Squeeze morph warning handling (#272)

* Squeeze morph warning handling

* Clarity update

* rework test

* Update test docstrings, warning message

* Update duplicate test docstring

Author: Sparks29032

news/squeeze_warnings.rst

@@ -0,0 +1,23 @@
+**Added:**
+
+* Warnings added to `squeeze` morph if the squeeze causes the grid to become non-monotonic.
+
+**Changed:**
+
+* <news item>
+
+**Deprecated:**
+
+* <news item>
+
+**Removed:**
+
+* <news item>
+
+**Fixed:**
+
+* <news item>
+
+**Security:**
+
+* <news item>

src/diffpy/morph/morph_io.py

@@ -408,9 +408,9 @@ def tabulate_results(multiple_morph_results):
     return tabulated_results
 
 
-def handle_warnings(squeeze_morph):
-    if squeeze_morph is not None:
-        extrapolation_info = squeeze_morph.extrapolation_info
+def handle_extrapolation_warnings(morph):
+    if morph is not None:
+        extrapolation_info = morph.extrapolation_info
         is_extrap_low = extrapolation_info["is_extrap_low"]
         is_extrap_high = extrapolation_info["is_extrap_high"]
         cutoff_low = extrapolation_info["cutoff_low"]
@@ -443,3 +443,35 @@ def handle_warnings(squeeze_morph):
                 wmsg,
                 UserWarning,
             )
+
+
+def handle_check_increase_warning(squeeze_morph):
+    if squeeze_morph is not None:
+        if not squeeze_morph.strictly_increasing:
+            wmsg = (
+                "Warning: The squeeze morph has interpolated your morphed "
+                "function from a non-monotonically increasing grid. "
+                "\nThis may not be an issue, but please check for your "
+                "particular case. "
+                "\nTo avoid squeeze making your grid non-monotonic, "
+                "here are some suggested fixes: "
+                "\n(1) Please decrease the order of your polynomial and "
+                "try again. "
+                "\n(2) If you are using initial guesses of all 0, please "
+                "ensure your objective function only requires a small "
+                "polynomial squeeze to match your reference. "
+                "(In other words, there is good agreement between the two "
+                "functions.) "
+                "\n(3) If you expect a large polynomial squeeze to be "
+                "needed, please ensure your initial parameters for the "
+                "polynomial morph result in good agreement between your "
+                "reference and objective functions. "
+                "One way to obtain such parameters is to "
+                "first apply a --hshift and --stretch morph. "
+                "Then, use the hshift parameter for a0 and stretch "
+                "parameter for a1."
+            )
+            warnings.warn(
+                wmsg,
+                UserWarning,
+            )

src/diffpy/morph/morphapp.py

@@ -707,9 +707,10 @@ def single_morph(
         chain(x_morph, y_morph, x_target, y_target)
 
     # THROW ANY WARNINGS HERE
-    io.handle_warnings(squeeze_morph)
-    io.handle_warnings(shift_morph)
-    io.handle_warnings(stretch_morph)
+    io.handle_extrapolation_warnings(squeeze_morph)
+    io.handle_check_increase_warning(squeeze_morph)
+    io.handle_extrapolation_warnings(shift_morph)
+    io.handle_extrapolation_warnings(stretch_morph)
 
     # Get Rw for the morph range
     rw = tools.getRw(chain)

src/diffpy/morph/morphs/morphsqueeze.py

@@ -1,6 +1,7 @@
 """Class MorphSqueeze -- Apply a polynomial to squeeze the morph
 function."""
 
+import numpy
 from numpy.polynomial import Polynomial
 from scipy.interpolate import CubicSpline
 
@@ -67,10 +68,36 @@ class MorphSqueeze(Morph):
     extrap_index_high = None
     squeeze_cutoff_low = None
     squeeze_cutoff_high = None
+    strictly_increasing = None
 
     def __init__(self, config=None):
         super().__init__(config)
 
+    def _check_strictly_increasing(self, x, x_sorted):
+        if list(x) != list(x_sorted):
+            self.strictly_increasing = False
+        else:
+            self.strictly_increasing = True
+
+    def _sort_squeeze(self, x, y):
+        """Sort x,y according to the value of x."""
+        xy = list(zip(x, y))
+        xy_sorted = sorted(xy, key=lambda pair: pair[0])
+        x_sorted, y_sorted = numpy.array(list(zip(*xy_sorted)))
+        return x_sorted, y_sorted
+
+    def _handle_duplicates(self, x, y):
+        """Remove duplicated x and use the mean value of y corresponded
+        to the duplicated x."""
+        x_unique, inv = numpy.unique(x, return_inverse=True)
+        if len(x_unique) == len(x):
+            return x, y
+        else:
+            y_avg = numpy.zeros_like(x_unique, dtype=float)
+            for idx, _ in enumerate(x_unique):
+                y_avg[idx] = y[inv == idx].mean()
+            return x_unique, y_avg
+
     def morph(self, x_morph, y_morph, x_target, y_target):
         """Apply a polynomial to squeeze the morph function.
 
@@ -82,9 +109,16 @@ def morph(self, x_morph, y_morph, x_target, y_target):
         coeffs = [self.squeeze[f"a{i}"] for i in range(len(self.squeeze))]
         squeeze_polynomial = Polynomial(coeffs)
         x_squeezed = self.x_morph_in + squeeze_polynomial(self.x_morph_in)
-        self.y_morph_out = CubicSpline(x_squeezed, self.y_morph_in)(
+        x_squeezed_sorted, y_morph_sorted = self._sort_squeeze(
+            x_squeezed, self.y_morph_in
+        )
+        self._check_strictly_increasing(x_squeezed, x_squeezed_sorted)
+        x_squeezed_sorted, y_morph_sorted = self._handle_duplicates(
+            x_squeezed_sorted, y_morph_sorted
+        )
+        self.y_morph_out = CubicSpline(x_squeezed_sorted, y_morph_sorted)(
             self.x_morph_in
         )
-        self.set_extrapolation_info(x_squeezed, self.x_morph_in)
+        self.set_extrapolation_info(x_squeezed_sorted, self.x_morph_in)
 
         return self.xyallout

tests/test_morphsqueeze.py

@@ -54,6 +54,8 @@ def test_morphsqueeze(x_morph, x_target, squeeze_coeffs):
     x_target_expected = x_target
     y_target_expected = y_target
     # actual output
+    # turn the coefficients into a list for passing to Polynomial
+    # the morphsqueeze function itself requires a dictionary
     coeffs = [squeeze_coeffs[f"a{i}"] for i in range(len(squeeze_coeffs))]
     squeeze_polynomial = Polynomial(coeffs)
     x_squeezed = x_morph + squeeze_polynomial(x_morph)
@@ -139,16 +141,16 @@ def test_morphsqueeze_extrapolate(user_filesystem, squeeze_coeffs, wmsg_gen):
     coeffs = [squeeze_coeffs[f"a{i}"] for i in range(len(squeeze_coeffs))]
     squeeze_polynomial = Polynomial(coeffs)
     x_squeezed = x_morph + squeeze_polynomial(x_morph)
-    with pytest.warns() as w:
+    with pytest.warns() as warning:
         morphpy.morph_arrays(
             np.array([x_morph, y_morph]).T,
             np.array([x_target, y_target]).T,
             squeeze=coeffs,
             apply=True,
         )
-        assert len(w) == 1
-        assert w[0].category is UserWarning
-        actual_wmsg = str(w[0].message)
+        assert len(warning) == 1
+        assert warning[0].category is UserWarning
+        actual_wmsg = str(warning[0].message)
     expected_wmsg = wmsg_gen([min(x_squeezed), max(x_squeezed)])
     assert actual_wmsg == expected_wmsg
 
@@ -170,3 +172,134 @@ def test_morphsqueeze_extrapolate(user_filesystem, squeeze_coeffs, wmsg_gen):
     )
     with pytest.warns(UserWarning, match=expected_wmsg):
         single_morph(parser, opts, pargs, stdout_flag=False)
+
+
+def test_non_unique_grid():
+    # Test giving morphsqueeze a non-unique grid
+    # Expect it to return a unique grid
+    squeeze_coeffs = {"a0": 0.01, "a1": 0.01, "a2": -0.1}
+    x_grid = np.linspace(0, 10, 101)
+
+    coeffs = [squeeze_coeffs[f"a{i}"] for i in range(len(squeeze_coeffs))]
+    squeeze_polynomial = Polynomial(coeffs)
+    x_morph = x_grid + squeeze_polynomial(x_grid)
+    x_gradient = np.diff(x_morph)
+    x_gradient_sign = np.sign(x_gradient)
+    # non-strictly increasing means the gradient becomes 0 or negative
+    assert not np.all(x_gradient_sign > 0)
+
+    x_target = np.linspace(min(x_morph), max(x_morph), len(x_morph))
+    y_target = np.sin(x_target)
+
+    y_morph = np.sin(x_morph)
+    # apply no squeeze, but the morph should sort the function
+    _, table = morphpy.morph_arrays(
+        np.array([x_morph, y_morph]).T,
+        np.array([x_target, y_target]).T,
+        squeeze=[0, 0, 0],
+        apply=True,
+    )
+    x_refined, _ = table[:, 0], table[:, 1]
+
+    # grid should be properly sorted
+    assert np.allclose(x_refined, x_target)
+    # note that the function itself may be distorted
+
+
+@pytest.mark.parametrize(
+    "squeeze_coeffs, x_morph",
+    [
+        # The following squeezes make the function non-monotonic.
+        # Expect code to work but issue the correct warning.
+        ([-1, -1, 2], np.linspace(-1, 1, 101)),
+        (
+            [-1, -1, 0, 0, 2],
+            np.linspace(-1, 1, 101),
+        ),
+    ],
+)
+def test_squeeze_warnings(user_filesystem, squeeze_coeffs, x_morph):
+    # call in .py
+    x_target = x_morph
+    y_target = np.sin(x_target)
+    squeeze_polynomial = Polynomial(squeeze_coeffs)
+    x_squeezed = x_morph + squeeze_polynomial(x_morph)
+    y_morph = np.sin(x_squeezed)
+    morph = MorphSqueeze()
+    morph.squeeze = squeeze_coeffs
+    with pytest.warns() as warning:
+        morphpy.morph_arrays(
+            np.array([x_morph, y_morph]).T,
+            np.array([x_target, y_target]).T,
+            squeeze=squeeze_coeffs,
+            apply=True,
+        )
+    assert len(warning) == 1
+    assert warning[0].category is UserWarning
+    actual_wmsg = str(warning[0].message)
+    expected_wmsg = (
+        "Warning: The squeeze morph has interpolated your morphed "
+        "function from a non-monotonically increasing grid. "
+        "\nThis may not be an issue, but please check for your "
+        "particular case. "
+        "\nTo avoid squeeze making your grid non-monotonic, "
+        "here are some suggested fixes: "
+        "\n(1) Please decrease the order of your polynomial and try again. "
+        "\n(2) If you are using initial guesses of all 0, please ensure "
+        "your objective function only requires a small polynomial "
+        "squeeze to match your reference. "
+        "(In other words, there is good agreement between the two "
+        "functions.) "
+        "\n(3) If you expect a large polynomial squeeze to be needed, "
+        "please ensure your initial parameters for the polynomial "
+        "morph result in good agreement between your reference and "
+        "objective functions. One way to obtain such parameters is to "
+        "first apply a --hshift and --stretch morph. "
+        "Then, use the hshift parameter for a0 and stretch parameter for a1."
+    )
+    assert expected_wmsg in actual_wmsg
+
+    # call in CLI
+    morph_file, target_file = create_morph_data_file(
+        user_filesystem / "cwd_dir", x_morph, y_morph, x_target, y_target
+    )
+    parser = create_option_parser()
+    (opts, pargs) = parser.parse_args(
+        [
+            "--squeeze",
+            ",".join(map(str, squeeze_coeffs)),
+            f"{morph_file.as_posix()}",
+            f"{target_file.as_posix()}",
+            "--apply",
+            "-n",
+        ]
+    )
+    with pytest.warns(UserWarning) as warning:
+        single_morph(parser, opts, pargs, stdout_flag=False)
+    assert len(warning) == 1
+    actual_wmsg = str(warning[0].message)
+    assert expected_wmsg in actual_wmsg
+
+
+@pytest.mark.parametrize(
+    "x_sampled",
+    [
+        # Expected output: all repeated datapoints are removed
+        # Test one duplicate per number
+        np.array([0, 0, 1, 1, 2, 2, 3, 3]),
+        # Test more than one duplicates per number
+        np.array([0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 2]),
+        # Test with only one grid number
+        np.array([0, 0, 0, 0]),
+        # Test no duplicates
+        np.array([0, 1, 2, 3, 4]),
+    ],
+)
+def test_handle_duplicates(x_sampled):
+    morph = MorphSqueeze()
+    y_sampled = np.sin(x_sampled)
+    x_handled, y_handled = morph._handle_duplicates(x_sampled, y_sampled)
+    x_target = np.unique(x_sampled)
+    y_target = np.array([y_sampled[x_sampled == x].mean() for x in x_target])
+    assert np.allclose(x_handled, x_target)
+    assert np.allclose(y_handled, y_target)