Fixed a bug where # applied to an associative array would return the largest that the array had been, rather than the current size. Reported by C. Doris.
Error checking when attempting to compute a function field of a reducible curve or a curve with no non-empty affine patches has been improved resulting in error messages rather than crashes.
Dirichlet and Hecke character evaluations that are 1 or -1 are now returned in the integers.
The RootNumbers intrinsic for a Hecke or Grossencharacter now has a AA vararg that if set returns the result as an associative array.
Equality check for Grossencharacters over totally real fields has been fixed.
A bug with RootNumber at a finite place for a Grossencharacter over a totally real field was fixed.
A crash when computing with homomorphisms having a matrix group domain has been fixed. Crash reported by D. Holt.
Due to technical issues, the database of primitive groups on Magma's website was incomplete (missing many groups of degree >= 2500) from 9 Sept 2016 to 23 Jan 2017. All primitive groups with degree < 2500 were present. Only sites that have this database from the website during this period will be affected. The command NumberOfPrimitiveGroups(4095); should return 6. If it returns another value then the database is incorrect, and this may be corrected by acquiring the current databases from the Magma website. Bug reported by D. Holt.
A number of crashes in permutation groups to do with a RandomSchreier followed by Verify, with intervening operations, have been fixed.
Fixed bugs in ClassicalSylow. Bugs reported by E. O'Brien.
A problem that rendered the last 696 of the 1396077 groups of order 3^8 inaccessible in the database has been fixed. Bug reported by Mark Lewis.
A bug in Subgroups, where some (soluble) subgroups were missed, has been fixed. Bug reported by William Cocke.
A bug with finitely presented groups and computing subgroup relationships has been fixed.
In NewformsOfDegree1 a trivial mistake caused a runtime error. Reported by J. Cremona.
TwoSelmerGroup and TwoSelmerGroupData returned an incorrect map from the Selmer group to the etale algebra, in the case of an odd degree curve. (The other items returned by these functions were not affected by the error.) Reported by M. Stoll.
The Bezoutian of a hypergeometric motive datum is now available.
The guesses for wild Euler factors have been arranged to avoid errors.
An error with tensor product L-functions has been fixed.
The L-series of a Hilbert modular form has been corrected in some cases, though some errors with (narrow) class group distinctions still exist.
The PHV (PrintHodgeVector) vararg has been added to a number of HodgeStructure intrinsics.
The Hash of a lattice is no longer computed from the user basis, as there is no canonical form that is easy to compute. This fixes bugs with sets of lattices and equality-checking. Reported by M. Kirschmer.
The eq operator for lattices of different ranks now returns false, rather than an incompatibility error.
The WittInvariants and HasseMinkowskiInvariants now have an AA vararg that when set returns the result as an associative array.
An inconsistency in the embedding of BCH codes of the same length has been fixed. Reported by M. Grassl.
The printing of elements of an extension of a p-adic field of precision 1 now uses the name of the generator, if set, instead of $.1. Reported by C. Doris.
Fixed a bug where the profiler might crash or return nonsense execution counts for profiled routines that were executed more than 230 times.