scikit-SUNDAE v1.1.3

v1.1.3

This release supports Python 3.10-3.14. Binary files on PyPI bundle SUNDIALS 7.5 (from conda-forge) built with int32, float64, OpenBLAS, and SuperLU_MT.

Bug Fixes

• Address memory leak with raised exceptions caused by persisting solver/data objects (#50)

scikit-SUNDAE v1.1.2

v1.1.2

This release supports Python 3.10-3.14. Binary files on PyPI bundle SUNDIALS 7.5 (from conda-forge) built with int32, float64, OpenBLAS, and SuperLU_MT.

Bug Fixes

• Fixes issue where jacfn is ignored when using sparse linear solver (#47)

scikit-SUNDAE v1.1.1

v1.1.1

This release supports Python 3.10-3.14. Binary files on PyPI bundle SUNDIALS 7.5 (from conda-forge) built with int32, float64, OpenBLAS, and SuperLU_MT.

Bug Fixes

• Ensures exception propagations work correctly with numpy 2.4 releases (#43)

scikit-SUNDAE v1.0.4

v1.0.4

Patch release only for v1.0.x. Available on PyPI for Python 3.9--3.13 and on conda-forge for Python 3.10--3.13.

Bug Fixes

• Fix conditional checks for constraints to allow numpy arrays (#36)

scikit-SUNDAE v1.1.0

v1.1.0

This release supports Python 3.10-3.14. Binary files on PyPI bundle SUNDIALS 7.5 (from conda-forge) built with int32, float64, OpenBLAS, and SuperLU_MT.

New Features

• Enforce constraints when **LSSparseDQJac funcs are used (#32)
• Drop Python 3.9 and add support for 3.14 in tests/release (#31)
• Move to newest SUNDIALS v7.5 for CI builds/tests (#30)
• Add version warning banner to docs for dev and older releases (#28)
• Move to newest SUNDIALS v7.4 for CI builds/tests (#23)
• Move to newest SUNDIALS v7.3 for CI builds/tests (#16)
• Add reduce_bandwidth function to help restructure sparse problems (#15)
• Implement interfaces for Jacobian-vector products (#13)
• Allow preconditioning for iterative solvers (#12)
• Enable OpenBLAS-linked LAPACK linear solvers (#11)
• Add iterative linear solvers to IDA and CVODE (#10)
• Allow numpy types in options checks for both CVODE and IDA (#8)
• Expose linear solver option that uses SuperLU_MT (#6)
• New jacband module for support finding sparsity/bandwidth (#6)
• Custom sparseDQJac routines available by supplying sparsity (#6)
• Changed signature inspections to support decorated jit functions (#3)

Optimizations

• Use np.testing where possible in tests for more informative fail statements (#14)
• Updates to be compliant with Cython deprecations of IF/ELIF/ELSE and DEF (#5)
• Use single-line memory views and pointer addressing for np2ptr and np2smat (#5)
• Use micromamba instead of miniconda in CI (#3)

Bug Fixes

• Move to cibuildwheel for releases due to segfaults on Linux (#33)
• Fixed import typo in docstring examples for RichResult (#29)
• Resolve exception propagation consistently for Cython v3.1 and up (#20)
• Fix memory leak when init_step is repeatedly called in CVODE and IDA (#19)
• Add sign_y terms and default to np.float64 for floating type in j_pattern (#7)

Breaking Changes

None.

scikit-SUNDAE v1.0.2

Optimizations

• Use micromamba over conda in CI workflow (#22)
• Remove IF and DEF Cython code for future-proofing builds (#22)

Bug Fixes

• Change Exception propagations to support Cython v3.1 (#21)
• Improve type support by using Integral and Real (#21)

scikit-SUNDAE v1.0.1

Bug Fixes

• Fix memory leak in CVODE and IDA (#18)

scikit-SUNDAE v1.0.0

This is the first official release of scikit-SUNDAE. Main features/capabilities are listed below.

Features

• Python errors can be propagated through Cython wrappers
• Implicit differential algebraic (IDA) solver for differential algebraic equations (DAEs)
• C-based variable-coeffecients ordinary differential equations (CVODE) solver
• Events functions with scipy-like API, including "terminal" and "direction" options
• Dense and banded linear solver options in both IDA and CVODE
• Option for user-supplied Jacobian function in both IDA and CVODE
• scipy-like RichResult output containers

Notes

• Implemented pytest with tests that directly compare against solutions generated using C programs
• Source/binary distributions available on PyPI
• Documentation available on Read the Docs