rht

I tried the backtracking linesearch using this code: ``` linesearch = LineSearches.BackTracking(iterations = 20) ``` 1.024 s (19868 allocations: 935.33 KiB) ``` * Status: success * Candidate solution Final objective...

I didn't specify any gradient function, so I assume it's finite difference. I can't give a rigorous answer about the smoothness of the function (it's a 30-dimensional function), but I...

> Try with an explicit gradient At least, I have got help in Julia Discourse to modify my code so that ForwardDiff works for calculating the gradient. The code is...

My model is a discrete-event simulation, so I'm not exactly sure how to derive the gradient. > Maybe Optim is less optimized for when gradients are not explicitly available (eg...

I encountered the same problem, even with the latest master ``` nix-shell -p '(callPackage (fetchTarball https://github.com/DavHau/mach-nix/tarball/master) {}).mach-nix' ```

Note that even when I tried a single-line requirements.txt ``` backoff-stubs==1.10.0 --hash=sha256:03e995de0a70016c6fe758498e1ca811f1db517c00cbd06e3039c9e4f6ea2566 ``` it still failed to parse the requirements.txt.

I tried with `cryptography>=3.4.8` just now, and was able to install it just fine.

It installed `cryptography` 36.0, which is the version provided by Nixpkgs, but it still failed to install 37.0. This is the error message: ``` Some requirements could not be resolved....

> I'm honestly surprised anyone uses this script still it's really, really, outdated; with respect to option selection. I suppose that no one uses the script because it's slow regardless...

Mypy 0.770 was out 22 days ago: https://github.com/python/mypy/releases/tag/v0.770.