Variable names used in IDTxl
Patricia Wollstadt edited this page Dec 2, 2018
·
3 revisions
We try to stick to our variable naming conventions to produce consistent code.
- Where applicable, IDTxl uses SI units for all inputs
- Internally, all time-series data are handled in samples not some time interval, i.e. user input should be translated into sample
- In general, we speak of arrays when we want to denote any non-scalar variable
Measures
| Var. name | Explanation |
|---|---|
| ent | entropy |
| mi | mutual information |
| cmi | conditional mutual information |
| multi | multiinformation (not yet implemented) |
| lais | local active information storage |
| ais | active information storage |
| lte | local transfer entropy |
| te | transfer entropy |
| syn | synergistic information |
| unq | unique information |
| shd | shared information |
Variable Types
| Var. name | Explanation |
|---|---|
| variable | generic random variable (RV) |
| realisations | realisations of a single RV in space or time |
| sample | single realisation of a variable |
| process | indexed series of variables (e.g., indexed by timestamps) |
| replication | copy (e.g., physical or in time) of a process |
| source(_set) | process(es) that have information about another process |
| target | process that receives information from a single or multiple sources |
| current_value | current sample in time in the target process that is predicted from the sources' past |
| past/history | past variables in source and target (relative to the current value) |
| conditional | variable to be conditioned on (e.g. in CMI estimation) |
Estimator Names
Estimator class names are composed of the backend/compute platform, estimator type, and measure estimated, e.g. JidtKraskovCMI.
| Name | Estimator type |
|---|---|
| kraskov | Kraskov estimator |
| kl | Kozachenko-Leonenko estimator |
| gaussian | Gaussian estimator |
| kernel | Kernel estimator |
Algorithm
| Var. name | Explanation |
|---|---|
| idx_*(_set) | index (or set of indices) of single variable |
| candidate(_set) | potential variable(s) for (non-)uniform embedding |
| selected_vars_* | candidate variables currently included in the conditioning set, * may be either 'full', 'sources', or 'target' to indicate all variables and sub-sets of variables coming from source or target processes respectively |
| max_lag | maximum lag for variables entering the candidate set |
| min_lag | minimum lag for variables entering the candidate set |
| theiler_k | n.o. samples to be excluded in neighbour searches, Theiler correction |
| kraskov_k | n.o. nearest neighbours for the Kraskov estimator |