hypotheses

Methods for finding hypotheses of the form (Y, T), that is, pixel Y seems to contain a planet signal at time T.

hsr4hci.hypotheses.get_all_hypotheses(roi_mask, residuals, parang, n_signal_times, frame_size, psf_template, minimum_similarity=0.0, n_roi_splits=1, roi_split=0)

This is a convenience function which wraps the loop over the ROI to call get_hypothesis_for_position() for every spatial pixel. See there for a full documentation of all parameters.

Parameters:

• roi_mask (ndarray) –

• residuals (Dict[str, ndarray]) –

• parang (ndarray) –

• n_signal_times (int) –

• frame_size (Tuple[int, int]) –

• psf_template (ndarray) –

• minimum_similarity (float) –

• n_roi_splits (int) –

• roi_split (int) –

Return type:

Tuple[ndarray, ndarray]

hsr4hci.hypotheses.get_hypothesis_for_position(residuals, position, parang, n_signal_times, frame_size, psf_template, minimum_similarity=0.0)

Take a dictionary containing the full training results and find, for a given spatial position, the time T at which we suspect there to be a planet at the position. If there is no time at which the residuals look like the contain a planet, return NaN.

Parameters:

• residuals (Dict[str, ndarray]) – A dictionary containing the residuals of all models that we have trained. The keys should be “default”, “0”, …, “N”, where the latter are the signal times of the temporal grid that was used during training, and each key should map to a 3D numpy array consisting of the residuals for the respective model.

• position (Tuple[int, int]) – A tuple (x, y) specifying the position for which we want to find the best hypothesis (i.e., the best guess for the time at which this pixel contains a planet).

• parang (ndarray) – A 1D numpy array of shape (n_frames, ) containing the parallactic angles for each frame.

• n_signal_times (int) – An integer specifying the number of different signal times, that is, the size of the temporal grid that was used during training.

• frame_size (Tuple[int, int]) – A tuple (x_size, y_size) specifying the size (in pixels) of the frames that we are working with.

• psf_template (ndarray) – A 2D numpy array containing the unsaturated PSF template for the data set.

• minimum_similarity (float) – Minimum cosine similarity between the expected and the observed signal for the signal to be counted as a hypothesis.

Returns:

A tuple (signal_time, similarity), where signal_time is the best guess for the signal_time of the planet signal in position, and score is the corresponding cosine similarity between the expected and the observed signal. If the position does not seem to contain a planet signal at any time, both of these values are NaN.

Return type:

Tuple[float, float]