mypy

.github/workflows/test.yml

@@ -5,7 +5,7 @@ on:
 
 jobs:
   nox:
-    timeout-minutes: 180
+    timeout-minutes: 240
     runs-on: ubuntu-latest
     if: >
       github.event_name == 'pull_request' ||

src/jafit/mag.py

@@ -169,6 +169,9 @@ def extract_H_c_B_r_BH_max(hm: npt.NDArray[np.float64], *, intrinsic: bool = Fal
     x = x[(x[:, 0] <= 0) & (x[:, 1] >= 0)]  # Keep only the second quadrant: H<=0, B>=0
     if len(x) > 0:
         H_c = np.abs(np.min(x[:, 0]))
+        # TODO: empirical data is often noisy and sparse, which causes this method to incur a large error.
+        # Interpolation could help with sparsity but not with noise.
+        # Perhaps we could use a high-order least-squares polynomial fit here?
         BH_max = -np.min(x[:, 0] * x[:, 1])
     else:
         H_c, BH_max = 0, 0

src/jafit/opt.py

@@ -138,7 +138,7 @@ def cb(intermediate_result: opt.OptimizeResult) -> None:
     _logger.info("Global optimization result:\n%s", res)
     # We have to check is_done because an early stop is considered an error by the optimizer (strange but true).
     if res.success or (is_done and np.all(np.isfinite(res.x))):
-        return conv(res.x)
+        return conv(res.x)  # type: ignore
     raise RuntimeError(f"Global optimization failed: {res.message}")
 
 
@@ -197,7 +197,7 @@ def cb(intermediate_result: opt.OptimizeResult) -> None:
 
     _logger.info("Local optimization result:\n%s", res)
     if res.success or (is_done and np.all(np.isfinite(res.x))):
-        return conv(res.x)
+        return conv(res.x)  # type: ignore
     raise RuntimeError(f"Local optimization failed: {res.message}")