FElupe: Finite Element Analysis

[8.7.0] - 2024-06-07

[!NOTE]
This is the first release with support for NumPy 2.0.

Added

• Add math.solve_nd(A, b, n=1) as a generalized function of math.solve_2d(A, b) with supported broadcasting on the elementwise-operating trailing axes.

Changed

• Rebase math.solve_2d(A, b) on math.solve_nd(A, b, n=2) with a batched-rhs supported in NumPy 2.0.
• Change ogden_roxburgh() and OgdenRoxburgh to use the Gauss error function erf instead of tanh as internal sigmoid function.
• Flatten the returned inverse indices-array of np.unique(..., return_inverse=True) in mesh.merge_duplicate_points() to ensure compatibility with NumPy 2.0.

Fixes

• Reset state variables in PlotMaterial.evaluate() after each completed load case.

What's Changed

• Reset state variables in PlotMaterial.evaluate() after each completed load case by @adtzlr in https://github.com/adtzlr/felupe/pull/784
• Add math.solve_nd(A, b, n=1) by @adtzlr in https://github.com/adtzlr/felupe/pull/785
• Support broadcasting in matrix-axes of math.solve_nd() by @adtzlr in https://github.com/adtzlr/felupe/pull/787
• Fix small typo focusses -> focuses by @tkoyama010 in https://github.com/adtzlr/felupe/pull/786
• Change ogden_roxburgh() and OgdenRoxburgh to use erf instead of tanh by @adtzlr in https://github.com/adtzlr/felupe/pull/788
• Add support for NumPy 2.0 by @adtzlr in https://github.com/adtzlr/felupe/pull/789

Full Changelog: https://github.com/adtzlr/felupe/compare/v8.6.0...v8.7.0