Table 5.
Components of the covariance model used in ML-GPR along with the priors and converged values of the parameters.
| Component | Covariance | Parameter | Description | Prior bounds | Estimated value |
|---|---|---|---|---|---|
| Intrinsic foregrounds | Radial basis function |
 |
Variance | [−1, 1] | −0.1 ± 0.02 |
| lint | Lengthscale | [20, 40] | 32.9 ± 1.4 | ||
| Mode-mixing foregrounds | Radial basis function |
 |
Variance | [−2, 0] | −1.32 ± 0.01 |
| lmix | Lengthscale | [0.1, 0.5] | 0.275 ± 0.001 | ||
| 21 cm signal | Trained ML Kernel | x1 | Latent space dimension | [−4, 4] | – |
| x2 | Latent space dimension | [−4, 4] | – | ||
 |
Variance | [−7, −1] | < − 4.97 | ||
| Excess power | Exponential function |
 |
Variance | [−3, −1] | −1.88 ± 0.02 |
| lex | Lengthscale | [0.2, 2] | 0.56 ± 0.03 | ||
Notes. All σ2 values are in logarithmic scale and are expressed as a fraction of the variance of the data input into GPR. 
reaches the lower bound and hence only the upper limit is shown. The priors are all uniform priors. All l values are in units of MHz.
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.