Summary of the GP model, the priors on its hyperparameters, and the estimated median and 68 per cent confidence intervals obtained using an MCMC procedure for the 10 nights data set (see Appendix B. All covariance functions are Matern functions.
Hyperparameter . | Prior . | MCMC estimate (10 nights) . |
---|---|---|
ηsky | +∞ | − |
|$\sigma ^2_{\mathrm{sky}} / \sigma ^2_{\mathrm{n}}$| | − | |$611\substack{+22\\-19}$| |
lsky | |$\mathcal {U}(10, 100)$| | |$47.5\substack{+3.1\\-2.8}$| |
ηmix | 3/2 | − |
|$\sigma ^2_{\mathrm{mix}} / \sigma ^2_{\mathrm{n}}$| | − | |$50.4\substack{+2.1\\-1.9}$| |
lmix | |$\mathcal {U}(1, 10)$| | |$2.97\substack{+0.09\\-0.08}$| |
ηex | 5/2 | − |
|$\sigma ^2_{\mathrm{ex}} / \sigma ^2_{\mathrm{n}}$| | − | |$2.18\substack{+0.09\\-0.14}$| |
lex | |$\mathcal {U}(0.2, 0.8)$| | |$0.26\substack{+0.01\\-0.01}$| |
η21 | 1/2 | − |
|$\sigma ^2_{\mathrm{21}} / \sigma ^2_{\mathrm{n}}$| | − | <0.77 |
l21 | |$\mathcal {U}(0.1, 1.2)$| | >0.73a |
αn | − | |$1.17\substack{+0.06\\-0.06}$| |
Hyperparameter . | Prior . | MCMC estimate (10 nights) . |
---|---|---|
ηsky | +∞ | − |
|$\sigma ^2_{\mathrm{sky}} / \sigma ^2_{\mathrm{n}}$| | − | |$611\substack{+22\\-19}$| |
lsky | |$\mathcal {U}(10, 100)$| | |$47.5\substack{+3.1\\-2.8}$| |
ηmix | 3/2 | − |
|$\sigma ^2_{\mathrm{mix}} / \sigma ^2_{\mathrm{n}}$| | − | |$50.4\substack{+2.1\\-1.9}$| |
lmix | |$\mathcal {U}(1, 10)$| | |$2.97\substack{+0.09\\-0.08}$| |
ηex | 5/2 | − |
|$\sigma ^2_{\mathrm{ex}} / \sigma ^2_{\mathrm{n}}$| | − | |$2.18\substack{+0.09\\-0.14}$| |
lex | |$\mathcal {U}(0.2, 0.8)$| | |$0.26\substack{+0.01\\-0.01}$| |
η21 | 1/2 | − |
|$\sigma ^2_{\mathrm{21}} / \sigma ^2_{\mathrm{n}}$| | − | <0.77 |
l21 | |$\mathcal {U}(0.1, 1.2)$| | >0.73a |
αn | − | |$1.17\substack{+0.06\\-0.06}$| |
Note. aThe upper confidence interval hits the prior boundaries, hence we report here only the lower limit.
Summary of the GP model, the priors on its hyperparameters, and the estimated median and 68 per cent confidence intervals obtained using an MCMC procedure for the 10 nights data set (see Appendix B. All covariance functions are Matern functions.
Hyperparameter . | Prior . | MCMC estimate (10 nights) . |
---|---|---|
ηsky | +∞ | − |
|$\sigma ^2_{\mathrm{sky}} / \sigma ^2_{\mathrm{n}}$| | − | |$611\substack{+22\\-19}$| |
lsky | |$\mathcal {U}(10, 100)$| | |$47.5\substack{+3.1\\-2.8}$| |
ηmix | 3/2 | − |
|$\sigma ^2_{\mathrm{mix}} / \sigma ^2_{\mathrm{n}}$| | − | |$50.4\substack{+2.1\\-1.9}$| |
lmix | |$\mathcal {U}(1, 10)$| | |$2.97\substack{+0.09\\-0.08}$| |
ηex | 5/2 | − |
|$\sigma ^2_{\mathrm{ex}} / \sigma ^2_{\mathrm{n}}$| | − | |$2.18\substack{+0.09\\-0.14}$| |
lex | |$\mathcal {U}(0.2, 0.8)$| | |$0.26\substack{+0.01\\-0.01}$| |
η21 | 1/2 | − |
|$\sigma ^2_{\mathrm{21}} / \sigma ^2_{\mathrm{n}}$| | − | <0.77 |
l21 | |$\mathcal {U}(0.1, 1.2)$| | >0.73a |
αn | − | |$1.17\substack{+0.06\\-0.06}$| |
Hyperparameter . | Prior . | MCMC estimate (10 nights) . |
---|---|---|
ηsky | +∞ | − |
|$\sigma ^2_{\mathrm{sky}} / \sigma ^2_{\mathrm{n}}$| | − | |$611\substack{+22\\-19}$| |
lsky | |$\mathcal {U}(10, 100)$| | |$47.5\substack{+3.1\\-2.8}$| |
ηmix | 3/2 | − |
|$\sigma ^2_{\mathrm{mix}} / \sigma ^2_{\mathrm{n}}$| | − | |$50.4\substack{+2.1\\-1.9}$| |
lmix | |$\mathcal {U}(1, 10)$| | |$2.97\substack{+0.09\\-0.08}$| |
ηex | 5/2 | − |
|$\sigma ^2_{\mathrm{ex}} / \sigma ^2_{\mathrm{n}}$| | − | |$2.18\substack{+0.09\\-0.14}$| |
lex | |$\mathcal {U}(0.2, 0.8)$| | |$0.26\substack{+0.01\\-0.01}$| |
η21 | 1/2 | − |
|$\sigma ^2_{\mathrm{21}} / \sigma ^2_{\mathrm{n}}$| | − | <0.77 |
l21 | |$\mathcal {U}(0.1, 1.2)$| | >0.73a |
αn | − | |$1.17\substack{+0.06\\-0.06}$| |
Note. aThe upper confidence interval hits the prior boundaries, hence we report here only the lower limit.
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.