3a31993-31c5-4572-8ef8-7d668a15fb87n%40googlegroups.com.