libsingular interface crashes when calling certain functions with integer coeffficients

[miguelmarco]

Steps To Reproduce

In a fresh sage session, run

sage: R.<x,y,z> = ZZ[]
sage: M = matrix(R,[[x,y,x],[x^2+z,y^2,x+y-z]])
sage: from sage.libs.singular.function import singular_function
sage: std = singular_function("std")
sage: std(M)

Expected Behavior

A result similar to what you get with rational coefficients:

sage: S.<x,y,z> = QQ[]
sage: N = matrix(S,[[x,y,x],[x^2+z,y^2,x+y-z]])
sage: std(N)
[(x, x + y - z), (y, y^2), (x^2 - x*y + x*z - x + y, 2*y*z - z^2 - z), (x*y*z - x*y + y^2 - 3*y*z, -y*z^2 - y*z), (y^3 - 2*y^2*z - 2*y^2 + 5*y*z, y*z^2 + y*z), (x*y^2 - x*y - y^2 + y*z, 0)]

Actual Behavior

Get this error message

---------------------------------------------------------------------------
SignalError                               Traceback (most recent call last)
Cell In [5], line 1
----> 1 std(M)

File ~/sage/src/sage/libs/singular/function.pyx:1312, in sage.libs.singular.function.SingularFunction.__call__()
   1310     if not (isinstance(ring, MPolynomialRing_libsingular) or isinstance(ring, NCPolynomialRing_plural)):
   1311         raise TypeError("cannot call Singular function '%s' with ring parameter of type '%s'" % (self._name, type(ring)))
-> 1312     return call_function(self, args, ring, interruptible, attributes)
   1313 
   1314 def _instancedoc_(self):

File ~/sage/src/sage/libs/singular/function.pyx:1491, in sage.libs.singular.function.call_function()
   1489     error_messages.pop()
   1490 
-> 1491 with opt_ctx:  # we are preserving the global options state here
   1492     if signal_handler:
   1493         sig_on()

File ~/sage/src/sage/libs/singular/function.pyx:1493, in sage.libs.singular.function.call_function()
   1491 with opt_ctx:  # we are preserving the global options state here
   1492     if signal_handler:
-> 1493         sig_on()
   1494         _res = self.call_handler.handle_call(argument_list, si_ring)
   1495         sig_off()

SignalError: Segmentation fault

Additional Information

No response

Environment

- **OS**: gentoo linux
- **Sage Version**: 10.2.rc0

Checklist

• [X] I have searched the existing issues for a bug report that matches the one I want to file, without success.
• [X] I have read the documentation and troubleshoot guide

[mkoeppe]

config.log please so we can see where your libraries are coming from

[kwankyu]

Is it supposed to work? Does it work on singular directly? I have no experience of working with singular for polynomials with integer coefficients.

[miguelmarco]

Is it supposed to work? Does it work on singular directly? I have no experience of working with singular for polynomials with integer coefficients.

Yes, it does work in singular. In fact, it does work with the old (pexpect-based) interface:

sage: R.<x,y,z> = ZZ[]
sage: M = matrix(R,[[x,y,x],[x^2+z,y^2,x+y-z]])
sage: M._singular_().std()._sage_()
[                            x                             y       x^2 - x*y + x*z - x + y     x*y*z - x*y + y^2 - 3*y*z y^3 - 2*y^2*z - 2*y^2 + 5*y*z       x*y^2 - x*y - y^2 + y*z]
[                    x + y - z                           y^2               2*y*z - z^2 - z                  -y*z^2 - y*z                   y*z^2 + y*z                             0]

[miguelmarco]

config.log please so we can see where your libraries are coming from

config.log.txt

[miguelmarco]

Does it work for you? It crashes in every system I tested.

[miguelmarco]

ping @kwankyu @mkoeppe

[kwankyu]

It works with Singular

                     SINGULAR                                 /  Development
 A Computer Algebra System for Polynomial Computations       /   version 4.3.2
                                                           0<
 by: W. Decker, G.-M. Greuel, G. Pfister, H. Schoenemann     \   Feb 2023
FB Mathematik der Universitaet, D-67653 Kaiserslautern        \
> ring R = integer,(x,y,z), dp;
> vector s1 = [x,y,x];
> vector s2 = [x^2+z,y^2,x+y-z];
> module m = s1,s2;
> std(m);
_[1]=x*gen(3)+x*gen(1)+y*gen(2)
_[2]=x2*gen(1)+y2*gen(2)-x*gen(1)+y*gen(3)-y*gen(2)-z*gen(3)+z*gen(1)

In Sage,

sage: R.<x,y,z> = ZZ[]
sage: M = matrix(R,[[x,y,x],[x^2+z,y^2,x+y-z]])
sage: std(M,ring=R)
[(x, x + y - z), (y, y^2), (x^2 - x*y + x*z - x + y, 2*y*z - z^2 - z), (x*y*z - x*y + y^2 - 3*y*z, -y*z^2 - y*z), (y^3 - 2*y^2*z - 2*y^2 + 5*y*z, y*z^2 + y*z), (x*y^2 - x*y - y^2 + y*z, 0)]

which seems wrong. For std(M), it seems to have never worked. I checked with 9.7 and later.

So if you want this "feature", you are on your own.

[miguelmarco]

I guess what happens in the example you posted has to do with considering the module as represented by the rows of the matrix vs the one corresponding to the columns:

sage: std(M.T,ring=R)
[(x, y, x), (x^2 - x + z, y^2 - y, y - z)]

So it seems that the problem comes from not using the ring keyword. That is interesting, since that keyword is not needed when the polynomial ring is defined over a field.

[kwankyu]

Right... I didn't see that. So the "fix" may be easy.

[kwankyu]

You may need to look into src/sage/libs/singular/function.pyx.

[miguelmarco]

I see. It seems that the difference from the field case is the type of the matrix:

sage: type(M)
<class 'sage.matrix.matrix_generic_dense.Matrix_generic_dense'>

whereas the code that tries to guess the ring of the input checks for more specific types:

https://github.com/sagemath/sage/blob/f10820ff877ea61a2fdbf0d7fc3c1445ead77d72/src/sage/libs/singular/function.pyx#L1388C1-L1391C38

[miguelmarco]

So I guess this could be fixed by using Matrix_mpolynomial_dense for matrices with coefficients in a integer polynomial ring.

In the meantime, just specifying the ring seems to solve it. Thank you!