# Unhandled SIGABRT when working with polynomial rings

Hello, experts!

I am working with multivariate polynomial rings in a finite field, such as

K.<a1,b1,a2,b2,x>=PolynomialRing(GF(32003))


But when I try to define an object like this one:

c1=numerator((a1+b1*I)+(a1-b1*I)*x^(-5))


I get a SIGABRT from Sage and the program terminates giving an Ubuntu error (something like "gdb stopped working").

This pastebin shows the output: http://pastebin.com/jKEM87aS. Even though it says Saved trace to /home/slenderman/.sage/crash_logs/crash_oImBtz.log, that file is totally empty.

More information: if I use the ring over Q it works perfectly

K.<a1,b1,a2,b2,x>=QQ[]


Also, I have Sage version 7.2 compiled from the git repository.

Thank you in advanced for any help!

edit retag close merge delete

What is I ? You should rather not use symbolic variables in polynomials. Otherwise, everything becomes symbolic, hence slow and better avoided.

( 2016-08-09 11:11:08 +0200 )edit

@FrédéricC You are right, but even if I declare the ring with I as a variable K.<a1,b1,a2,b2,x,I>=PolynomialRing(GF(32003)) I get the same SIGABRT error.

( 2016-08-09 11:19:14 +0200 )edit

Sort by » oldest newest most voted

I cannot reproduce your error, the error I get is :

PariError: I already exists with incompatible valence


which is fine. Indeed, the product of a polynomial indeterminate defined over a finite field and I (which belongs to a field of characteristic 0) is not well defined. If I is defined as another indeterminate, everything is fine:

sage: K.<a1,b1,a2,b2,x,I>=PolynomialRing(GF(32003))
sage: numerator((a1+b1*I)+(a1-b1*I)*x^(-5))
b1*x^5*I + a1*x^5 - b1*I + a1


In your case, the error is more annoying since it is not handled by Sage, and as the log says "his probably occurred because a compiled module has a bug in it and is not properly wrapped with sig_on(), sig_off(). Python will now terminate."

When you compiled Sage, and especially Pari, did you get any error ? Coud you look at the file $SAGE_ROOT/logs/pkgs/pari-2.8-*.log to see if something wrong happened there ? If you do not find anything, could you try, from the $SAGE_ROOT direcory:

 ./sage -i -f -c pari


To rebuild pari and run a self-check ?

more

The Pari build log didn't have any errors in it, but your suggestion of rebuilding Sage and running a self check worked like a charm, thank you!

( 2016-08-10 10:13:39 +0200 )edit