Proposes a move in state space.
Operator with a flexible kernel distribution
Operator that moves many parameters (possibly, after transformation to make them more normally distributed). It learns the correlation structure among these parameters during the MCMC run and updates parameters accordingly. doi:10.1093/bioinformatics/btx088
transformations, scaleFactor, coefficient, beta, initial, burnin, every, optimise, store, allowNonsense, kernelDistribution, weight
| type: beast.base.inference.operator.kernel.Transform*** |
| one or more transformed parameters to be moved. For scale parameters use LogTransform (where e.g. scale operators were used). For location parameter use NoTransform (where e.g. random walk operators were used). For parameters that sum to a constant use LogConstrainedSumTransform (where e.g. delta-exchange operators were used). |
| Optional input |
| type: java.lang.Double |
| start scaling factor, larger values give bolder moves (this is tuned during the run) |
| Optional input. Default: 1.0 |
| type: java.lang.Double |
| determines diagonal correlation for variance matrix |
| Optional input. Default: 1.0 |
| type: java.lang.Double |
| fraction of proposal determined by non-covariance matrix |
| Optional input |
| type: java.lang.Integer |
| Number of proposals before covariance matrix is considered in proposal. Must be larger than burnin, if specified. If not specified (or < 0), the operator uses 200 * parameter dimension |
| Optional input. Default: -1 |
| type: java.lang.Integer |
| Number of proposals that are ignored before covariance matrix is being updated. If initial is not specified, uses half the default initial value (which equals 100 * parameter dimension) |
| Optional input. Default: 0 |
| type: java.lang.Integer |
| update interval for covariance matrix, default 1 (that is, every step) |
| Optional input. Default: 1 |
| type: java.lang.Boolean |
| flag to indicate that the scale factor is automatically changed in order to achieve a good acceptance rate (default true) |
| Optional input. Default: true |
| type: java.lang.Boolean |
| flag to indicate if covariance and mean should be stored to the state file (default true) |
| Optional input. Default: true |
| type: java.lang.Boolean |
| flag to indicate if transforms may accept nonsensical inputs eg. 0 parameters (default false) |
| Optional input. Default: false |
| type: beast.base.inference.operator.kernel.KernelDistribution |
| provides sample distribution for proposals |
| Optional input |
| type: java.lang.Double |
| weight with which this operator is selected |
| Required input |