Merge pull request #194 from h2020charisma/bugfix_and_tests

Bugfix and tests

src/ramanchada2/spectrum/spectrum.py

@@ -31,6 +31,8 @@ def __init__(self,
                  metadata: Union[SpeMetadataModel, None] = None,
                  applied_processings: Union[SpeProcessingListModel, None] = None):
         super(Plottable, self).__init__()
+        self._xdata = None
+        self._ydata = None
         if x is not None:
             if isinstance(x, int):
                 self.x = np.arange(x) * 1.
@@ -51,14 +53,15 @@ def __init__(self,
         self._cachefile = cachefile
         self._metadata = deepcopy(metadata or SpeMetadataModel(root={}))
         self._applied_processings = deepcopy(applied_processings or SpeProcessingListModel(root=[]))
-        if self.x is not None and self.y is not None:
-            if len(self.x) != len(self.y):
-                raise ValueError(f'x and y shold have same dimentions len(x)={len(self.x)} len(y)={len(self.y)}')
+        if self._xdata is not None and self._ydata is not None:
+            if len(self._xdata) != len(self._ydata):
+                raise ValueError(
+                    f'x and y shold have same dimentions len(x)={len(self._xdata)} len(y)={len(self._ydata)}')
 
     def __copy__(self):
         return Spectrum(
-            x=self.x,
-            y=self.y,
+            x=self._xdata,
+            y=self._ydata,
             cachefile=self._cachefile,
             metadata=self._metadata,
             applied_processings=self._applied_processings,
@@ -125,7 +128,10 @@ def _sort_x(self):
             self.y = self.y[idx]
 
     @property
-    def x(self): return np.array(self._xdata)
+    def x(self):
+        if self._xdata is None:
+            raise ValueError('x of the spectrum is not set. self._xdata is None')
+        return np.array(self._xdata)
 
     @x.setter
     def x(self, val: npt.NDArray[np.float64]):
@@ -142,6 +148,8 @@ def x_bin_boundaries(self):
 
     @property
     def y(self) -> npt.NDArray[np.float64]:
+        if self._ydata is None:
+            raise ValueError('y of the spectrum is not set. self._ydata is None')
         return np.array(self._ydata)
 
     @y.setter
@@ -190,6 +198,8 @@ def x_err(self):
 
     @x_err.setter
     def x_err(self, val: Union[npt.NDArray, None]):
+        if self._xdata is None:
+            raise ValueError('x of the spectrum is not set. self._xdata is None')
         if val is not None:
             if val.shape != self._xdata.shape:
                 raise ValueError(
@@ -205,6 +215,8 @@ def y_err(self):
 
     @y_err.setter
     def y_err(self, val: Union[npt.NDArray, None]):
+        if self._ydata is None:
+            raise ValueError('y of the spectrum is not set. self._ydata is None')
         if val is not None:
             if val.shape != self._ydata.shape:
                 raise ValueError(

tests/end_to_end/test_generate_and_fit.py

@@ -1,5 +1,7 @@
 import numpy as np
+import pandas as pd
 
+import ramanchada2 as rc2
 from ramanchada2 import spectrum
 
 
@@ -20,3 +22,29 @@ def test_generate_and_fit():
     assert len(true_pos) == len(calc_pos), 'wrong number of peaks found'
     assert np.max(np.abs(true_pos - fit_pos - shift)) < 3, 'fit locations far from truth'
     assert np.max(np.abs(true_pos - calc_pos - shift)) < 5, 'find_peaks locations far from truth'
+
+
+def test_gaussian_fit_parameters():
+    x = np.linspace(-100, 3500, 2500)
+    params = [
+        dict(center=400, sigma=6, amplitude=500),
+        dict(center=430, sigma=3, amplitude=300),
+        dict(center=600, sigma=10, amplitude=500),
+        dict(center=625, sigma=5, amplitude=300),
+        dict(center=700, sigma=10, amplitude=500),
+        dict(center=780, sigma=5, amplitude=300),
+    ]
+    spe = rc2.spectrum.from_theoretical_lines(shapes=['gaussian']*len(params),
+                                              params=params, x=x)
+    spe = spe.add_gaussian_noise_drift(sigma=.5, coef=.2, rng_seed=1111)
+    spe = spe.subtract_moving_minimum(100)
+    params_df = pd.DataFrame.from_records(params)
+
+    cand = spe.find_peak_multipeak(prominence=spe.y_noise_MAD()*10, width=3, wlen=100)
+    fitres = spe.fit_peak_multimodel(profile='Gaussian', candidates=cand, vary_baseline=True)
+    df = fitres.to_dataframe_peaks()
+
+    assert np.all(np.isclose(df['amplitude'], params_df['amplitude'], atol=df['amplitude_stderr']*5))
+    assert np.all(np.isclose(df['center'], params_df['center'], atol=df['center_stderr']*5))
+    assert np.all(np.isclose(df['sigma'], params_df['sigma'], atol=df['sigma_stderr']*5))
+    assert np.all(np.isclose(df['sigma']*np.sqrt(8*np.log(2)), df['fwhm']))

tests/spectrum/creators/test_spectrum_init.py

@@ -0,0 +1,24 @@
+import numpy as np
+import pytest
+
+import ramanchada2 as rc2
+
+
+def test_spectrum_init():
+    with pytest.raises(ValueError):
+        spe = rc2.spectrum.Spectrum(x=np.arange(100), y=np.arange(101))
+    with pytest.raises(ValueError):
+        spe = rc2.spectrum.Spectrum(x=np.arange(100), y=np.arange(99))
+    spe = rc2.spectrum.Spectrum(x=np.arange(100, dtype=int), y=np.random.uniform(size=100))
+    assert spe.x.dtype == float
+    assert spe.y.dtype == float
+    spec = spe.__copy__()
+    assert np.all(spe.x == spec.x)
+    assert np.all(spe.y == spec.y)
+    assert spe._xdata is spec._xdata
+    assert spe._ydata is spec._ydata
+
+    spe = rc2.spectrum.Spectrum(x=np.arange(100, dtype=int))
+    assert spe._ydata is None
+    with pytest.raises(ValueError):
+        spe.y