Table 5
Summary of retrievals run on synthetic data.
| Dataset | χ2/ndata | logℒ | BIC | dm |
|---|---|---|---|---|
| KLIP, C | 0.69 | 2692 | −5289 | 2.35 |
| KLIP, diag(C) | 0.41 | 2656 | −5218 | 3.29 |
| PynPoint, C | 1.72 | 2646 | −5220 | 5.63 |
| PynPoint, diag(C) | 1.43 | 2641 | −5208 | 9.60 |
| Andromeda, diag(C) | 0.39 | 2650 | −5266 | 4.62 |
| Gaussian, diag(C) | 1.06 | 2651 | −5267 | 2.21 |
| Noise Free, diag(C) | 0.06 | 2686 | −5336 | 1.69 |
Notes. We compare best fit reduced χ2/ndata the (negative) log likelihood which includes the covariance weighting term of Eq. (9), the Bayesian Information Criterion of Eq. (10) and the Mahalanobis dM from Eq. (11). Retrievals were performed on data processed with each algorithm, both using the full and diagonal only terms of the covariance matrix. Toy models using univariate Gaussian scatter about the input and no scatter are also included, with the uncertainties defined as diag(C) from the KLIP data.
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.