regression in solve

[zimmermann6]

the following example, which we use in our book (in french) about Sage, does not work any more with Sage 5.7:

With Sage 5.1:

sage: var('y z')
sage: solve([x^2*y*z==18,x*y^3*z==24,x*y*z^4==6],x,y,z)                        
[[x == 3, y == 2, z == 1], [x == (1.33721506733 - 2.68548987407*I), y == (-1.70043427146 + 1.05286432575*I), z == (0.932472229404 - 0.361241666187*I)], [x == (1.33721506733 + 2.68548987407*I), y == (-1.70043427146 - 1.05286432575*I), z == (0.932472229404 + 0.361241666187*I)], [x == (-2.55065140719 - 1.57929648863*I), y == (-0.547325980144 + 1.92365128635*I), z == (-0.982973099684 - 0.183749517817*I)], [x == (-2.55065140719 + 1.57929648863*I), y == (-0.547325980144 - 1.92365128635*I), z == (-0.982973099684 + 0.183749517817*I)], [x == (0.27680507839 - 2.98720252889*I), y == (1.47801783444 - 1.34739128729*I), z == (-0.85021713573 - 0.526432162877*I)], [x == (0.27680507839 + 2.98720252889*I), y == (1.47801783444 + 1.34739128729*I), z == (-0.85021713573 + 0.526432162877*I)], [x == (-0.820988970216 + 2.88547692952*I), y == (-1.20526927276 - 1.59603445456*I), z == (0.0922683594633 - 0.995734176295*I)], [x == (-0.820988970216 - 2.88547692952*I), y == (-1.20526927276 + 1.59603445456*I), z == (0.0922683594633 + 0.995734176295*I)], [x == (-1.80790390914 - 2.39405168184*I), y == (0.891476711553 - 1.79032658271*I), z == (0.739008917221 - 0.673695643647*I)], [x == (-1.80790390914 + 2.39405168184*I), y == (0.891476711553 + 1.79032658271*I), z == (0.739008917221 + 0.673695643647*I)], [x == (2.21702675166 + 2.02108693094*I), y == (1.86494445881 + 0.722483332374*I), z == (-0.273662990072 - 0.961825643173*I)], [x == (2.21702675166 - 2.02108693094*I), y == (1.86494445881 - 0.722483332374*I), z == (-0.273662990072 + 0.961825643173*I)], [x == (2.79741668821 - 1.08372499856*I), y == (-1.96594619937 + 0.367499035633*I), z == (-0.602634636379 - 0.79801722728*I)], [x == (2.79741668821 + 1.08372499856*I), y == (-1.96594619937 - 0.367499035633*I), z == (-0.602634636379 + 0.79801722728*I)], [x == (-2.94891929905 + 0.55124855345*I), y == (0.184536718927 + 1.99146835259*I), z == (0.445738355777 - 0.895163291355*I)], [x == (-2.94891929905 - 0.55124855345*I), y == (0.184536718927 - 1.99146835259*I), z == (0.445738355777 + 0.895163291355*I)]]

With Sage 5.7:

sage: var('y z')                                    
(y, z)
sage: solve([x^2*y*z==18,x*y^3*z==24,x*y*z^4==6],x,y,z)                        
[x^2*y*z == 18, x*y^3*z == 24, x*y*z^4 == 6]

Cf http://sagebook.gforge.inria.fr/

Upstream: Fixed upstream, in a later stable release.

CC: @mezzarobba

Component: calculus

Author: Karl-Dieter Crisman

Branch/Commit: 0585302

Reviewer: Peter Bruin

Issue created by migration from https://trac.sagemath.org/ticket/14306

[kcrisman]

comment:1

I've reported this at Maxima bug 2564.

[kcrisman]

Upstream: Reported upstream. No feedback yet.

[zimmermann6]

Changed upstream from Reported upstream. No feedback yet. to Fixed upstream, but not in a stable release.

[zimmermann6]

comment:3

see #13973

[jdemeyer]

Changed upstream from Fixed upstream, but not in a stable release. to Fixed upstream, in a later stable release.

[kcrisman]

comment:8

Wait a minute. This is already almost a doctest in Sage.

Sage example in ./calculus.tex, line 583::

    sage: x, y, z = var('x, y, z')
    sage: solve([x^2 * y * z == 18, x * y^3 * z == 24,\
    ....:        x * y * z^4 == 3], x, y, z)
    [[x == (-2.76736473308 - 1.71347969911*I), y == (-0.570103503963 + 2.00370597877*I), z == (-0.801684337646 - 0.14986077496*I)], ...]

and it passes

$ ./sage -t src/sage/tests/french_book/calculus_doctest.py 
Running doctests with ID 2014-05-30-08-56-36-7a1afdf4.
Doctesting 1 file.
sage -t src/sage/tests/french_book/calculus_doctest.py
    [246 tests, 7.59 s]
----------------------------------------------------------------------
All tests passed!
----------------------------------------------------------------------
Total time for all tests: 7.7 seconds
    cpu time: 5.6 seconds
    cumulative wall time: 7.6 seconds

But assuming we want to add this back: Paul, where should this appear - presumably in src/sage/tests/french_book/calculus_doctest.py ?

[zimmermann6]

comment:9

in our book we originally had the example in the description (at that time we were using Sage 5.1) but
we had to change the last equation from x*y*z^4==6 to x*y*z^4==3 to get Sage to really solve the system (the final version of the book is based on Sage 5.9).

With x*y*z^4==6 as last equation, Sage 6.2 is still unable to solve the system as Sage 5.1 did in the description, thus the problem is still open for me.

Paul

[kcrisman]

comment:10

Ah, but with #13973 we have

sage: var('y,z')
(y, z)
sage: solve([x^2*y*z==18,x*y^3*z==24,x*y*z^4==6],x,y,z)                        
[[x == 3, y == 2, z == 1], [x == (1.33721506733 - 2.68548987407*I), y == (-1.70043427146 + 1.05286432575*I), z == (0.932472229404 - 0.361241666187*I)],

As you see, Maxima fixed it. So are you saying that you don't need this for your book? We can still add this as a doctest elsewhere, of course.

[zimmermann6]

comment:11

yes we don't need this (with x*y*z^4==6) for the current version of our book.
But it would be good to add this as a doctest, to check the Maxima fix remains ok.

Paul

[kcrisman]

Branch: u/kcrisman/14306-maxima_solve_doctest

[kcrisman]

Commit: 9834929

[kcrisman]

Author: Karl-Dieter Crisman

[kcrisman]

Last 10 new commits:

13d48c2	Trac 13973: fix doctests with changed floating point format/precision
a130eed	Trac 13973: realpart should be real_part
3d927bf	Merge remote-tracking branch 'trac/u/pbruin/13973-maxima_update' into ticket/13712
c715aab	Merge branch 'public/ticket/inf_sum_doctest-13712' of git://trac.sagemath.org/sage into ticket/11894-maxima_sum_zero_division
1dd0f05	Trac 11894: add doctest for error detection in Maxima sum
0ede45c	Merge branch 'u/pbruin/11894-maxima_sum_zero_division' of trac.sagemath.org:sage into maxima_upgrade
b6657b9	Trac #15386 - document that Maxima 5.33 fixes this limit
54d0609	Trac #14209 - verify Maxima fix for doing improper integral twice
999e21f	Trac #14209 - Forget assumptions from earlier doctest
9834929	Trac 14306 - doctest Maxima bugfix in solve

[kcrisman]

comment:13

Once again sorry for the looong branch - I'll do this better after the current Maxima upgrade things are done.

[pjbruin]

comment:14

The last colon before the example should be a double colon.

[sagetrac-git]

Changed commit from 9834929 to 0585302

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

0585302

Add second colon for doctest

[kcrisman]

comment:16

The last colon before the example should be a double colon.

Right you are! I've fixed this now.

---

I just want to point out that in Mercurial, I could always "roll back" the previous commit and fix such trivialities. Now it has to clutter up the log/tree/directed acyclic graph/whatever for a one-character fix.

---

New commits:

0585302

Add second colon for doctest

---

New commits:

0585302

Add second colon for doctest

[nbruin]

comment:17

Replying to @kcrisman:

I just want to point out that in Mercurial, I could always "roll back" the previous commit and fix such trivialities. Now it has to clutter up the log/tree/directed acyclic graph/whatever for a one-character fix.

You still can. You'll just have to "force push" and deal with a grumpy Volker. It's a trade-off: do you want a clean history or a (small) chance for more difficult merge issues later on, in case someone based some work on your now abandoned commit?

[pjbruin]

Reviewer: Peter Bruin

[vbraun]

Changed branch from u/kcrisman/14306-maxima_solve_doctest to 0585302