Move everything into single PySRRegressor

Assert _set_globals can work with both X and n_features

Ensure that equations are up-to-date when getting model representations

Add .latex() representation to PySRRegressor

Add jax, pytorch, sympy output from Regressor

Test __repr__ of sklearn

Improve printed string

Add missing equation in __repr__

Add __repr__ method that lists selected equation

Autogenerate docstring for PySR Regressor

Add missing weights argument to fit()

Create scikit-learn API

Automatically instantiate Project.toml

Cut frequency of simplification 1/5x

Bump SymbolicRegression version with SymbolicUtils fix

Fix variable_names in test

Format code with black

Fix varMap for no feature selection

Add test for Xresampled as pd.DataFrame

Merge pull request #90 from MilesCranmer/deepsource-transform-14177f66

Update TODOs

Format code with black

Ignore install function in coverage since already ran

Have PySR generate Project.toml from Python

Add missing alread_ran=True

Add warning about restarting PySR after install

Add workaround for statically-linked libpython

Get dockerfile working with PyJulia

Format code with black

Include additional kwargs if passed

Cleanup unused imports

Clean up use of global

Clean up Feynman problems

Add missing top-level import

Make pysr.install() install/update packages

Don't return julia output, store in global state

Force update and precompile on all launches

Only activate env on first run

Remove non-PyJulia parts of codebase

PyJulia actually works for distributed if done internally

Use more pure-PyJulia calls rather than evals

Fix issue with re-running outside pyjulia

Let Julia not PyJulia evaluate runfile

Avoid activating package twice with pyjulia

Initial working version with PyJulia

Clean up unused imports

Fix bug with curMSE equal to 0.0

Change default procs to be # of cores

Run all tests with multithreading set to True

Make multithreading the default mode