pnl-1.15.0

Changed

• Remove the full tensor representation of PnlBasis and only use the sparse tensor (#9).

Fixed

• Fix unpacking of sparse matrices.

pnl-1.14.0

Added

• Apply optional non-linear pre-treatment to the data before renormalisation in PnlBasis (#8).

Changed

• Remove internal C translation of Blas & Lapack.
• Improve basis memory management (#7).

Fixed

• Fix packing/unpacking of PnlBasis objects.

pnl-1.13.1

Fixed

• pnl_mat_fprint was printing to stdout instead of the file descriptor

pnl-1.13.0

Added

• Add missing documentation on local bases.
• Add tests for basis tensor constructors.

Changed

• For the sake of consistency, the constructors pnl_basis_create_local and pnl_basis_create_local_regular are renamed to pnl_basis_local_create and pnl_basis_local_create_regular.
• The function pnl_basis_i is not marked inline anymore.
• In functions to compute a basis derivative, first check that Df is not NULL.
• Prevent standard PnlBasis constructors to be used to create a local basis (type PNL_BASIS_LOCAL).
• Refactor PnlBasis tensor constructors to drop the recursive and copy approach.

pnl-1.12.1

Added

• Add local bases (#5)

pnl-1.11.0

Changed

• The macros RETRIEVE and CREATE are prefix with PNL_RAND. This is a breaking change.
• Add PNL_ prefix to FALSE, TRUE, OK and FAIL. This is a breaking change.

Added

• Add pnl_mat_print_csv to print a matrix in a csv format.

pnl-1.10.4

Fixed

• Add missing libwinpthread-1.dll to post_build event

pnl-1.10.3

Fixed

• Add missing libwinpthread-1.dll to Windows binaries

pnl-1.10.2

Fixed

• Fix libgfortran version for Windows binary release
• Fix README typos

pnl-1.10.1

Fixed

• Fix strcasecmp not available on MSVC
• Fix PNL_RANGE_CHECK_OFF in Release mode

Added

• Add new functions for sparse matrices
• Add pnl_vect_set_subblock
• Add FFT api with work spaces