desolve_system unable to interpret ilt and Maxima's temporary variables

[rhinton]

desolve_system sometimes generates a Maxima result that includes temporary variables that Sage does not parse correctly.

sage: t = var('t')
sage: x1 = function('x1', t)
sage: x2 = function('x2', t)
sage: de1 = (diff(x1,t) == -3*(x2^2-1))
sage: de2 = (diff(x2,t) == 1)
sage: desolve_system([de1, de2], [x1, x2], ivar=t)
...
TypeError: unable to make sense of Maxima expression 'x1(t)=ilt(-((3*laplace(x2(t)^2,t,?g1543)-x1(0))*?g1543-3)/?g1543^2,?g1543,t)' in Sage 

Upstream: Fixed upstream, in a later stable release.

CC: @robert-marik

Component: calculus

Keywords: maxima, symbolics

Author: Emmanuel Charpentier

Branch/Commit: 74d88f4

Reviewer: Karl-Dieter Crisman

Issue created by migration from https://trac.sagemath.org/ticket/9825

[kcrisman]

comment:1

On [ this Maxima list thread] we get the original system in Maxima notation - thanks to Stavros Macrackis:

de1: diff(x1(t),t)=-3*(x2(t)^2-1);
de2: diff(x2(t),t)=1;
desolve([de1,de2],[x1(t),x2(t)]);

He also provides a simpler example which does this:

desolve([diff(f(x),x)=f(x^2)],[f(x)]);

The suggestion is that the ilt should be replacing the ?g1234 type variables (which are indeed dummy variables, but native Lisp ones) by Maxima-type ones, so I am putting to reported upstream, developers acknowledge bug. However, my feeling is that we should fix this by parsing these things as well, should they come up again.

[kcrisman]

comment:2

Replying to @kcrisman:

On [ this Maxima list thread] we get the original system in Maxima notation - thanks to Stavros Macrackis:

Meaning this thread.

[kcrisman]

comment:3

I've followed up again at this new thread - apparently it never actually made it to their bug tracker?

[kcrisman]

comment:4

See also this ask.sagemath question.

[kcrisman]

comment:5

And this ask.sagemath question, though here Maxima is actually asking a question about these variables!

[rwst]

Upstream: Reported upstream. No feedback yet.

[kcrisman]

comment:10

Did you report it upstream to their bug tracker? I never heard on either of these emails, so I think this is how it will have to be reported.

[rwst]

comment:11

I took your repeated mail to the list as report.

[kcrisman]

comment:12

Sadly, that isn't always enough :( Reported upstream here, however, just now. There was internal discussion in the original Maxima thread so I took it that the experts had several possible resolutions.

[kcrisman]

Changed upstream from Reported upstream. No feedback yet. to Fixed upstream, in a later stable release.

[kcrisman]

comment:14

Upstream seems to have made a change that would do something about this. Anyone want to give it a whirl?

[rwst]

comment:15

This now returns [x1(t) == ilt(-(3*g3390*laplace(x2(t)^2, t, g3390) - g3390*x1(0) - 3)/g3390^2, g3390, t), x2(t) == t + x2(0)] so it becomes an issue to fix our Maxima interface.

[nbruin]

comment:16

Replying to @rwst:

This now returns [x1(t) == ilt(-(3*g3390*laplace(x2(t)^2, t, g3390) - g3390*x1(0) - 3)/g3390^2, g3390, t), x2(t) == t + x2(0)] so it becomes an issue to fix our Maxima interface.

According to the documentation of inverse_laplace this is probably more or less correct. We might want to do something about "ilt" so that it is more closely tied to inverse_laplace, though.

[EmmanuelCharpentier]

comment:17

This is now fixed (probably due to upstream upgrade) :

sage: x1, x2=function("x1, x2")
sage: de1=x1(t).diff(t)==-3*(x2(t)-1)
sage: de2=x2(t).diff(t)==1
sage: Sol=desolve_system([de1, de2],[x1(t),x2(t)],ivar=t) ; Sol
[x1(t) == -3/2*t^2 - 3*t*x2(0) + 3*t + x1(0), x2(t) == t + x2(0)]

==> invalidation of the bug and review query in order to get this bug closed.

HTH,

[kcrisman]

comment:18

I guess as usual doctest needed? Looks like it was a combination of their upstream fix and something we did to parse it right.

[EmmanuelCharpentier]

Branch: u/charpent/desolve_system_unable_to_interpret_ilt_and_maxima_s_temporary_variables

[EmmanuelCharpentier]

Commit: 74d88f4

[EmmanuelCharpentier]

comment:20

Doctest added.

HTH,

---

New commits:

74d88f4

Document the fix of #9825

[kcrisman]

Reviewer: Karl-Dieter Crisman

[kcrisman]

comment:21

Thanks, this is great. Despite patchbot not yet reporting and my own Sage install being too brittle to test, Cell server says it's fine so let's do it.

[kcrisman]

Author: Emmanuel Charpentier

[vbraun]

Changed branch from u/charpent/desolve_system_unable_to_interpret_ilt_and_maxima_s_temporary_variables to 74d88f4