Open Access
Table D.1.
Numerical values of the parameters and hyperparameters used in the different algorithms.
| Parameter | Value | Description |
|---|---|---|
| Algorithm 1: Data reduction | ||
| medfilter | 3 × 3 | Window size of the median filter |
| nit | 6 | Number of optimisation loops |
| − | 20 | Number of iterations of the VMLM-B algorithm to solve Eq. (1) |
| wth | 0.5 | Threshold on the equivalent weights defined in Eq. (10) to identify outlier pixels |
| ∀k, sk = s0 | 5 for RY lup/25 for others | Scaling factor of the robust penalisation in Eq. (9) (∝ the standard deviation of the residuals as seen on Figs. 7o,p for the HR 8799 and RY lup datasets) |
 |
Eq. (A.7),  |
Noise model used in Eq. (4) (Readout noise and Poisson noise) |
| 𝜚 |  |
Normalisation factor of the hyperparameters to balance the unknowns and the measurements |
 |
5 × 10−6 | Hyperparameter of the spectral regularisation defined in Eq. (13) |
 |
2.5 × 10−2 | Hyperparameter of the spatial regularisation defined in Eq. (13) |
| ϵ | 10−16 | Spatial regularisation threshold defined in Eq. (11) |
| Algorithm 2: Remnant signal estimation | ||
| nrs | 100 | Number of iterations of the algorithm for the HR 8799 and RY lup datasets (the algorithm is not run for Ganymede and Io, the background obtained with the shutter opened is used instead). |
| Algorithm 3: Defective pixel identification | ||
| medfilter | 5 × 5 | Window size of the median filter |
| αa | 15 | Tolerance factor |
| Algorithm 4: Spectral dispersion calibration | ||
| nc | 10 | Number of optimisation loops |
| − | 30 | Number of iterations of the VMLM-B algorithm to minimize Eq. (B.8) |
| − | 100 | Number of iterations of the VMLM-B algorithm to minimize Eqs. (B.6) and (B.7) |
| scal | 350 | Scaling factor of the robust penalisation in Eqs. (B.6)–(B.8) |
 |
0|1 | No noise model used in Eqs. (B.6)–(B.8) |
| 𝜚Λ |  |
Normalisation factor of the hyperparameters to balance the unknowns and the measurements |
 |
1.5 × 10−3 | Hyperparameter of the spectral regularisation defined in Eq. (B.8) |
| μw |  |
Hyperparameter to balance the spectral flat-field acquisition with the wavelength calibration file in Eq. (B.7) |
| Autocalibration algorithm | ||
| nr | 3 × 3 | Number of regions of interest |
| mr | 61 | Size of the regions of interest |
| − | 3 | Number of optimisation loops |
| − | 100 | Number of iterations of the VMLM-B algorithm to estimate  via the adapted Eq. (B.8) |
| − | 100 | Number of iterations of a simplex search method to estimate δk via the adapted Eq. (B.6) |
| scal | 350 | Scaling factor of the robust penalisation in Eqs. (B.6) and (B.8) |
 |
0|1 | No noise model used in Eqs. (B.6) and (B.8) |
| 𝜚Λ |  |
Normalisation factor of the hyperparameters to balance the unknowns and the measurements |
 |
8 × 10−4 | Hyperparameter of the spectral regularisation defined in Eq. (B.8) |
Notes. The parameters that are unnamed in the text are noted “−”.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.