Bumped version to 2.5
Added sparse grid option for not reaching domain boundary.
Added option to not store old samples in marginal approximants, using standard…
Removed unused imports.
Added conversion to list of vector-like approxParameters to fix comparison…
Streamlined setup of trainedModel. Avoided repeated computation of old projMat…
Reweighted residue angle in distance table.
Fixed marginal coefficient support for collapsed models.
Enforced marginal coefficient support through sparse matrices. Reduced Psupp to…
Added missing index in virtual padding.
Fixed residue transposition.
Introduced virtual padding to improve memory usage.
Reduced space wasted by coalescing samples in marginal sampling engines.
Collapsed marginal interpolator in matched pivoted trained model. Now corrected…
Added assignment of musMarginal on model load. Turned musMarginal into…
Fixed evaluation of hat function for non-uniform grids. Renamed lobatto to…
Fixed sample batch bug in rational greedy.
Added check for recompression in trainedModels.
Fixed wrong coeff size in interpolators. Added length check in interpolator…
Fixed wrong depth selection bug in piecewise linear marginalization. Improved…
Added piecewise linear interpolant in marginal approximants.
Improved naming convention of piecewise linear interpolator. Fixed reordering…
Changed name of marginal sampler in pivoted greedy.
Updated tests and examples after change in pivoted approximant parameter fields.
Removed specialized nearest neighbor pivoted approximants. Compressed…
Allowed assignment of approxParameters in fragile mode. Setting up removal of…
Added piecewise linear interpolation engine based on sparse grids.
Allowed computation of scale factors in fragile mode. Several minor bugs fixed.
Debugged corrector iterations. Fixed scaling issue with radial weight.
Fixed inconsistent definition of nmus of trainedModel.
Removed now useless pivoted POD_global.
Added approximant postprocessing by recompression.
Updated installer info and other meta files.
Simplified pivoted greedy error estimator. Fixed broken test.