A bug in computing the complex roots of a polynomial has been fixed. Reported by S. Humphries.
Some small fixes have been made for GL(n, Z) conjugacy testing.
The production of fp-groups with trivial relations by the Simplify functions has been fixed. Bug reported by D. Holt.
The free group subgroup code has been extended to recognise some free subgroups of non-free groups. In response to an example from D. Holt.
Arithmetic on Jacobians over inexact rings is now disallowed, as the underlying algorithm (Cantor's) is not suitable for this, leading to crashes and superfluous results. Noted by M. Stoll.
The TorsionBound for Jacobians of hyperelliptic curves over the rationals now never considers the prime 2 to be good. This caused problems with Order in some instances. Reported by A. Sutherland.
A memory leak with searching for points on Jacobians was fixed. Reported by M. Stoll.
Added a simpler version of Seek that assumes the desired offset is relative to the start of the channel.
Fixed a bug that caused WaitForIO to disallow server sockets.
Fixed a bug that prevented version exchange from working.
Fixed a bug that would prevent further writes to a channel after attempting to write an object of a type not yet supported.
Conductors of orders of quadratic fields > 230 are now handled better.
A crash in RiemannSurface involving root finding has been fixed. Reported by R. van Bommel.
JacobianMatrix and JacobianIdeal now use a minimal basis of the defining ideal of a scheme if this is smaller than the set of original defining polynomials. This change also affects IsNonsingular and JacobianSubrankScheme. Change suggested by M. Reid.
The use of shorthands such as [RngIntElt] instead of SeqEnum[RngIntElt] in intrinsic arguments now works properly with reference arguments. Reported by M. Joaquin.
The term "multi-iteration" has been changed to "dual iteration" in error messages.
Some problems with twisted polynomials have been fixed. Reported by A. Joux.