bug in Ehv::isSubset(): misusage of size()

[kroeker@…]

Hello,

the (public) routine

isSubset() in "ehv.lib" is incorrect due to incorrect usage of size() instead of ncols() when iterating over all ideal generators.

Here is a failing example:

    ring  r = 0,(x,y),dp;
    ideal J = x;
    ideal I = x, 0 , y;
    isSubset( I, std(J) ); //=1, wrong!
    isSubset( std(I), std(J) ) ; //=0, ok
    ASSUME( 0, isSubset( I, std(J) ) == isSubset( std(I), std(J) ) ); //fails

Also isSubset() can be modified to be more user friendly and less error-prone; that means:

• as a user I would expect to get the correct answer, even if J is not standard basis
• the std computation of J should be done by isSubset(), but only if necessary.

If a nonconservative version of isSubset is required, for whatever reasons, it could be added e.g. as isSubsetUnsafe

A test, if an ideal I is contained is ideal J is not special for a primary decomposition, therefore this method should move from "ehv.lib" to "ring.lib".

Bugfix is in preparation.

Jack

[kroeker@…]

see pull request

​https://github.com/Singular/Sources/pull/495