An underlying change to how IsIrreducible works when returning false caused various examples to break. The code here was changed to use CompositionFactors (of a G-module) instead.
A bug involving taking chi_p instead of 1/chi_p was fixed.
A fix has been made to Expand and the application of Completion maps.
The construction of sets of ideals of orders of associative algebras has been fixed. Reported by M. Kirschmer.
The testing of equality of homomorphisms between algebras which do not inherit from AlgGen has been fixed.
A bug with IsIsomorphicSmallPeriodMatrices was fixed.
A crash in ReedMullerCodeQRMZ4 has been fixed. Reported by M. Kiermaier.
A bug in the code computing the maximal module of a sheaf has been fixed. This was causing errors in some cases in FullModule and the intrinsics depending on this, like IsLocallyFree. Reported by Claude Quitte as an IsLocallyFree error applied to the tangent sheaf of a cubic surface.
An occasional incorrectness in the equality test for modules over euclidean rings has been fixed. Reported by C. Quitte.
Removed stray timing printings from some Subfield calculations.
A new version of the matrix group CompositionTree code has been installed. Code supplied by E. O'Brien.
A crash in SolvableQuotient has been fixed. Bug reported by S. Humphreys.
A bug in the creation of subgroups of pc-groups has been fixed. Bug reported by J. Klüners.
A crash when GenericAbelianGroup was given a non-structure argument now results in a runtime error.
TwoSelmerGroup for Jacobians of hyperelliptic curves of even degree other than 6 over the rationals had been disallowed, by mistake. In fact, TwoDescent is implemented for any degree (at least 4) over the rationals and over number fields, with certain provisos.
A bug in TwoCoverDescent (2-descent on curves) has been fixed; the precision in some local calculations is now increased as necessary.
A bug with computing the automorphism group of a lattice when the Gram matrix has entries over 2^30 has been fixed. Bug reported by G. Nebe.
A crash when computing the automorphism group of a lattice using the Vectors parameter to give the set of vectors to use has been fixed.
A crash in ShortVectorsProcess for lattices over the reals with large entries has been fixed. Reported by I. Wagner.
Added an optional argument Virtual to the Branch function, to allow computation of virtual decompositions of representations.
Fixed a bug in SimplyConnectedVersion (and AdjointVersion) that caused those functions to return incorrect results in case a non-semisimple root datum was given as input. This could subsequently cause a crash in other procedures, as reported by B. van Geemen.
A bug was fixed that could cause Magma to interface incorrectly with the linear programming package, which could result in incorrect solutions being returned. In particular, this affected integral problems solved by MaximalSolution, MinimalSolution, MaximalZeroOneSolution, and MinimalZeroOneSolution.
Lists may now be tested for equality.
Fix a problem in Automorphisms for non-simple fields. Reported by V. Busch.
A hang in multivariate factorization over finite fields has been fixed. Reported by A.-S. Elsenhans.
A fix has been made to the preimage computation of class group maps.
A crash in representations over quadratic orders has been fixed. Reported by A. Bartel.
SimplicialComplex now also allows a set of sets as its argument; this fixes a bug in AddSimplex. Reported by F. Lunnon.