Come analyze HEASARC, IRSA, and MAST data in the cloud! The Fornax Initiative is now welcoming all interested beta users.
Glossary
- Component – A part of a model.
- Additive – A component which produces a theoretical spectrum.
- Multiplicative – A component which produces multiplicative factors to apply to the current theoretical spectrum.
- Convolution – A component which modifies the current theoretical spectrum
- Mixing – If there are multiple datasets with separately calculated theoretical models this component modifies all of them simultaneously.
- Covariance group – A set of spectra (for example the Stokes I, Q and U spectra of a source) whose bin-by-bin covariances are carried together in the data files and used by the chistokes (chicov) fit statistic instead of treating the spectra as independent.
- Datagroup – A set of observed spectra which will all be fit using the same model and parameter values.
- Dataset – The observed spectrum and its associated information.
- Evidence – The marginal likelihood of the data given a model, integrated over the prior. It is returned by the nest command and is the basis for Bayesian model comparison.
- Fit statistic – The statistic used to determine the best fit value. The options are based on maximizing the likelihood, or, when priors are set with the bayes command, the posterior.
- Live points – The set of active samples maintained by the nest (nested sampling) algorithm. More live points explore the prior volume more thoroughly at a proportionally higher cost.
- Model – The parameterized theoretical model for the source spectrum. There is a default unnamed model but additional named models can be defined.
- Plotgroup – A set of spectra which are summed together for plotting purposes only.
- Posterior – The probability distribution of the parameters given the data, proportional to the likelihood times the prior. It is explored by the MCMC (chain, hmc) and nested-sampling (nest) methods.
- Prior – The probability distribution assigned to a parameter before the data are considered, set with the bayes command. It is required for Bayesian inference.
- Response Model – A model which modifies the response. At the moment this is limited to the gain.
- Source – In some observations several objects can contribute to the same spectrum. In this case XSPEC refers to them as different sources and each source can have its own model and response.
- Table Model – A model component stored as a FITS file containing model spectra tabulated on a grid of parameter values.
- Test Statistic – The statistic to provide the
goodness-of-fit criterion after the parameters have been optimized
based on the fit statistic