A bug that was causing runtime errors in Parametrization for curves that are already lines has been fixed.

Fixed some leaks in the quadratic residue code creation.

Fixed some leaks when re-creating an already-existing code.

Fixed a leak when testing whether a code is quasi-twisted cyclic.

Complex numbers with imaginary part equal to zero could erroneously be coerced into the integers. Due to the automatic coercion strategy used by Magma, this had many side effects, such as when multiplying an integral matrix by a complex scalar. Noted by A.-S. Elsenhaus.

A bug in the computation of the full module of a sheaf has been fixed. This was causing errors in some cases in many of the sheaf intrinsics: a particular case was with IsArithmeticallyCohenMacaulay that was reported by Eric Rains.

A bug in the cohomology code with a trivial generator has been fixed. Reported by K. Andersen.

Fixed a minor problem in PointsQI with local points not being tested for global solubility in all cases.

Fixed a problem (incorrectly re-using a pre-computed polynomial) in local solubility in 2-descent when points were close to infinity.

A logical problem in the code to exclude candidate groups during the computation of Galois groups has been fixed. Without this fix, certain computation would fail. Reported by J. Klueners.

Some memory leaks in MaximalOrder have been fixed.

It is now possible to compute the maximal order, using the special code for Artin-Schreier extensions, of an Artin-Schreier extension which is an extension of an extension of a rational function field.

Orders of function fields are now recognized as being domains and some computation such as matrix determinants over these orders will gain as a result.

Enforce upper bound of 2¹⁶ on prime argument to pQuotient. Bug reported by N. Barker.

A memory leak when both evaluating homomorphisms and using the Evaluate function has been fixed.

A crash that sometimes occurred when invalid relations were given to a PC-group constructor has been fixed. Bug reported by D. Howden.

A bug when testing membership of a trivial automorphism group has been fixed. Bug reported by K. Andersen.

A runtime error connected to homomorphisms with domain a matrix group has been corrected. Bug reported by K. Andersen.

An error in the maximal subgroups of Alt(17) returned by Magma has been noticed by John Bray and corrected by Colva Roney-Dougal. The affine group 17:8 is not maximal in Alt(17) (it is contained in PGammaL(2,16)) and the function has been fixed to not return this group. Note that the group 17:16 is maximal in Sym(17) (a novelty maximal).

Bugs in constructive recognition of linear and symplectic groups have been fixed by Derek Holt.

Fixed quo constructor for finitely presented Lie algebras. Bug reported by Jason S. Williford.

Improved error messages for several other constructors for finitely presented Lie algebras.

A bug in IsDivisibleBy for local ring elements has been fixed. Reported by M. Stoll.

The Log intrinsic has been extended to arbitrary units rather than 1-units only.

A problem with SplittingField over p-adic rings for non-monic polynomials has been fixed. Reported by T. Dokchitser.

An error in the computation of AbsoluteTotallyRamifiedExtension for a general p-adic field has been fixed. Reported and fix provided by T. Dokchitser.

The mapping of a general local field to its ramified representation has been fixed to avoid unnecessary precision loss.

Fixed a memory issue when using ideal<> of an order of a number field.

A bug in Automorphisms of non-normal numberfields has been fixed. Previously an internal error could happen in such cases.

A bug in Compositum if one of the fields is non-monic has been fixed.

A bug in DistinctDegreeFactorization over GF(2) has been fixed. Reported by T. Dokchitser.

A crash in Discriminant over small finite fields of characteristic 3 has been fixed.

Fixed an incorrect memory allocation size that could lead to occasional crashes.

Fixed a problem where CrossCorrelation with sequences of non-identical length would give an internal error, rather than a user error.

Evaluation of a straight-line polynomial at elements not in the coefficient ring has been fixed.