The intrinsic AlgebraGenerators (and code that calls it) will no longer reset the random number generator.
QuaternionAlgebra(I, S) entered an infinite loop in a special situation.
A bug which caused a runtime error has been fixed. For example, this occurred at level 16+6*sqrt-3. Reported by John Cremona.
A bug has been fixed in the IsLocallyFree intrinsic for sheaves that was causing a runtime error in the case where the sheaf argument was already known to be arithmetically Cohen-Macualay. Reported by C. Quitte.
A crash in FreeResolution when the input could be written over a proper subfield has been fixed.
Fixed a crash in IsIsomorphic and IsEquivalent which could occur for higher-dimensional polytopes with unusual facet configurations.
A crash in the CosetGeometry function caused by incorrect input has been fixed. The input now causes a runtime error. Crash found by an unknown user of the online calculator.
A bug in Minimize for elements of the maximal order of cyclotimic fields has been fixed. The element would wrongly reported to be a field element rather than a ring element, thus causing internal errors.
Fixed a bug that could cause certain torsion points to erroneously be reported as having infinite order when the torsion group was already known. Reported by J. Cremona.
Fixed a bug with Reduction of a GenusOneModel in the Windows version. A work-around for avoiding indefinite LLL was no longer needed, and could occasionally run into precision problems.
An unnecessary bottleneck has been avoided in the routine EllipticCurve(C, pt) for a genus 1 curve C.
A runtime error in RootNumber (due to insufficient p-adic precision) has been fixed.
The use of the Check parameter has been made possible from ext< FldFunRat | >.
A bug in AssignWeight has been fixed that could cause weights for other edges to be reset.
A crash in StandardGenerators for Aut(F4(2):2) and Aut(G2(4):2) has been fixed.
The intrinsic CorrectForm (and code that calls it) will no longer reset the random number generator. Reported by E. O'Brien.
A crash when creating orthogonal (plus type) groups of degree 2 has been fixed. Reported by E. O'Brien.
Code for UnitaryForm has been improved by D. Holt.
The performance of the function UnipotentStabiliser for matrix groups has been improved.
The transitive group database access and identification have been significantly improved, in terms of both speed and memory usage.
A new database type for the transitive group databases has been added. This provides faster access for multiple groups than the old process approach, the use of which is now recommended against.
The transitive group database has been extended slightly to include degree 31.
A bug in matrix groups IsConjugate has been fixed. The bug showed up when asked to test conjugacy of two elements with respect to a subgroup not containing the elements. It was possible for the two elements to be conjugate, but the answer false was returned. Bug reported by Paul Taylor.
The performance of MaximalSubgroups has been greatly improved.
A bug where the NormalLattice function could incorrectly compute inclusions has been fixed. Reported by A. Steel.
A problem due to not using a known base when computing ChiefFactors of a matrix group has been fixed. Reported by D. Holt.
A bug when enumerating elements of a matrix group of order ≥ 230 has been fixed. The bug also afflicted the numbering maps of such groups. Bug reported by Yves Edel.
A bug in the IntersectionWithNormalSubgroup function for permutations groups has been fixed. Bug reported by A. Steel.
A bug in ClassicalMaximals caused by recent standardisations to ChevalleyGroup has been fixed. Bug fix by D. Holt.
A bug in CharacterTable which allowed the routine to finish without finding the full table, has been fixed. Reported by A. Steel.
The optional third argument of Intseq (the minimum length of the result) is now allowed to be zero.
Fixed a bug in HighestWeightRepresentation for Lie algebras. Reported by A. Previtali.
Fixed a bug in Generators for groups of Lie type that would cause an incomplete set of generators to be returned if the group was simply connected and not semisimple.
Fixed a number of bugs in StandardRepresentation for groups of Lie type.
Fixed a number of bugs in RowReductionHomomorphism and Inverse for representations of groups of Lie type. These arose in particular in the non-semisimple case. Bug reported by D. Loeffler.
An bug in inverse of matrices over rings with zero divisors has been fixed.
HeckePolynomial gave incorrect answers for non-cuspidal spaces (due to incorrect bounds on eigenvalues). Reported by Alan Lauder.
Null sequences are now properly checked for and produce an error message instead of crashing.
ReducedForm now returns the correct transformation when a=b or a=c. Reported by Justin Walker.
Representation is now allowed for G-modules over rings which are not domains. Reported by K. Andersen.
A bug in the application of a hom from a function field of a scheme has been fixed. Reported by M. Bright.
LinearSystem with multiplicity now gives an appropriate error, instead of garbage, in cases where it is not implemented.
One may now do S*A for sparse matrix S and dense matrix A.
A bug has been fixed where import would sometimes incorrectly report that a package file had not been attached due to paths being equivalent but not identical.
A bug in equality testing of intrinsics that could cause it to erroneously return true for unequal intrinsics has been fixed.