(Normal)Intersection produces wrong result, probably due to bug in IsNormal

[stertooy]

Given the following group G and subgroups H and K, it seems the intersection being calculated is wrong:

gap> G := ExamplesOfSomePcpGroups( 1 );;
gap> H := Subgroup( G, [ G.1*G.4^5, G.1*G.2^4*G.3*G.4 ] );;
gap> K := Subgroup( G, [ G.1^-1*G.2^2*G.3^-1*G.4^-2, G.1^-1*G.2^2*G.3^-2 ] );;
gap> I := Intersection( H, K );;
gap> IsSubgroup( H, I );
true
gap> IsSubgroup( K, I );
false

The Intersection command will call NormalIntersection( K, H ), which is where the error happens. Turning on all the assertion variables (CHECK_IGS, etc) doesn't detect any problems, unfortunately.

After some digging, I found the following awkward result:

gap> IsNormal( H, K );
true
gap> HK := ClosureGroup( H, K );;
gap> IsNormal( HK, K );
false

The source code of IsNormal for PcpGroups is:

function ( H, U )
    local u, h;
    for h in GeneratorsOfPcp( Pcp( H, U ) ) do
        for u in Igs( U ) do
            if not u ^ h in U then
                return false;
            fi;
        od;
    od;
    return true;
end

I suspect that the usage of Pcp( H, U ) is to blame, and this should be Pcp( H ) instead?

[stertooy]

Oops, apparently this was already fixed, but I was working on 2.16 again in my fresh install of GAP 4.13.1. Apologies.