Base Model für CITT erstellt, PTM Dortmund ergänzt, Tests hinzugefügt

tests/analysis/citt_test.py

@@ -0,0 +1,50 @@
+import logging
+import os
+
+import toml
+from src.paveit.helper import read_file_to_bytesio
+from src.paveit.labtest.citt import CITT_PTMDortmund
+
+logger = logging.getLogger(__name__)
+
+
+def test_base_class():
+    pass
+
+
+def test_citt_ptmdortmund():
+
+    data_path = 'tests/data/citt/PTM_Dortmund'
+
+    res_dict = toml.load(os.path.join(data_path, 'meta.toml'))
+    logger.info(res_dict)
+
+    for filename, meta in res_dict.items():
+
+        logger.info(f'run test on: {filename}, {meta}')
+
+        file = os.path.join(data_path, filename)
+
+        buf = read_file_to_bytesio(file)
+
+        metadata = {'org': 'pytest_ptm_dortmund'}
+
+        res = CITT_PTMDortmund(filename, metadata, archive=False,
+                               data=buf)
+        res.run()
+
+        fit = res.fit.reset_index()
+        
+        assert len(fit) == 5
+        
+        m = res_dict[filename]
+        
+        for col in ['F', 's_hor_sum', 's_hor_1', 's_hor_2']:
+            assert all(fit[f'fit_{col}_r2'] >= m['min_r2'])
+        
+        sel = fit[(fit['f']==10.0) & (fit['sigma']==0.2) & (fit['T']==20.0)].iloc[0]
+        
+        Emin = (1-m['max_diff'])*m['stiffness_10Hz']
+        Emax = (1+m['max_diff'])*m['stiffness_10Hz']
+        
+        assert Emin <= sel['E'] <= Emax

tests/analysis/sine_test.py

@@ -0,0 +1,116 @@
+from random import uniform
+
+import numpy as np
+from paveit.analysis.regression import fit_cos, fit_cos_eval
+
+
+def fit(freq: float = 10,
+        ampl: float = 100.0,
+        offset: float = 20.0,
+        slope: float = 0.1,
+        phase: float = 0.05,
+        error: float = 0.001) -> None:
+
+    N: int = 5
+    num_samples_per_cycle: int = 50
+
+    t = np.linspace(0, N / freq, N * num_samples_per_cycle)
+    y = ampl * np.cos(2 * np.pi * freq * t + phase) + slope * t + offset
+
+    r = fit_cos(t, y)
+
+    error_min = (1 - error)
+    error_max = (1 + error)
+
+    # ampltude
+    rel_error = (r['amp'] / ampl)
+    assert error_min <= rel_error <= error_max
+
+    # offset
+    rel_error = (r['offset'] / offset)
+    assert error_min <= rel_error <= error_max
+
+    # slope
+    rel_error = (r['slope'] / slope)
+    assert error_min <= rel_error <= error_max
+
+    # phase
+    rel_error = (r['phase'] / phase)
+    assert error_min <= rel_error <= error_max
+
+    # freq
+    rel_error = (r['freq'] / freq)
+    assert error_min <= rel_error <= error_max
+
+
+def test_fit_simple_sine(ntest: int = 50) -> None:
+    """
+    fit a simple sine signal and evaluate amplitude
+    
+    error: percentage error of ampl, Error max 0.1 %
+    """
+
+    fit()
+
+    #run multiple tests with random parameters
+    for i in range(ntest):
+
+        fit(
+            ampl=uniform(1e-3, 1000),
+            offset=uniform(1e-3, 1),
+            slope=uniform(1e-5, 1),
+            phase=uniform(1e-5, 1),
+        )
+
+
+def fit_noise(freq: float = 10,
+              ampl: float = 100.0,
+              offset: float = 20.0,
+              slope: float = 0.1,
+              phase: float = 0.05,
+              noise_level: float = 0.01,
+              error: float = 0.01) -> None:
+
+    N: int = 5
+    num_samples_per_cycle: int = 50
+
+    t = np.linspace(0, N / freq, N * num_samples_per_cycle)
+    y = ampl * np.cos(2 * np.pi * freq * t + phase) + slope * t + offset
+    y_noise = np.random.normal(0, noise_level * ampl, len(t))
+
+    y = y + y_noise
+
+    r = fit_cos(t, y)
+
+    error_min = (1 - error)
+    error_max = (1 + error)
+
+    # ampltude
+    rel_error = (r['amp'] / ampl)
+    assert error_min <= rel_error <= error_max
+    
+    # freq
+    rel_error = (r['freq'] / freq)
+    assert error_min <= rel_error <= error_max
+
+
+def test_fit_simple_sine_with_noise(ntest: int = 50) -> None:
+    """
+    fit a simple sine signal and evaluate amplitude
+    
+    error: percentage error of ampl, Error max 0.1 %
+    """
+
+    fit_noise()
+
+    #run multiple tests with random parameters
+    for i in range(ntest):
+
+        fit_noise(
+            ampl=uniform(1e-3, 1000),
+            offset=uniform(1e-3, 1),
+            slope=uniform(1e-5, 1),
+            phase=uniform(1e-5, 1),
+            noise_level=uniform(0.01, 0.1),
+            error=0.02,
+        )