Andrew Knyazev, Professor Emeritus

ENH: added the spectral 2-norm to _norm.py

2

Added the spectral norm #### Reference issue Closes gh-7268 #### What does this implement/fix? Rejuvenates https://github.com/scipy/scipy/pull/8472

BUG: Update _svds.py: orthogonalize eigenvectors from arpack to eliminate edge case failures with clustered eigenvalues and update docstring examples

8

#### Reference issue Closes gh-11406, gh-16741. #### What does this implement/fix? As proposed in https://github.com/scipy/scipy/pull/16693#issuecomment-1193369299, copying line 297 eigvec, _ = np.linalg.qr(eigvec) to before line 337, so that the postprocessing...

ENH: improved accuracy and orthonormality of output eigenvectors in lobpcg

2

#### Reference issue Closes gh-16319, gh-16408. gh-15461, gh-15935 #### What does this implement/fix? - Introduced a free and simple tunable restart to potentially increase the attainable accuracy for edge cases...

BUG: DOC: editing docstring example in svds

3

### Is your feature request related to a problem? Please describe. Changed this issue to BUG, since coincidently every PR now fails on exactly this docstring example in refguide_asv_check ,...

### Describe your issue. When the tolerance parameter tol is passed or internally generated in svds, the actual tolerance in calls to the solvers is squared: tol=tol ** 2 which...

### Describe your issue. svds call of arpack for complex case is redirected to eig if k >= N - 1, but any input format of the data matrix for...

hypre availability as a standard Python library

1

It would greatly increase hypre value and usage if it was available as a standard Python library, just like https://github.com/pyamg/pyamg with examples of calling it, e.g., like in amg solver...

no plot in networkx/examples/algorithms /plot_iterated_dynamical_systems.py - delete?

3

### Current Behavior https://github.com/networkx/networkx/blob/main/examples/algorithms/plot_iterated_dynamical_systems.py produces no plots and many functions in the module are never called. ### Expected Behavior Doing something useful as an example in the Gallery. ### Steps...

https://github.com/lobpcg/blopex/blob/master/blopex_abstract/krylov/lobpcg.c is a somewhat simplified C version, following https://arxiv.org/abs/0705.2626, of my original MATLAB implementation https://github.com/lobpcg/blopex/blob/master/blopex_tools/matlab/lobpcg/lobpcg.m Comparing my original MATLAB implementation https://github.com/lobpcg/blopex/blob/master/blopex_tools/matlab/lobpcg/lobpcg.m with https://github.com/lobpcg/blopex/blob/master/blopex_abstract/krylov/lobpcg.c code, I see an important step missing:...

TST: sparse.linalg.lobpcg: add unit tests for accuracy

4

#### Reference issue Closes gh-19355 #### What does this implement/fix? See gh-19355