Patch Notes
Magma V2.15-8 Patch Notes
Patch release for Magma V2.15-8. Updated areas: Coding Theory, Elliptic Curves, Function Fields, Galois Groups, Groups and 7 more.
Coding Theory¶
- A crash in
LDPCDecodehas been fixed. Reported by M. Grassl.
Elliptic Curves¶
TwoDescenthas been modified to also return maps. Similarly withTwoCover.DualIsogenyhas been rewritten. Previously it was correct only up to an automorphism; now the composite of the given and its dual is guaranteed to equal multiplication by m, where m is the degree of the isogeny.- Some errors in
IsogenyFromKernelhave been corrected. PushThroughIsogenynow works faster.- Improved the equality comparison of isogenies so that different presentations of the same isogeny are correctly identified as equal more often.
- Improved the speed of
MultiplicationByMMapfor larger values ofm, and fixed several bugs in the characteristic 2 case. - A change to the amount of real precision in
QuarticReducewas made, which was a problem in some cases of using the descent machinery. - An bug has been fixed concerning reduction of plane cubics (occurring as genus one models of degree 3, or as curves returned by
ThreeDescent).
Function Fields¶
- Some incorrect answers from
Expandand the application of completion maps have been fixed. Reported by G. Moehlmann.
Galois Groups¶
- A crash in
SolveByRadicalshas been fixed. Reported by A. Previtali.
Groups¶
- A crash when comparing two group homomorphisms for equality has been corrected. Bug reported by S. Guest.
- A bug when computing conjugacy classes of a matrix group has been corrected. Bug reported by S. Guest.
- A crash computing
NormalLatticeof a matrix group has been fixed. Bug reported by H. Coutts.
Hilbert Modular Forms¶
- A serious oversight in the initial choices has been corrected. This would only have affected Hecke operators at primes dividing the level. (The choice of right ideal classes for a definite quaternion order was not made carefully enough.)
- The precomputation phase is now massively faster for number fields of degree $\ge 8$, due to greater care in lattice searches.
Hyperelliptic Curves¶
- The intrinsics taking a sequence of 3 G2 invariants now accept sequences of integers which are converted to rationals. The geometric autorphism groups returned for genus 2 curves are now permutation groups (
GrpPerm) rather than finitely-presented groups (GrpFP).
Invariant Theory¶
- A bug in
FundamentalInvariantsfor invariant rings of reductive groups has been fixed. Reported by M. Vogt.
Lattices¶
- A long-standing bug in the 2-adic genus of a lattice was fixed; e.g.,
pAdicDiagonalization(LatticeWithGram(DiagonalMatrix([4,6])),2)gave the wrong result. Reported by N. Scheithauer. - A crash in
ShortestVectorshas been fixed. Reported by R. Marinier. - A crash when taking an
ActionGeneratorof a G-Lattice has been fixed. Bug reported by M. Watkins.
Polynomial Rings¶
- A crash when creating polynomial rings over fields of fractions has been fixed. Reported by K. Binder.
- A crash when computing the discriminant of the zero multivariate polynomial has been fixed. Reported by A. Brumer.
- An incorrect result for squarefree factorization of bivariate polynomials in small characteristic has been fixed. Reported by P. Lisonek.
Real Numbers¶
- A convergence problem when using
Rootin fixed low-precision real fields has been fixed. Reported by R. Cossett.
Sparse Matrices¶
- A bug in
ModularSolutionwould result in a crash whenever Lanczos was selected without also selecting verbose output. Reported by C. Diem.