A bug in FunctionField, which was creating invalid algebraic function fields defined by reducible polynomials for some curves in positive characteristic, has been fixed.
A problem computing residue fields in fields whose constant field is a univariate rational function field at prime polynomials over that rational function field which were inseparable has been fixed.
PseudoMatrix now again accepts as input a sequence of ideals of an order and a matrix over an order. Reported by D. Yasaki.
Ideals of orders of associative algebras constructed using ideal<|> now have a basis in hermite normal form.
On repeated calls, ClassGroup did not always use the bound specified by the user via SetClassGroupBounds.
The ClassGroup of a number field defined by a non-monic polynomial sometimes crashed; this has been fixed. Reported by A.-S. Elsenhans.
For quadratic fields and quadratic forms, ClassGroup and ClassNumber did not terminate in very rare cases, e.g., for discriminant 202556. Reported by M. Wanless.
A bug in StandardForm applied to Z4 codes has been fixed. This bug would cause the result to not always have the required diagonal property. Reported by M. Villanueva.
Bugs in Radical and PrimaryDecomposition for positive dimensional ideals over fields of small characteristic have been fixed. Reported by E. Rains.
A fix has been made to CoveringStructure and hence automatic coercion so that if the result has the same order as one of the inputs it will that input rather possibly a non-simple representation of it.
A fix has been made to coercion from cyclotomic fields having simple representation to those having non-simple representation.
A crash when computing IsConjugate for two unordered partitions under the action of a permutation group, has been fixed.
A crash when working with double cosets of a permutation group has been fixed.
A trivial bug (resulting in a runtime error) in TwoCoverDescent was fixed. Reported by E. Gonzalez-Jimenez.
A precision problem with CanonicalHeight (and HeightPairing) that caused crashes has been fixed (via try/catch and increasing the precision). Reported by J. S. Müller.
A problem in the LLL function with the DeepInsertions parameter has been fixed.
Some issues with p-adic ring and field creation have been fixed; those could have caused two equal p-adic structures to be erroneously considered different, causing coercion problems.
A fix has been made to Roots for a polynomial when the ring or field as the second argument has precision 1.
Asking for the defining polynomial of a local ring with respect to itself no longer crashes.
ColumnSubmatrix can now handle [ .. ] sequences as input.
A crash in Determinant for sparse matrices has been fixed.
DegeneracyMap and DegeneracyMatrix now require the two arguments to be spaces of the same sign. (Previously, repeated calls to these could sometimes return the answer for the wrong sign.)
The maps returned by MultiQuotientMaps are now equipped with inverses.
Some minor improvements have been made to the usability of IsTwist and IsMinimalTwist.
A crash in DistinctDegreeFactorization when the Degree parameter was specified has been fixed.
A bug with PointSearch was been fixed. For affine schemes, points whose first coordinate was zero could be missed (this bug was introduced in the latest major Magma release). Note that PointSearch is not guaranteed to find points, but the points in question should have been found. Reported by E. González Jiménez.
Submatrix with indices is now supported. Omission reported by N. Dunfield.
Several of the minimisation routines for surfaces failed to accept input consisting of polynomials over Q with denominators. Reported by A.-S. Elsenhans.