Skip to content

Bug in CosetLeadersMatFFE #5923

@osj1961

Description

@osj1961

For matrices with entries in 8-bit fields, this function can return a list of coset leaders w/ missing entries.
A minimal example is

gap> M := One(GF(4))*[[1,0,Z(4)],[0,1,Z(4)^2]];
[ [ Z(2)^0, 0*Z(2), Z(2^2) ], [ 0*Z(2), Z(2)^0, Z(2^2)^2 ] ]
gap> L := CosetLeadersMatFFE(M,GF(4));
[ [ 0*Z(2), 0*Z(2), 0*Z(2) ],, [ 0*Z(2), Z(2^2), 0*Z(2) ], 
  [ 0*Z(2), Z(2)^0, 0*Z(2) ],, [ 0*Z(2), Z(2)^0, Z(2^2) ], 
  [ 0*Z(2), Z(2^2), Z(2^2) ], [ 0*Z(2), 0*Z(2), Z(2^2) ], 
  [ Z(2^2), 0*Z(2), 0*Z(2) ], [ 0*Z(2), 0*Z(2), Z(2)^0 ], 
  [ 0*Z(2), Z(2^2)^2, Z(2)^0 ], [ 0*Z(2), Z(2)^0, Z(2)^0 ], 
  [ Z(2)^0, 0*Z(2), 0*Z(2) ] ]

where the 2nd and 5th entries of L are unbound.

I discovered this issue in trying to track down a problem with the covering radius computation for error-correcting codes in the Guava package. It could well be the case that Guava is the only place where this function is used. If so, it would seem to be reasonable to remove this function from GAP and recreate the functionality in Guava. (I've done exactly that as a stop-gap measure for the latest Guava release - but my hack loses the faster approach available for 8-bit fields.) If, on the other hand, other parties are also using this function it should be repaired.

Metadata

Metadata

Assignees

No one assigned

    Labels

    kind: bugIssues describing general bugs, and PRs fixing themkind: bug: wrong resultIssues describing bugs that result in mathematically or otherwise wrong results, and PRs fixing them

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions