Merge pull request monocongo#497 from WeatherGod/use-abs-tolerance

Force get_tolerance() to always be greater than zero.

src/climate_indices/__main__.py

@@ -88,11 +88,6 @@ def _validate_args(args):
                                          ("division", "time"),
                                          ("division")]
 
-    # dynamic threshold absolute tolerance parameter np.allclose
-    # derived from (smallest) grid size along dimension dim
-    def get_tolerance(dim):
-        return np.diff(dim).min() / 10
-
     # all indices except PET require a precipitation file
     if args.index != "pet":
 
@@ -218,11 +213,11 @@ def get_tolerance(dim):
                         raise ValueError(msg)
 
                     # verify that the coordinate variables match with those of the precipitation dataset
-                    if not np.allclose(lats_precip, dataset_pet["lat"][:], atol=get_tolerance(lats_precip)):
+                    if not np.allclose(lats_precip, dataset_pet["lat"][:], atol=utils.get_tolerance(lats_precip)):
                         msg = "Precipitation and PET variables contain non-matching latitudes"
                         _logger.error(msg)
                         raise ValueError(msg)
-                    elif not np.allclose(lons_precip, dataset_pet["lon"][:], atol=get_tolerance(lons_precip)):
+                    elif not np.allclose(lons_precip, dataset_pet["lon"][:], atol=utils.get_tolerance(lons_precip)):
                         msg = "Precipitation and PET variables contain non-matching longitudes"
                         _logger.error(msg)
                         raise ValueError(msg)
@@ -300,11 +295,11 @@ def get_tolerance(dim):
                         raise ValueError(msg)
 
                     # verify that the coordinate variables match with those of the precipitation dataset
-                    if not np.allclose(lats_precip, dataset_temp["lat"][:], atol=get_tolerance(lats_precip)):
+                    if not np.allclose(lats_precip, dataset_temp["lat"][:], atol=utils.get_tolerance(lats_precip)):
                         msg = "Precipitation and temperature variables contain non-matching latitudes"
                         _logger.error(msg)
                         raise ValueError(msg)
-                    elif not np.allclose(lons_precip, dataset_temp["lon"][:], atol=get_tolerance(lons_precip)):
+                    elif not np.allclose(lons_precip, dataset_temp["lon"][:], atol=utils.get_tolerance(lons_precip)):
                         msg = "Precipitation and temperature variables contain non-matching longitudes"
                         _logger.error(msg)
                         raise ValueError(msg)
@@ -378,11 +373,11 @@ def get_tolerance(dim):
                         raise ValueError(msg)
 
                     # verify that the coordinate variables match with those of the precipitation dataset
-                    if not np.allclose(lats_precip, dataset_awc["lat"][:], atol=get_tolerance(lats_precip)):
+                    if not np.allclose(lats_precip, dataset_awc["lat"][:], atol=utils.get_tolerance(lats_precip)):
                         msg = "Precipitation and AWC variables contain non-matching latitudes"
                         _logger.error(msg)
                         raise ValueError(msg)
-                    elif not np.allclose(lons_precip, dataset_awc["lon"][:], atol=get_tolerance(lons_precip)):
+                    elif not np.allclose(lons_precip, dataset_awc["lon"][:], atol=utils.get_tolerance(lons_precip)):
                         msg = "Precipitation and AWC variables contain non-matching longitudes"
                         _logger.error(msg)
                         raise ValueError(msg)

src/climate_indices/utils.py

@@ -489,4 +489,15 @@ def count_zeros_and_non_missings(
     return zeros, non_missings
 
 
+# ------------------------------------------------------------------------------
+def get_tolerance(dim: np.ndarray) -> float:
+    """
+    dynamic threshold absolute tolerance parameter np.allclose
+    derived from (smallest) absolute grid size along dimension dim.
+    Always greater than zero.
+    """
+    tol = np.abs(np.diff(dim).min() / 10)
+    return max(tol, np.finfo(tol.dtype).resolution)
+
+
 _logger = get_logger(__name__, logging.DEBUG)

tests/test_utils.py

@@ -488,3 +488,26 @@ def test_transform_to_366day():
                              values_365,
                              1972,
                              24)
+
+def test_tolerance():
+    lons, dlon = np.linspace(-180.0, 180.0, 250, retstep=True)
+    tol = utils.get_tolerance(lons)
+    assert tol > 0, "Tolerance must always be greater than zero"
+    assert tol < abs(dlon), "Tolerance must always come out smaller than the coordinate delta"
+
+    lats, dlat = np.linspace(25.0, -15.0, 50, retstep=True)
+    tol = utils.get_tolerance(lats)
+    assert tol > 0, "Tolerance must always be greater than zero"
+    assert tol < abs(dlat), "Tolerance must always come out smaller than the coordinate delta"
+
+    meshlat, meshlon = np.meshgrid(lats, lons)
+    tol = utils.get_tolerance(meshlat)
+    assert tol > 0, "Tolerance must always be greater than zero"
+    assert tol < abs(dlat), "Tolerance must always come out smaller than the coordinate delta"
+    # Tricky situation because np.diff() on this grid returns all zeros.
+    # Not the greatest situation, but we can at least allow a tiny bit of tolerance.
+    # Would be nice to be smarter about this situation.
+    tol = utils.get_tolerance(meshlon)
+    assert tol > 0, "Tolerance must always be greater than zero"
+    assert tol < abs(dlon), "Tolerance must always come out smaller than the coordinate delta"
+