--- title: "Magma V2.17-4 Patch Notes" canonical: "https://docs.magma-maths.org/patchnotes/2.17-4.html" source: "https://docs.magma-maths.org/patchnotes/2.17-4.md" --- # Magma V2.17-4 Patch Notes ## Algebraic Number Fields - A fix has been made to `Subfields` of algebraic number fields in relative representation. - A fix has been made to memory management for class group computations. - A fix has been made to `GaloisGroup` for reducible polynomials over algebraic number fields. - The prime chosen for the creation of lattices of subfields has been changed. The old prime could produce splitting fields too large to be useful. Reported by J. Klüners. - A bug in `PellEquation` has been fixed. `NormEquation` was called with the wrong parameters. Reported and fixed by N. Bruin. - The changed signature for [`GetAssertions`](../MagmaLanguage/EnvironmentAndOptions/Set-Get.md#function-getassertions) caused a runtime error in certain computations of Galois groups. ## Algebras - Argument checking in `BasisProduct` for structure constant algebras has been improved. - A crash in `MaximalIdeals` for group algebras has been fixed. Reported by S. Humphries. ## Commutative Algebra - An incorrect result for `ColonIdeal` for ideals with gradings has been fixed. Reported by E. Rains. ## Convex Polytopes - The ID numbers used by `PolytopeReflexiveFanoDim3` now coincide with the PALP database, with the exception that PALP indexes from 0..4318, whilst Magma indexes from 1..4319. ## Databases of Groups - A crash (due to incomplete porting of the small groups database to the new garbage collection scheme) has been fixed. Reported by M. Newman. ## Function Fields - A problem in the `ext<...>` constructor for function fields has been fixed: it would cause an internal error on zero (or constant) polynomials. ## Graphs - A crash has been fixed in the [`MultiGraph`](../Combinatorics/Multigraphs/multig-constr.md#constructor-constructor-multigraph-rngintelt-list-grphmultund-grphvertset-grphedgeset) constructor when using an indexed set to describe the edges. Reported by C. Quitte. ## Group Representations - A bug in the `IrreducibleModules` function has been fixed. Reported by U. Thiel. ## Groups - A bug in the implementation of the `IsPerfect` filter for the `Subgroups` intrinsic, which allowed the return of subgroups that were not perfect, has been fixed. Reported by D. Holt. - A crash to do with degree 2 orthogonal groups over finite fields has been fixed. - A crash in the permutation group Jellyfish functions has been fixed. Reported by E. O’Brien. ## L-Series - Fixed the definition of the LSeries of a Hilbert modular form at bad primes - Fixed a series-precision problem in LTaylor ## Lie Theory - A bug in the map returned by the `quo<...>` constructor for Matrix Lie algebras has been fixed: it would return objects of incorrect type. Reported by S. Murray. - The performance of [`BruhatLessOrEqual`](../LieTheory/CoxeterGroups/group-op.md#function-bruhatlessorequal), [`BruhatDescendants`](../LieTheory/CoxeterGroups/group-op.md#function-bruhatdescendants), [`KLPolynomial`](../LieTheory/RepresentationsOfLieGroupsAndAlgebras/representation-other.md#function-klpolynomialxy) and [`RPolynomial`](../LieTheory/RepresentationsOfLieGroupsAndAlgebras/representation-other.md#function-rpolynomialxy) has been greatly improved. Moreover, these methods now apply to elements of Coxeter groups presented as permutation groups, fp groups, or matrix groups. ## Module Theory - A bug in [`Submodules`](../RepresentationTheory/ModulesOverAnAlgebraAndGroupRepresentations/submodule-lattice.md#function-repthy-submodules) with codimension limit (coming from a call to [`LowIndexNormalSubgroups`](../FinitelyPresentedGroups/FPGroups/subgroup-fin-index.md#function-lowindexnormalsubgroups-grpfp-rngintelt)) has been fixed. Reported by T. Kliem. ## Polynomial Rings - Creating `HenselProcess` objects no longer causes crashes. ## Series Rings - `Reverse` of a series no longer loses a digit of precision. Reported by A. Previtali. - In fixed-precision series rings, evaluating a full-precision series at a term with valuation zero no longer reduces the result to have no precision. Reported by A. Lauder.