Glossary

Bayesian methods

analytical methodology that takes advantage of Bayesian logic

credible intervals

The Bayesian equivalent to a confidence interval, the parameter is likely to fall within this interval with the given percentage

homoscedastic

the variance for each data point is the same for all values

likelihood

the probability distribution of some observed data in terms of model parameters

Markov chain Monte Carlo

a random sampling technique that is used to investigate probability distributions

maximum likelihood estimation

the model and parameters that obtain the maximum of the likelihood function for the data

model dataset

the data that arises from our mathematical model

nested sampling

a apporach to estimate the multi-dimension integral that gives the Bayesian evidence

normal distribution

a probability distribution that is often used in the natural sciences to represent real-valued random variables, i.e. experimental measurements, also known as a Gaussian distribution

optimisation algorithm

the process used to try and get the best agreement between our model and experimental datasets

parameters

values within our mathematical model that may be changed

posterior distribution

the result of the product of the likelihood and prior probabilities

prior knowledge

what we already know about our system before looking at our data, e.g., from other measurements of underlying physics/chemistry

thinning

the sub-sampling of MCMC chains to remove correlation between samples

walkers

unique samplers in a Markov chain Monte Carlo sampling