ppigazzini

Errors found on DEV with `pymongo 4.x`: 1. ``` File "/home/usr00/fishtest/server/fishtest/userdb.py", line 77, in add_user_group self.users.save(user) File "/home/usr00/fishtest/server/env/lib/python3.8/site-packages/pymongo/collection.py", line 2584, in __call__ raise TypeError("'Collection' object is not callable. If you...

Tentative solutions working on DEV: 1. https://github.com/glinscott/fishtest/blob/593c3d4004a190ee6bd75bf0b2de5c0be309f2d9/server/fishtest/userdb.py#L74-L77 ```python self.users.replace_one({'_id': user['_id']}, user) ``` 2. https://github.com/glinscott/fishtest/blob/593c3d4004a190ee6bd75bf0b2de5c0be309f2d9/server/fishtest/rundb.py#L480-L485 ```python else: # Otherwise, the count is slow count = self.runs.count_documents(q) ```

I need to review the `delta_update_users.py` changes in #902

Windows support seems to be still a work in progress (`msys2 mingw64` shell): ``` $ python3 make.py -c x86_64-w64-mingw32-gcc -t test % rm -f c-chess-cli c-chess-cli.1.log log out1.pgn out2.pgn stdout...

On Windows msys2 `python3 make.py -c x86_64-w64-mingw32-gcc -t test` and `python3 make.py -c clang -t test` still have the problems reported in July.

@MJZ1977 the companion code of the seminal paper asks for the number of averaged SP gradients to be used per iteration. List updated, thank you :)

The experimental options "careful clipping" and "randomized rounding" don't seems to have a first order effect, so we could keep only one method to clip and to round. - careful...

@MJZ1977 "careful clipping" https://github.com/glinscott/fishtest/commit/7eebda7e6d1f47f2672aefe46db35baee7cb5b1f and randomized rounding https://github.com/glinscott/fishtest/commit/5f63500db3f40569ea406a8b8b4b987f054ee79f are theoretical improvements with little/no effect on SPSA convergence wrt other parameters. People stuck to default, so the GUI was simplified dropping...

@xoto10 the SPSA at very low tc can be also done locally :)

@vondele SPSA claims to minimize the number of function evaluations. Classic SPSA evaluates the function only at "variables_values_k+delta; variables_values_k-delta" for the gradient estimation, so SPSA obviously diverges with wrong delta....