Bugs in Automorphism and AutomorphismGroup have been fixed. Reported by M. Grassl.
Large exponents in divisors have been disallowed.
A bug in ZetaFunction, Order and other point counting functions on hyperelliptic curves and their Jacobians has been fixed.
A bug has been fixed in the functions CommonEigenspaces and Diagonalisation for diagonalising a commutative matrix algebra over an extension field.
An missing error check in AdditiveCode has been inserted. Problem reported by M. Grassl.
A bug in the returned parent of a parity check matrix of an additive code (incorrect number of rows) has been fixed. Reported by M. Grassl.
A bug in Order([FldCycElt]) that resulted in internal errors has been fixed. Reported by R. Bruner.
A problem with embedding one finite field in another (where the generators were not primitive) has been fixed.
Improvements and bug fixes to Expand of function field elements at a place (including fixes to the preimage of the completion map of a function field order). The bug fix also fixes some problems in Automorphism and AutomorphismGroup reported by M. Grassl.
A bug in IsInvertible for scheme maps, which caused errors for weighted projective domains/codomains, has been fixed. Reported by Tobias Beck.
A boolean Vararg FullCheck, with default false, has been added to Is(Non)singular and SingularSubscheme. If it is false, then the scheme is assumed to be equidimensional (i.e., all irreducible components of the same dimension). If FullCheck is true, then the scheme is decomposed into its equidimensional constituents, which can be potentially very expensive, for the singularity computations. Note that if the scheme isn't equidimensional and FullCheck is false then the results of the functions are likely to be wrong. This has caused some errors in the past along with some "knock on" errors as, for example, internally, the labelling of a scheme as non-singular causes it to be labelled as reduced also.
Loop index bug fixed in HighMap and HighProduct in the A-infinity computation code.
A crash in Trace for very large matrices has been fixed. Reported by M. Grassl.
A bug in SubOrder for equation orders has been fixed. Without this fix, any call to the function with an equation order would trigger an internal error. Reported by R. Bruner.
A bug in UnitEquation has been fixed. The function wrongly allowed relative extensions to be passed in - which then triggered internal errors.
A bug in ext<Q| ... in the case when the argument is a polynomial which cannot be coerced into a polynomial over Q has been fixed. This would result in internal errors when unfixed.
A bug in the function SplittingField([
A bug in polynomial multiplication over non-commutative rings has been fixed. Reported by Stephen Tawn.
The implemetation of Shank's algorithm for the computation of class groups of quadratic forms of moderate discriminant has been improved and should now be more correct. Fixes a bug reported by Jeremy Rouse.
A bug in the computation of automorphisms has been fixed. In particular when the local ring of field was not in "standard form" (i.e., a totally ramified extension over an unramified extension over the prime field), the automorphisms computed were frequently wrong. Reported by Ruben Debeerst.
A bug in testing for squarefreeness of polynomials over p-adic rings resulted in the inability of Magma to compute roots of non-squarefree polynomials. Reported by Ruben Debeerst.
A bug and inefficiency in the isomorphism testing for p-adic rings was fixed. In particular when at least one field was not in "standard form" (i.e., a totally ramified extension over an unramified extension over the prime field), the test would produce wrong results, or fail with runtime errors. Reported by Ruben Debeerst.