The matrix representation of U5(2) of dimension 55 over GF(11) was incorrect and has now been fixed.
The intrinsics ATLASGroupNames and AtlasGroupnames now correctly indicate that their return type is SetIndx, not SeqEnum.
A minor regression: The first two matrix representations of E6(2) have been swapped. Similarly, the two permutation representations of O8-(3).2_3 have been swapped.
Error checking has been improved in Automorphisms of an algebraic function field over a given coefficient field. This intrinsic is currently only for function fields whose constant field is the rational field.
A precision problem with p-adic determinants has been fixed. Reported by C. Birkbeck.
The functions Representation and Submodules now work for G-modules of dimension zero. Problem reported by G. Tornaria.
Compatibility of modules over Dedekind domains has been improved. This, in particular, allows more flexibility in the combination of inputs to + and meet.
The OptimalSkewness intrinsic now checks that its input is defined over the integers.
A bug that could cause crashes when computing the roots of a polynomial over a Laurent series algebra has been fixed.
A crash in factorization of polynomials over a chain of function fields in characteristic zero has been fixed.
A crash in multivariate GCD has been fixed. Reported by R. Lewis.
The Kernel of a matrix over an order of a quadratic field which is also a Euclidean ring has been fixed.
The IsPower intrinsic for series has been improved to consider scaling by a leading coefficient that is not 1. Reported by M. Stoll.
A bug that could cause require directives to report the wrong error location has been fixed.