fix: resolve ruff violations blocking CI (I001, F401, E501, E702, F841)

src/bssunfold/core/__init__.py

@@ -131,7 +131,6 @@
 from .unfold_maxed import solve_maxed, unfold_maxed
 from .unfold_mcmc import solve_bayesian_mcmc, unfold_mcmc
 from .unfold_mlem import solve_mlem, unfold_mlem
-from .unfold_nnqp import solve_nnqp, unfold_nnqp
 from .unfold_mlem_bs import (
     AUTO_BETA_RELATIVE_GRID,
     build_bspline_basis,
@@ -157,6 +156,7 @@
     solve_tikhonov_nnls,
     unfold_nnksvd,
 )
+from .unfold_nnqp import solve_nnqp, unfold_nnqp
 from .unfold_nsduaz import (
     builtin_catalogue,
     select_catalogue_initial,

src/bssunfold/core/_gnowee.py

@@ -36,9 +36,8 @@
 
 from __future__ import annotations
 
-import copy as cp
 from collections.abc import Callable
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from math import gamma, sqrt
 
 import numpy as np
@@ -316,7 +315,8 @@ def initialize(self, num_samples: int, method: str | None = None) -> np.ndarray:
         elif method in ("lhc", "lhs"):
             try:
                 from scipy.stats.qmc import LatinHypercube
-                sampler = LatinHypercube(d=n_dim, seed=int(self.rng.integers(0, 2**31 - 1)))
+                seed_int = int(self.rng.integers(0, 2**31 - 1))
+                sampler = LatinHypercube(d=n_dim, seed=seed_int)
                 unit = sampler.random(n=num_samples)
             except Exception:
                 # Fallback: crude Latin-hypercube via permutation per dimension
@@ -442,7 +442,8 @@ def mutate(self, pop: list[Parent]) -> list[np.ndarray]:
         perm1 = self.rng.permutation(n)
         perm2 = self.rng.permutation(n)
         r = float(self.rng.random())
-        k = self.rng.random((n, dim)) > (self.s.frac_mutation * float(self.rng.random()))
+        k = self.rng.random((n, dim)) > (
+            self.s.frac_mutation * float(self.rng.random()))
         diff = pop_arr[perm1] - pop_arr[perm2]
         step = r * diff
         children = pop_arr + step * k
@@ -456,7 +457,8 @@ def population_update(self, parents: list[Parent], children: list[np.ndarray],
                           timeline: list[Event] | None = None,
                           adopted_parents: list[int] | None = None,
                           mh_frac: float = 0.0,
-                          random_parents: bool = False) -> tuple[list[Parent], int, list[Event] | None]:
+                          random_parents: bool = False,
+                          ) -> tuple[list[Parent], int, list[Event] | None]:
         """Evaluate children and replace worse parents (port of ``population_update``).
 
         Keeps the parents sorted ascending by fitness.  Implements:
@@ -474,7 +476,6 @@ def population_update(self, parents: list[Parent], children: list[np.ndarray],
 
         replace = 0
         feval = 0
-        worst = max((p.fitness for p in parents), default=self.s.penalty)
         for i, child_vars in enumerate(children):
             fnew = float(self.objective(np.asarray(child_vars, dtype=float)))
             if fnew > self.s.penalty:
@@ -551,7 +552,8 @@ def run_gnowee(lb: np.ndarray, ub: np.ndarray,
                settings: GnoweeSettings | None = None,
                rng: np.random.Generator | None = None,
                seed_solution: np.ndarray | None = None,
-               extra_starting: np.ndarray | None = None) -> tuple[np.ndarray, float, list[Event]]:
+               extra_starting: np.ndarray | None = None,
+               ) -> tuple[np.ndarray, float, list[Event]]:
     """Run the Gnowee optimizer on a continuous problem.
 
     Parameters
@@ -606,8 +608,6 @@ def run_gnowee(lb: np.ndarray, ub: np.ndarray,
     timeline.append(Event(0, len(pop), pop[0].fitness,
                           np.array(pop[0].variables, dtype=float)))
 
-    fe = gh.s.frac_elite
-    fl = gh.s.frac_levy
     converge = False
     while not converge:
         # Gnowee re-samples the elite/levy fractions each generation for MI
@@ -669,7 +669,8 @@ def run_gnowee(lb: np.ndarray, ub: np.ndarray,
                 converge = True
                 if settings.verbose:
                     print("Gnowee: fitness convergence (absolute).")
-        elif abs((pop[0].fitness - settings.optimum) / settings.optimum) <= settings.opt_conv_tol:
+        elif (abs((pop[0].fitness - settings.optimum) / settings.optimum)
+              <= settings.opt_conv_tol):
             converge = True
             if settings.verbose:
                 print("Gnowee: fitness convergence (relative).")

src/bssunfold/core/detector.py

@@ -102,6 +102,7 @@
 from .unfold_mystic import unfold_mystic as unfold_mystic_impl
 from .unfold_mystic import unfold_mystic_hybrid as unfold_mystic_hybrid_impl
 from .unfold_nnksvd import unfold_nnksvd as unfold_nnksvd_impl
+from .unfold_nnqp import unfold_nnqp as unfold_nnqp_impl
 from .unfold_nsduaz import unfold_nsduaz as unfold_nsduaz_impl
 from .unfold_nspline import unfold_nspline as unfold_nspline_impl
 from .unfold_odl_advanced import (
@@ -116,7 +117,6 @@
 from .unfold_pspline_reml import (
     unfold_pspline_reml as unfold_pspline_reml_impl,
 )
-from .unfold_nnqp import unfold_nnqp as unfold_nnqp_impl
 from .unfold_qpmad import unfold_qpmad as unfold_qpmad_impl
 from .unfold_qpsolvers import unfold_qpsolvers as unfold_qpsolvers_impl
 from .unfold_qubo import unfold_qubo as unfold_qubo_impl

src/bssunfold/core/unfold_gnowee.py

@@ -69,7 +69,8 @@ def _build_seed(A: np.ndarray, b: np.ndarray, x0: np.ndarray | None) -> np.ndarr
     return seed
 
 
-def _build_log_bounds(seed: np.ndarray, half_range: float) -> tuple[np.ndarray, np.ndarray]:
+def _build_log_bounds(seed: np.ndarray, half_range: float,
+                      ) -> tuple[np.ndarray, np.ndarray]:
     """Return ``(lb, ub)`` in log space centred on the seed."""
     y0 = np.log(np.maximum(np.asarray(seed, dtype=float), 1e-300))
     span = half_range * np.log(10.0)

src/bssunfold/core/unfold_qpmad.py

@@ -48,7 +48,7 @@
 from typing import Any
 
 import numpy as np
-from scipy.linalg import cho_factor, cho_solve, solve_triangular
+from scipy.linalg import solve_triangular
 
 from ..logging_config import get_logger
 from ._base_unfolder import make_solve_wrapper, run_unfolding
@@ -154,11 +154,15 @@ def solve_H(rhs: np.ndarray) -> np.ndarray:
         ub_arr = np.asarray(ub, dtype=float)
         for i in range(n):
             if np.isfinite(lb_arr[i]):
-                r = np.zeros(n); r[i] = 1.0
-                rows_C.append(r); vals_d.append(float(lb_arr[i]))
+                r = np.zeros(n)
+                r[i] = 1.0
+                rows_C.append(r)
+                vals_d.append(float(lb_arr[i]))
             if np.isfinite(ub_arr[i]):
-                r = np.zeros(n); r[i] = -1.0
-                rows_C.append(r); vals_d.append(-float(ub_arr[i]))
+                r = np.zeros(n)
+                r[i] = -1.0
+                rows_C.append(r)
+                vals_d.append(-float(ub_arr[i]))
 
     if A is not None and lb_A is not None and ub_A is not None:
         A_arr = np.asarray(A, dtype=float)
@@ -396,7 +400,8 @@ def _solve_qp_qpmad_cpp(
             )
             return np.asarray(x, dtype=float), "OK" if status == 0 else "INFEASIBLE"
         except Exception as e:
-            logger.warning(f"qpmad.solve call failed: {e}; falling back to python backend")
+            logger.warning(
+                f"qpmad.solve call failed: {e}; falling back to python backend")
 
     if hasattr(qpmad_module, "Solver"):
         # OOP form: solver = qpmad.Solver(); solver.solve(...)
@@ -416,7 +421,9 @@ def _solve_qp_qpmad_cpp(
                 return np.zeros(n), "INFEASIBLE"
             return np.asarray(x, dtype=float), "OK" if status == 0 else "INFEASIBLE"
         except Exception as e:
-            logger.warning(f"qpmad.Solver.solve call failed: {e}; falling back to python backend")
+            logger.warning(
+                f"qpmad.Solver.solve call failed: {e};"
+                " falling back to python backend")
 
     raise ImportError(
         "qpmad Python bindings found but no compatible solve API. "

tests/test_coverage.py

@@ -2546,7 +2546,8 @@ def _make_detector(self):
 
     @pytest.mark.parametrize(
         "method",
-        ["cvxpy", "qpsolvers", "mlem", "landweber", "genetic", "gnowee", "nnqp", "qpmad"],
+        ["cvxpy", "qpsolvers", "mlem", "landweber", "genetic",
+         "gnowee", "nnqp", "qpmad"],
     )
     def test_spectrum_zero_above_emax(self, method):
         det = self._make_detector()

tests/test_gnowee.py

@@ -20,7 +20,6 @@
 from bssunfold.core._gnowee import (
     GnoweeHeuristics,
     GnoweeSettings,
-    Parent,
     levy,
     rejection_bounds,
     run_gnowee,