Patch Notes

Magma V2.14-14 Patch Notes

Patch release for Magma V2.14-14. Updated areas: Abelian Groups, Coding Theory, Commutative Algebra, Cyclotomic Fields, Elliptic Curves and 11 more.

11 June 2008 16 areas

Abelian Groups ¶

A few improvements to GenericAbelianGroup have been made, so that it should work for more universes. Also, coercion into the generic Abelian group has been improved.

A crash in free resolution computation for modules defined over non-fields has been fixed.
A memory leak in Gröbner basis computations with field polynomials has been fixed.

A crash in inversion in cyclotomic fields has been fixed. Reported by M. Grassl.

A bug in the elliptic curve rank computations that could lead to random crashes on "large" curves has been fixed.

A crash in finite field discrete log computation has been fixed. Reported by R.-P. Weinmann.

A bug in computation of representation matrices has been fixed.
Product representation of elements have been improved and fixed. Norms of elements in product representation are now returned in product representation when they lie in a function field. Mapping of elements in product representation to the residue class field is now faster.
Converting elements to rational functions has been fixed.

Some specialized optimized invariants were wrongly disabled, resulting in too long running times for certain examples.
An unexpected loss of precision during the initial stages of the Galois group computation has been fixed. This would result in runtime errors in certain cases.
Internally, a bound used to estimate the precision neccessary to reconstruct elements in GaloisSubfieldTower was too low. In certain trivial examples this would result in wrong anwsers.
The computation of Galois groups of polynomials over the integers and rationals has been adopted to allow handling of arbitrary polynomials, ie. the polynomials are no longer assumed to be square-free.

A bug in OutEdges for multi-digraphs that would cause Magma to issue runtime errors has been fixed.

The algorithm for the computation of gcds and lcms of elements of local rings with variable precision was broken and resulted in internal errors (eg. computing the contents of a polynomial defined over such a ring). Bug reported by N. Bruin.

A bug in the computation of the field fixed by a a sequence of maps has been fixed. Internally, a wrong field basis was choses which could produce wrong results.
The algorithm for adding orders that have non-coprime indices has been changed. This removed a very bad inefficiency in the old code. Problem reported by N. Bruin.
The intrinsic Reconstruct when tasked with reconstructing elements with denominators would wrongly type the elements as order elements. This bug would for example result in Denominator to wrongly report a denominator of 1.

Some bugs in trivial subgroups of PC-groups have been fixed. Bugs reported by Derek Holt.
The algorithms used by the $p$-group functions Agemo and JenningsSeries have been changed so that these functions can now deal with much larger groups. Problem reported by Harris Nover. Thanks to Eamonn O'Brien for advice on algorithms.

TransitiveGroupIdentification for groups of degree 22 was completely broken, the data for degree 23 was wrongly used. This would generally result in runtime errors ("Group not found").
TransitiveGroupIdentification when asked to return the conjugating element (by setting Raw := false) as well would occaisonly return the inverse of the element instead.

New intrinsics (SetIloadAllowEsc and GetIloadAllowEsc) have been added to allow printing of escape sequences in iload files. Without those fixes escape characters would be translated into printable characters. Requested/reported by M. Kida.
A crash in Read for very large files has been fixed with a proper check. Reported by H. Nover.