Coercion of elements in an exact constant field of a function field into the function field has been fixed for relative extensions.

The composition of automorphisms of number fields has been fixed. Reported by J. Sijsling and W. Bley.

A bug has been fixed in computing the adjoints of surfaces Reported by M.J. Martins.

A major slowdown in FormallyResolveProjectiveHypersurface for some kinds of inputs has been fixed. Example inputs provided by M.J. Martins.

The upper bound returned by RankBounds or RankBound for the rank of the Jacobian of a hyperelliptic curve or a cyclic cover has been refined due to a new result of Brendan Creutz. The upper bound will now always agree with parity. Example: for the curve y^2 = 5*x^6 + x^5 + x^4 - 4*x^3 - 4*x^2 + 5*x - 1, the upper bound was previously 1 and is now 0.

The omission of square-free groups from the small groups database has been fixed. This created the bug of square-free groups of order > 2000 being mistaken for groups of order 3⁸.

A database of some automatic groups created from hyperbolic manifolds has been installed. Currently this contains 5389 such automatic groups. The database is contained in the file GrpAtc.tar.gz at the download page http://magma.maths.usyd.edu.au/magma/download/db/ .

A problem with crashes in lattices with standard basis and inner product has been remedied. This often showed up in number field computations.

A problem with verbose printing has been fixed.

A problem with BKZ in high dimension has been fixed. The problem was that in rare circumstances the internal kappa could change when a new short vector was found, causing a discrepancy between the approximate Gram matrix (which was reset before the kappa change) and the basis matrix.

Some superfluous printing in the sparse Hermite normal form algorithm has been fixed. Reported by E. Thome.

The function Submatrix(A, I, J) where I and J are sequences has been fixed in the case that there are zero rows or columns in the result. Reported by M. Villanueva.

A crash in testing whether a multivariate polynomial is square has been fixed. The algorithm to test squareness has also been sped up at the same time. Reported by A-S. Elsenhans.

The algorithm for GCD of polynomials over number fields has been sped up.

The algorithm for evaluation of a multivariate polynomial at a sequence of points in a larger ring has been sped up.

The Integral of a series over a ring is now returned as a series over the field of fractions of the ring.

Precision handling has been improved in Composition of series. Reported by G. Blanco.