Sync master and dev

.github/workflows/code-quality.yml

@@ -0,0 +1,23 @@
+name: Code quality
+
+on:
+  push:
+    branches: [master, dev]
+  pull_request:
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  lint-format-typecheck:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: '3.12'
+      - name: Install pre-commit
+        run: python -m pip install -r requirements-dev.txt
+      - name: Lint, format, and type-check all source files
+        run: pre-commit run --all-files --show-diff-on-failure

.gitignore

@@ -109,3 +109,5 @@ ENV/
 
 # mypy
 .mypy_cache/
+# Ruff
+.ruff_cache/

.pre-commit-config.yaml

@@ -0,0 +1,18 @@
+default_language_version:
+  python: python3
+repos:
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.17.0
+    hooks:
+      - id: ruff-check
+        args: [--fix]
+      - id: ruff-format
+  - repo: local
+    hooks:
+      - id: mypy
+        name: mypy (all Python sources)
+        entry: python scripts/typecheck.py
+        language: python
+        additional_dependencies: [mypy==2.4.0]
+        pass_filenames: false
+        always_run: true

CONTRIBUTING.md

@@ -11,6 +11,38 @@ Ground Rules
 2. Make the PR detailed and reference [specific issues](https://github.com/cavalab/srbench/issues) if the PR is meant to address any. 
 3. **Please be kind and please be patient**. We will be, too.  
 
+Code quality
+============
+
+Install the development tools and enable the commit hooks from the repository root:
+
+```bash
+python -m pip install -r requirements-dev.txt
+pre-commit install
+pre-commit run --all-files
+```
+
+The hooks run Ruff linting (with safe automatic fixes), Ruff formatting, and mypy.
+CI runs the same hooks over the entire repository on every pull request and on
+pushes to `master` and `dev`. If a hook changes files, review and stage the changes,
+then rerun it before committing. Python scripts and Jupyter notebook code cells
+are linted and formatted; Markdown Python examples are also formatted. Notebook
+outputs are preserved.
+
+Ruff rules and mypy settings live in `pyproject.toml`. Tool versions are pinned
+in `.pre-commit-config.yaml`; update Ruff's hook revision and `required-version`
+together when upgrading. The formatter targets Python 3.7 syntax to accommodate
+older algorithm environments; development hooks and CI run on Python 3.12.
+
+For type checking alone, install `mypy==2.4.0` and run
+`python scripts/typecheck.py`. The runner checks all Python files, including
+hyphenated method directories that mypy's CLI cannot treat as package names.
+Mypy checks unannotated function bodies without requiring type annotations
+throughout this existing codebase. Algorithm libraries are installed in separate
+containers, so third-party imports are treated as `Any` and site packages are
+ignored for consistent results across environments. Python source in notebooks
+is covered by Ruff, but is not type-checked by mypy.
+
 How to contribute an SR method
 ==============================
 
@@ -108,10 +140,10 @@ specify a mapping in the `model` function such as:
 
 ```python
 def model(est, X=None):
-    mapping = {'x_'+str(i):k for i,k in enumerate(X.columns)}
+    mapping = {"x_" + str(i): k for i, k in enumerate(X.columns)}
     new_model = est.model_
-    for k,v in reversed(mapping.items()):
-        new_model = new_model.replace(k,v)
+    for k, v in reversed(mapping.items()):
+        new_model = new_model.replace(k, v)
 ```
 
 2. The operators/functions in the model are available in [sympy's function set](https://docs.sympy.org/latest/modules/functions/index.html). 

algorithms/gp-elite/environment.yml

@@ -0,0 +1,5 @@
+channels:
+  - conda-forge
+dependencies:
+  # Temporary ceiling: pyarrow has no Python 3.15 Conda package yet.
+  - python<3.15

algorithms/gp-elite/requirements.txt

@@ -1 +1 @@
-gp-elite>=0.5.0
+gp-elite>=0.8.0

algorithms/keplearn/environment.yml

@@ -1,6 +1,8 @@
 channels:
   - conda-forge
 dependencies:
+  # Temporary ceiling: pyarrow has no Python 3.15 Conda package yet.
+  - python<3.15
   - numpy
   - sympy
   - scikit-learn

algorithms/pir/environment.yml

@@ -0,0 +1,5 @@
+channels:
+  - conda-forge
+dependencies:
+  # Temporary ceiling: pyarrow has no Python 3.15 Conda package yet.
+  - python<3.15

algorithms/tir/regressor.py

@@ -1,45 +1,50 @@
-import sys
-import os
 import pyTIR as tir
-from itertools import product
-#os.environ["LD_LIBRARY_PATH"] = os.environ["CONDA_PREFIX"] + "/lib"
+
+# os.environ["LD_LIBRARY_PATH"] = os.environ["CONDA_PREFIX"] + "/lib"
 
 hyper_params = [
     {
-        'transfunctions' : ('Id,Tanh,Sin,Cos,Log,Exp,Sqrt',),
-        'ytransfunctions' : ('Id,Sqrt,Exp,Log,ATan,Tan,Tanh',),
-        'exponents' : ((-5,5),)
+        "transfunctions": ("Id,Tanh,Sin,Cos,Log,Exp,Sqrt",),
+        "ytransfunctions": ("Id,Sqrt,Exp,Log,ATan,Tan,Tanh",),
+        "exponents": ((-5, 5),),
     },
     {
-        'transfunctions' : ('Id,Tanh,Sin,Cos,Log,Exp,Sqrt',),
-        'ytransfunctions' : ('Id,Sqrt,Exp,Log,ATan,Tan,Tanh',),
-        'exponents' : ((-2,2),)
+        "transfunctions": ("Id,Tanh,Sin,Cos,Log,Exp,Sqrt",),
+        "ytransfunctions": ("Id,Sqrt,Exp,Log,ATan,Tan,Tanh",),
+        "exponents": ((-2, 2),),
     },
 ]
 
 # Create the pipeline for the model
-eval_kwargs = {'scale_x': False, 'scale_y': False}
-est = tir.TIRRegressor(npop=1000, ngens=500, pc=0.3, pm=0.7, exponents=(-5,5), error="R^2", alg="MOO")
+eval_kwargs = {"scale_x": False, "scale_y": False}
+est = tir.TIRRegressor(
+    npop=1000, ngens=500, pc=0.3, pm=0.7, exponents=(-5, 5), error="R^2", alg="MOO"
+)
+
 
 def pre_train(est, X, y):
     """Adjust settings based on data before training"""
-    if X.shape[0]*X.shape[1] <= 1000:
+    if X.shape[0] * X.shape[1] <= 1000:
         est.penalty = 0.01
 
+
 def complexity(e):
     return e.len
 
+
 def model(e, X):
-    new_model = e.sympy.replace("^","**")
-    for i,f in reversed(list(enumerate(X.columns))):
-        new_model = new_model.replace(f'x{i}',f)
+    new_model = e.sympy.replace("^", "**")
+    for i, f in reversed(list(enumerate(X.columns))):
+        new_model = new_model.replace(f"x{i}", f)
     return new_model
 
+
 def get_population(est):
     pop = []
-    for i in range(min(100, len(est.front))-1,-1,-1):
+    for i in range(min(100, len(est.front)) - 1, -1, -1):
         pop.append(est.create_model_from(i))
-    return pop 
+    return pop
+
 
 def get_best_solution(est):
     return est

docs/CompetitionGuide.md

@@ -79,10 +79,10 @@ specify a mapping in the `model` function such as:
 
 ```python
 def model(est, X):
-    mapping = {'x_'+str(i):k for i,k in enumerate(X.columns)}
+    mapping = {"x_" + str(i): k for i, k in enumerate(X.columns)}
     new_model = est.model_
-    for k,v in reversed(mapping.items()):
-        new_model = new_model.replace(k,v)
+    for k, v in reversed(mapping.items()):
+        new_model = new_model.replace(k, v)
 ```
 
 2. The operators/functions in the model are available in [sympy's function set](https://docs.sympy.org/latest/modules/functions/index.html). 
@@ -92,8 +92,8 @@ If your method uses protected operators (e.g., `my_prot_div(a,b)= a/b if b!=0 el
 
 ```python
 def model(est, X):
-    ... # any previous pre-processing
-    new_model = new_model.replace("my_prot_div","/").replace("my_prot_log","log")
+    ...  # any previous pre-processing
+    new_model = new_model.replace("my_prot_div", "/").replace("my_prot_log", "log")
 ```
 
 ### CLI methods
@@ -146,7 +146,8 @@ In python, one can grab this value as follows:
 
 ```python
 import os
-n_cpus = os.environ['OMP_NUM_THREADS']
+
+n_cpus = os.environ["OMP_NUM_THREADS"]
 ```
 
 ### Time Budget
@@ -172,18 +173,17 @@ function in `regressor.py` as part of `eval_kwargs`.
 As an example, one could define the following in `regressor.py`:
 
 ```python
-def pre_train_fn(est, X, y): 
+def pre_train_fn(est, X, y):
     """set max_time in seconds based on length of X."""
-    if len(X)<=1000:
-        max_time = 360 - 1 # 1 second of slack
+    if len(X) <= 1000:
+        max_time = 360 - 1  # 1 second of slack
     else:
-        max_time = 3600 - 1 # 1 second of slack
+        max_time = 3600 - 1  # 1 second of slack
     est.set_params(max_time)
 
+
 # pass the function to eval_kwargs
-eval_kwargs = {
-    'pre_train': pre_train_fn
-}
+eval_kwargs = {"pre_train": pre_train_fn}
 ```
 
 #### Timeout Handling
@@ -195,22 +195,25 @@ Nonetheless, timeout handling looks like this:
 ```python
 import signal
 
+
 class TimeOutException(Exception):
     pass
 
+
 def alarm_handler(signum, frame):
     print(f"raising TimeOutException")
     raise TimeOutException
 
+
 def evaluate(est):
     """model evaluation"""
     # alarm that sends SIGALRM
     signal.signal(signal.SIGALRM, alarm_handler)
-    signal.alarm(MAXTIME) # maximum time, defined above 
+    signal.alarm(MAXTIME)  # maximum time, defined above
     try:
         est.fit()
     except TimeOutException:
-        print('evaluate timeout')
+        print("evaluate timeout")
 ```
 
 Example approaches are the following: 
@@ -280,39 +283,45 @@ Here is an example of how to do so:
 
 ```python
 from sklearn.model_selection import GridSearchCV
-#from mylibrary import MyMethod
+
+# from mylibrary import MyMethod
 from sklearn import linear_model
 
 
-hyper_params = { 'alpha': (1e-04,0.001,0.01,0.1,1,) }
+hyper_params = {
+    "alpha": (
+        1e-04,
+        0.001,
+        0.01,
+        0.1,
+        1,
+    )
+}
 
 # define your base estimator
-base_est=linear_model.LassoLars()
+base_est = linear_model.LassoLars()
 
 # set est to be a GridSearchCV estimator
-est = GridSearchCV(estimator=base_est, 
-                   param_grid=hyper_params, 
-                   cv=5,
-                   refit=True)
+est = GridSearchCV(estimator=base_est, param_grid=hyper_params, cv=5, refit=True)
 
 # set max time w.r.t. hyper-param tuning
 num_combos = 5  # five values of alpha
-num_cv = 5      # 5 cross-validation splits as above
-total_runs = num_combos * num_cv + 1 # +1 because refit=True
+num_cv = 5  # 5 cross-validation splits as above
+total_runs = num_combos * num_cv + 1  # +1 because refit=True
+
 
-def pre_train_fn(est, X, y): 
+def pre_train_fn(est, X, y):
     """set max_time in seconds based on length of X & hyper-parameter tuning"""
-    if len(X)<=1000:
+    if len(X) <= 1000:
         # account for hyper-parameter tuning, remove 1 extra second of slack to be sure it will terminate on time
-        max_time = 360 // total_runs - 1 
+        max_time = 360 // total_runs - 1
     else:
         max_time = 3600 // total_runs - 1
     est.set_params(max_time)
 
+
 # pass the function eval_kwargs
-eval_kwargs = {
-    'pre_train': pre_train_fn
-}
+eval_kwargs = {"pre_train": pre_train_fn}
 
 # additional definitions
 # ...

docs/CompetitionGuide2023.md

@@ -72,7 +72,7 @@ The final score of each competitor will be composed of:
 The rank will be calculated for each data set independently such that, with N participants, the k-th ranked competitor (k=1 being the best) will be assigned a value of *N - k + 1*. The final score will be the harmonic mean of all of the scores and each participant will be ranked accordingly:
 
 ```python
-score = 2*n / sum([ (1/acc[i]) + (1/simpl[i]) for i in (1..n)])
+score = 2 * n / sum([(1 / acc[i]) + (1 / simpl[i]) for i in ((1.0).n)])
 ```
 
 # Track 2 - Interpretability