Fix tests on compute_delays

src/solartoolbox/signalproc.py

@@ -412,8 +412,6 @@ def compute_delays(ts_in, ts_out, mode='loop', scaling='coeff'):
         Time lag between the two timeseries at the maximum value of the cross
         correlation. Values are always integer multiples of the sampling period
         as the max correlation values are limited to the discrete time steps.
-    corr : array(float)
-        The peak value of the cross correlation at the identified delay.
     extra_data : dict{} or None
         if compute_extras was true, additional info will be returned. Fields
         are:

tests/test_signalproc.py

@@ -52,17 +52,15 @@ def test_correlation_illegal(corr_data):
     with raises(ValueError):
         c, lag = correlation(x1, x2, scaling="illegal")
 
+
 @pytest.mark.parametrize("scaling", ['coeff', 'none'])
 def test_fftcorrelate_identity(corr_data, scaling):
     dt, t, x1, x2, dly = corr_data
     c = _fftcorrelate(x1, x1, scaling)
     cr, lag = correlation(x1, x1, scaling=scaling)
-    # import matplotlib.pyplot as plt
-    # plt.plot(c)
-    # plt.plot(cr)
-    # plt.show()
     assert np.allclose(c, cr)
 
+
 @pytest.mark.parametrize("scaling", ['coeff', 'none'])
 def test_fftcorrelate_shift(corr_data, scaling):
     d, t, x1, x2, dly = corr_data
@@ -71,7 +69,6 @@ def test_fftcorrelate_shift(corr_data, scaling):
     assert np.allclose(c, cr)
 
 
-
 def test_averaged_psd():
     # Create a simple sinusoidal signal
     fs = 10  # sample rate
@@ -216,8 +213,6 @@ def test_apply_delay(delay):
     assert np.allclose(tf_delayed['tf'], tf_expected['tf'], atol=1e-5)
 
 
-
-
 def test_averaged_psd_multi():
     np.random.seed(2023)
     # Create a simple sinusoidal signal
@@ -239,8 +234,6 @@ def test_averaged_psd_multi():
     scaling = 'density'
     psds = [averaged_psd(tsig, navgs, overlap, window, detrend, scaling) for tsig in tsigs]
 
-
-
     # Calculate the PSD directly using scipy.signal.welch
     freqs, psd_direct = signal.welch(xs, fs, window, nperseg=len(x1)//navgs,
                                      noverlap=int(overlap*len(x1)//navgs),
@@ -258,7 +251,6 @@ def test_averaged_tf_multi():
     t = np.linspace(0, T, int(T*fs), endpoint=False)  # time variable
     delay = 3
 
-
     x = 0.5 * np.sin(2 * np.pi * 2 * t) + np.random.random(len(t))
 
     y1 = np.roll(x, int(delay*fs))
@@ -288,11 +280,11 @@ def test_averaged_tf_multi():
                               noverlap=int(overlap*len(x)//navgs),
                               detrend=detrend)
     sometf_direct = somePxy / somePxx
-    import matplotlib.pyplot as plt
-    plt.plot(freqs, np.abs(sometf_direct).T)
-    plt.figure()
-    plt.plot(freqs, np.angle(sometf_direct).T)
-    plt.show()
+    # import matplotlib.pyplot as plt
+    # plt.plot(freqs, np.abs(sometf_direct).T)
+    # plt.figure()
+    # plt.plot(freqs, np.angle(sometf_direct).T)
+    # plt.show()
 
 
     # Calculate the PSD directly using scipy.signal.welch
@@ -308,42 +300,17 @@ def test_averaged_tf_multi():
     # Check that the PSDs match
     assert np.allclose(tfs, tf_direct, atol=1e-5)
 
-def test_correlation_multi():
-    np.random.seed(2023)
-    # Create a simple sinusoidal signal
-    fs = 10  # sample rate
-    T = 5.0  # seconds
-    t = np.linspace(0, T, int(T * fs), endpoint=False)  # time variable
-    delay = 0.25
-
-    x = 0.5 * np.sin(2 * np.pi * 2 * t) + np.random.random(len(t))
-
-    y1 = np.roll(x, int(delay * fs))
-    y2 = np.roll(x, int(2 * delay * fs))
-    y3 = np.roll(x, int(3 * delay * fs))
-    y4 = np.roll(x, int(4 * delay * fs))
-
-    ys = [y1, y2, y3, y4]
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in ys]
-
-    corr = [correlation(x_tsig, y_tsig, scaling='none')[0] for y_tsig in ysigs]
-    corr = np.array(corr)
-
-    c = np.flip(signal.correlate([x_tsig], ysigs, method='fft'), axis=0)
-
-    assert np.allclose(corr, c)
 
 @pytest.fixture(params=['loop', 'fft'])
 def compute_delays_modes(request):
     return request.param
 
 
-@pytest.mark.parametrize("delay", [-200,-50.0, -25, -5, 0.0, 5, 25, 50,200])
+@pytest.mark.parametrize("delay", [-200, -50.0, -25, -5, 0.0, 5, 25, 50, 200])
 def test_compute_delays(delay, compute_delays_modes):
     np.random.seed(2023)
     # Create a simple sinusoidal signal
-    fs = 250  # sample rate
+    fs = 100  # sample rate
     T = 500.0  # seconds
     t = np.linspace(0, T, int(T * fs), endpoint=False)  # time variable
     # delay = 0.25
@@ -358,16 +325,12 @@ def test_compute_delays(delay, compute_delays_modes):
         yi = np.roll(x, int(deli * fs))
         ys.append(yi)
         delay_ins.append(deli)
-    # y2 = np.roll(x, int(2 * delay * fs))
-    # y3 = np.roll(x, int(3 * delay * fs))
-    # y4 = np.roll(x, int(4 * delay * fs))
-    #
-    # ys = [x, y1, y2, y3, y4]
+
     x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
     ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in ys]
     xsigs = [x_tsig for y in ys]
     xsigs = pd.DataFrame(np.array(xsigs).T, index=x_tsig.index)
 
-    delays, corrs, _, _ = compute_delays(xsigs, ysigs, compute_delays_modes)
+    delays, _ = compute_delays(xsigs, ysigs, compute_delays_modes)
 
     assert np.allclose(delays, delay_ins)