negative leading coefficients still occur for std over integers

origin: https://github.com/surface-smoothers/lipmanresolve/issues/316

example:

option("redSB");
LIB("primdecint.lib");
proc stdZW(I)
{ return(  Primdecint::stdZ(I) );
}
ring rng = integer,(x, y, z, v),dp;

ideal Q =  8*y-2*z-5*v,
 2*x*z+5*x*v+2*z-3*v,
 x*y-7*y+2*z+4*v,
 z*v^2-2*y^2,
 5*x*v^2-3*v^2+4*y,
 4*y*z*v-z^2*v-5*y^2,
 x*z*v+z*v-2*y,
 5*x^2*v+2*x*v-3*v+4,
 x*z^2+z^2-4*z*v+5*y,
 x^2*z-5*x*v-z+3*v-2;
 Q = std(Q); 
qring q = Q;

ideal I = 2, v,y^2;
ideal J = v,2*z,2*y,y*v,y^2;

ideal IJ = intersect(I,J);
ideal stdIJ = std(IJ);
stdIJ;
ideal stdIJZ = stdZW(IJ);
stdIJZ;
int k;
 for(k=1; k <= ncols(stdIJ); k++)
{
 if(leadcoef(stdIJ[k]) < 0)
 {
    ERROR(" Warning: negative leadcoef occured");
 }
}

update: error occurs if option "redSB" is set! jakobkroeker jakobkroeker added the bug label on Jul 12 fruhbis-kruger-anne fruhbis-kruger-anne was assigned by jakobkroeker on Jul 12 jakobkroeker jakobkroeker added the question label on Jul 12 jakobkroeker jakobkroeker added this to the Singular 4.1.0 milestone on Jul 12 fruhbis-kruger-anne fruhbis-kruger-anne commented on Jul 13

First a remark on the input: Please always make sure that Q is a standard basis before defining a qring using it.

Now to the problem: The negative leading coefficient is indeed very annoying. Adi needs to know about it. On the other hand, I get the following output

std(IJ);
_[1]=4
_[2]=v
_[3]=2y
_[4]=y2
_[5]=-2z-4v

which is even more annoying, because even with redTail set I do not get rid of the -4v which obviously is a product of two other generators. jakobkroeker jakobkroeker commented on Jul 14

which is even more annoying, because even with redTail set I do not get rid of the -4v which obviously is a product of two other generators.

update: error occurs if option "redSB" is set, otherwise not! fruhbis-kruger-anne fruhbis-kruger-anne commented on Jul 14

That fits. In the C-code, this option forces a different strategy and hence certain different internal routines than in the other case.

--> need to look at it together with Adi Adi Popescu adipopescu referenced this issue in adipopescu/To-Do-List on Jul 15 Open Negative Coeffcient #19 Adi Popescu adipopescu commented on Jul 16

Fixed: in https://github.com/adipopescu/Sources/tree/NegativeCoef

Now the output is stdIJ[1]=4 stdIJ[2]=v stdIJ[3]=2z stdIJ[4]=2y stdIJ[5]=y2

for both ideals. Adi Popescu adipopescu referenced this issue in Singular/Sources on Jul 16 Closed Negative coef #617 jakobkroeker jakobkroeker added the pending label on Jul 18 Adi Popescu adipopescu commented on Oct 10

This bug was solved completely in my other branch (SignatureSB). There were some bugs in the redtailBba which where solved there. Therefore i close this push request.

jakobkroeker / test_singular

negative leading coefficients still occur for std over integers #85