zhuwq0
commented
7 months ago sorry, I cannot reproduce your error. I no longer maintain this synthetic test example. Please use the real data examples.
Closed dh4931 closed 2 months ago
zhuwq0
commented
7 months ago sorry, I cannot reproduce your error. I no longer maintain this synthetic test example. Please use the real data examples.
When i run the script example_synthetic.ipynb,I found that the script fails to execute successfully on my computer. I did not modify any parameters in the script, while executing it .
In the first step( Prepare synthetic data), the cell runs successfully. The output parameter _phase_type_ has a shape of (137,), which includes 65 'p' items and 72 's' items.
However, in the second step(Association with GaMMA ), the cell fails to run successfully. The Error occurs at "gmm= BayesianGaussianMixture(...)". and I have verified that the input parameter _num_event_init_ is 41.
my Python version is 3.7.13, my Numpy version is 1.21.5.
error report is as follow #
error report
--------------------------------------------------------------------------- IndexError Traceback (most recent call last) /tmp/ipykernel_98731/1795965897.py in
33 loss_type="l1",
34 # max_covar=10**2,
---> 35 ).fit(data)
36
37 # gmm = GaussianMixture(n_components=num_event_init,
~/software/Anaconda3/envs/phasenet/lib/python3.7/site-packages/gamma/_base.py in fit(self, X, y)
202 self
203 """
--> 204 self.fit_predict(X, y)
205 return self
206
~/software/Anaconda3/envs/phasenet/lib/python3.7/site-packages/gamma/_base.py in fit_predict(self, X, y)
247
248 if do_init:
--> 249 self._initialize_parameters(X, random_state)
250
251 lower_bound = (-np.infty if do_init else self.lower_bound_)
~/software/Anaconda3/envs/phasenet/lib/python3.7/site-packages/gamma/_base.py in _initialize_parameters(self, X, random_state)
166 % self.init_params)
167
--> 168 self._initialize(X, resp)
169
170 @abstractmethod
~/software/Anaconda3/envs/phasenet/lib/python3.7/site-packages/gamma/_bayesian_mixture.py in _initialize(self, X, resp)
541 self._estimate_weights(nk)
542 self._estimate_means(nk, xk)
--> 543 self._estimate_precisions(nk, xk, sk)
544 self.centers_ = centers
545
~/software/Anaconda3/envs/phasenet/lib/python3.7/site-packages/gamma/_bayesian_mixture.py in _estimate_precisions(self, nk, xk, sk)
596 "diag": self._estimate_wishart_diag,
597 "spherical": self._estimate_wishart_spherical
--> 598 }[self.covariance_type](nk, xk, sk)
599
600 self.precisions_cholesky_ = _compute_precision_cholesky(
~/software/Anaconda3/envs/phasenet/lib/python3.7/site-packages/gamma/_bayesian_mixture.py in _estimate_wishart_full(self, nk, xk, sk)
624
625 for k in range(self.n_components):
--> 626 diff = xk[k] - self.mean_prior_
627 self.covariances_[k] = (self.covariance_prior_ + nk[k] * sk[k] +
628 nk[k] * self.mean_precision_prior_ /
IndexError: index 39 is out of bounds for axis 0 with size 39
Additionally, I would like to understand why the parameter num_event_init is defined as
min(int(len(data)/min(num_station, 20) * 6), len(data)). Why is there a comparison with min(num_station, 20)? What is the significance of the value 6? Does it represent six components related to 'p' and 's'?"