use ruff format

.github/workflows/lint.yml

@@ -21,9 +21,9 @@ jobs:
         uses: astral-sh/setup-uv@v6
       - name: Install project
         run: uv sync --locked --dev
-      - name: Run ruff
+      - name: Run ruff lint
         run: uv run ruff check
-      - name: Run black
-        run: uv run black --check .
+      - name: Run ruff format check
+        run: uv run ruff format --check
       - name: Run mypy
         run: uv run make mypy

Makefile

@@ -1,11 +1,11 @@
 RUFF_FLAGS =
-BLACK_FLAGS =
+FORMAT_FLAGS =
 MYPY_FLAGS =
 
 MYPY_TARGETS = \
     src/odil/history.py \
 
-default:
+default: lint format
 
 data:
 	git clone -b data --single-branch git@github.com:cselab/odil.git data
@@ -14,10 +14,14 @@ release:
 	V=$$(sed -rn 's/^version = "(.*)"$$/\1/p' pyproject.toml) && git archive --prefix="odil-$$V/" -o "odil-$$V.tar.gz" HEAD
 
 lint:
-	ruff check --fix $(RUFF_FLAGS) .
-	black $(BLACK_FLAGS) .
+	ruff check --fix $(RUFF_FLAGS)
+
+format:
+	ruff format $(FORMAT_FLAGS)
+
+check: lint format
 
 mypy:
 	mypy $(MYPY_FLAGS) $(MYPY_TARGETS)
 
-.PHONY: default release lint mypy
+.PHONY: default data release lint format check mypy

examples/basic/fields.py

@@ -57,7 +57,6 @@ def parse_args():
 
 
 def plot(problem, state, epoch, frame, cbinfo=None):
-
     domain = problem.domain
     fig, ax = plt.subplots(figsize=(4, 2))
     kw = dict(vmin=0, vmax=1, cmap="Greys", clip_on=False, lw=0.5)

pyproject.toml

@@ -40,27 +40,27 @@ jax = ["jax[cuda13]>=0.7.0"]
 [dependency-groups]
 dev = [
     "pytest>=7.4.4",
-    "black[jupyter]>=25.0.0",
     "ruff>=0.14.0",
     "mypy>=1.18.2",
 ]
 
-[tool.black]
+[tool.ruff]
 line-length = 120
-target-version = ['py312']
+target-version = "py312"
 
-[tool.ruff]
-lint.select = [
+[tool.ruff.lint]
+select = [
     "E",  # Style errors
     "W",  # Style warnings
     "F",  # Logical errors
     "I",  # Import sorting
 ]
-lint.ignore = [
+ignore = [
     "E741",  # Ambiguous variable name
 ]
-output-format = 'concise'
-line-length = 120
+
+[tool.ruff.format]
+quote-style = "double"
 
 [tool.mypy]
 python_version = "3.12"

src/odil/backend.py

@@ -10,7 +10,6 @@
 
 
 class ModBase:
-
     def __init__(self, mod=None):
         assert mod is not None
         self.mod = mod
@@ -43,7 +42,6 @@ def __init__(self, mod=None):
 
 
 class ModNumpy(ModBase):
-
     def __init__(self, mod=None, modsp=None, jax=None):
         mod = mod or np
         super().__init__(mod)
@@ -183,7 +181,6 @@ def conv_transpose(input, filters, output_shape=None, strides=None, padding=None
 
 
 class ModTensorflow(ModBase):
-
     def __init__(self, mod=None, modsp=None):
         super().__init__(mod)
         self.batch_to_space = mod.batch_to_space
@@ -280,7 +277,6 @@ def csr_matrix(*args, **kwargs):
 
 
 class ModCupy(ModBase):
-
     def __init__(self, mod=None, modsp=None):
         super().__init__(mod)
         self.cast = lambda x, dtype: mod.array(x, dtype=dtype)

src/odil/cache.py

@@ -61,7 +61,6 @@ def save(content, path):
         raise ValueError("Unrecognized extension '{}', expected .pickle.".format(ext))
 
     def decorator(func):
-
         def inner(*args, **kwargs):
             name, ext = os.path.splitext(targetbase)
             if len(args) and name_with_arg0:

src/odil/core.py

@@ -9,7 +9,6 @@
 
 
 class Domain:
-
     def __init__(
         self,
         cshape,
@@ -504,7 +503,6 @@ def get_context(self, state, extra=None, tracers=None):
 
 
 class Field:
-
     def __init__(self, array=None, loc=None, cshape=None):
         """
         array: `ndarray`
@@ -527,7 +525,6 @@ def __repr__(self):
 
 
 class MultigridField:
-
     def __init__(self, terms=None, loc=None, factors=None, axes=None, method=None):
         """
         terms: `list` of `Field`
@@ -550,7 +547,6 @@ def __init__(self, terms=None, loc=None, factors=None, axes=None, method=None):
 
 
 class NeuralNet:
-
     def __init__(self, weights=None, biases=None, func_in=None, func_out=None, activation=None):
         """
         weights: `list` of `ndarray`
@@ -574,7 +570,6 @@ def __init__(self, weights=None, biases=None, func_in=None, func_out=None, activ
 
 
 class Array:
-
     def __init__(self, array=None, shape=None):
         """
         array: `ndarray`
@@ -593,7 +588,6 @@ def __repr__(self):
 
 
 class State:
-
     def __init__(self, fields=None, initialized=False):
         """
         fields: A dict mapping field names to state variables
@@ -863,9 +857,7 @@ def eval_neural_net(net: NeuralNet, inputs, mod, frozen=False):
 
 
 class Context:
-
     class Raw:
-
         def __init__(self, value):
             self.value = value
 
@@ -991,7 +983,6 @@ def res(*inputs):
 
 
 class Problem:
-
     def __init__(self, operator, domain, extra=None, tracers=None, jit=None):
         """
         operator: callable(mod, ctx)
@@ -1055,7 +1046,8 @@ def func(arrays):
                 assert len(nonempty) == len(set(nonempty)), "Name of fields must be unique, got {}".format(nonempty)
                 values = [f[1] if isinstance(f, tuple) else f for f in ff]
                 terms = [
-                    mod.mean(v.value) if isinstance(v, Context.Raw) else mod.mean(mod.square(v)) for v in values  #
+                    mod.mean(v.value) if isinstance(v, Context.Raw) else mod.mean(mod.square(v))
+                    for v in values  #
                 ]
                 loss = sum(terms)
                 norms = [t if isinstance(v, Context.Raw) else mod.sqrt(t) for t, v in zip(terms, values)]  #
@@ -1321,7 +1313,6 @@ def _eval_operator_grad_tf(self, state):
 
         def func(arrays, tracers):
             with tf.GradientTape(persistent=True, watch_accessed_variables=False) as tape:
-
                 domain.arrays_to_state(arrays, cache["state"])
                 ctx = Context(
                     domain,
@@ -1477,7 +1468,6 @@ def latin_hypercube(ndim, size, dtype):
 
 
 class Approx:
-
     def __init__(self, domain):
         self.domain = domain
         self.mod = domain.mod

src/odil/core_min.py

@@ -4,7 +4,6 @@
 
 
 class Domain:
-
     def __init__(self, domain=None, ndim=None, lower=None, upper=None, dimnames=None, dtype=None, cshape=None):
         domain = domain or Namespace(ndim=None, lower=0.0, upper=1.0, dimnames=None, dtype=None, cshape=None)
         dtype = dtype or domain.dtype

src/odil/history.py

@@ -7,7 +7,6 @@
 
 
 class History:
-
     def __init__(self, csvpath: str | None = None, warmup: int = 0):
         """
         warmup: `int`

src/odil/linsolver.py

@@ -119,9 +119,7 @@ def add_arguments(parser):
     parser.add_argument("--smooth_pre", type=int, default=2, help="Pre-smoothing steps")
     parser.add_argument("--smooth_post", type=int, default=2, help="Post-smoothing steps")
     parser.add_argument("--omega", type=float, default=0.6, help="Jacobi smoother relaxation factor")
-    parser.add_argument(
-        "--ndirect", type=int, default=3, help="Systems on smaller grids " "are solved with direct solver"
-    )
+    parser.add_argument("--ndirect", type=int, default=3, help="Systems on smaller grids are solved with direct solver")
     parser.add_argument(
         "--restriction",
         type=str,

src/odil/optimizer.py

@@ -4,7 +4,6 @@
 
 
 class Optimizer:
-
     def __init__(self, name=None, displayname=None, dtype=None):
         self.name = name
         self.displayname = displayname if displayname is not None else name
@@ -20,14 +19,12 @@ def run(self, x0, loss_grad, epochs, callback=None, epoch_start=0, **kwargs):
 
 
 class EarlyStopError(Exception):
-
     def __init__(self, msg, optinfo):
         super().__init__(msg)
         self.optinfo = optinfo
 
 
 class LbfgsbOptimizer(Optimizer):
-
     def __init__(self, pgtol=1e-16, m=50, maxls=50, factr=0, dtype=None, mod=None, **kwargs):
         """
         pgtol: `float`
@@ -118,7 +115,6 @@ def func_wrap(x):
 
 
 class LbfgsOptimizer(Optimizer):
-
     def __init__(self, pgtol=1e-16, m=50, maxls=50, factr=0, dtype=None, mod=None, **kwargs):
         """
         pgtol: `float`
@@ -206,18 +202,15 @@ def stopping_condition(converged, failed):
 
 
 class AdamTfOptimizer(Optimizer):
-
     def __init__(self, dtype=None, mod=None, **kwargs):
         super().__init__(name="adam_tf", displayname="AdamTf", dtype=dtype)
         self.mod = mod
 
     def run(self, x0, loss_grad, epochs=None, callback=None, lr=1e-3, epoch_start=0, **kwargs):
-
         mod = self.mod
         tf = mod.tf
 
         class CustomModel(tf.keras.Model):
-
             def __init__(self, x):
                 super().__init__()
                 self.x = x
@@ -238,7 +231,6 @@ def train_step(self, _):
         model = CustomModel(x)
 
         class CustomCallback(tf.keras.callbacks.Callback):
-
             def on_epoch_end(self, epoch, logs=None):
                 model.epoch = epoch_start + epoch
                 if epoch > 0 and callback:
@@ -254,13 +246,11 @@ def on_epoch_end(self, epoch, logs=None):
 
 
 class GdOptimizer(Optimizer):
-
     def __init__(self, dtype=None, mod=None, **kwargs):
         super().__init__(name="gd", displayname="GD", dtype=dtype)
         self.mod = mod
 
     def run(self, x0, loss_grad, epochs=None, callback=None, lr=1e-3, epoch_start=0, **kwargs):
-
         mod = self.mod
         x = [mod.copy(e) for e in x0]
         for epoch in range(epoch_start + 1, epoch_start + epochs + 1):
@@ -278,7 +268,6 @@ def run(self, x0, loss_grad, epochs=None, callback=None, lr=1e-3, epoch_start=0,
 
 
 class AdamNativeOptimizer(Optimizer):
-
     def __init__(self, dtype=None, mod=None, **kwargs):
         super().__init__(name="adamn", displayname="AdamNative", dtype=dtype)
         self.mod = mod

src/odil/runtime.py

@@ -7,7 +7,7 @@
 
 if not int(os.environ.get("ODIL_MT", 0)):
     os.environ["OMP_NUM_THREADS"] = "1"
-    os.environ["XLA_FLAGS"] = "--xla_cpu_multi_thread_eigen=false " "intra_op_parallelism_threads=1"
+    os.environ["XLA_FLAGS"] = "--xla_cpu_multi_thread_eigen=false intra_op_parallelism_threads=1"
     os.environ["TENSORFLOW_INTER_OP_PARALLELISM"] = "1"
     os.environ["TENSORFLOW_INTRA_OP_PARALLELISM"] = "1"