`simulate.ctmm.Rd`

Given a `ctmm`

movement model (and optional `telemetry`

data to condition upon) these functions predict or simulate animal locations over a prescribed set of times.

predict(object,...) # S3 method for ctmm predict(object,data=NULL,VMM=NULL,t=NULL,dt=NULL,res=1,complete=FALSE,...) # S3 method for telemetry predict(object,CTMM=NULL,VMM=NULL,t=NULL,dt=NULL,res=1,complete=FALSE,...) simulate(object,nsim=1,seed=NULL,...) # S3 method for ctmm simulate(object,nsim=1,seed=NULL,data=NULL,VMM=NULL,t=NULL,dt=NULL,res=1,complete=FALSE, precompute=FALSE,...) # S3 method for telemetry simulate(object,nsim=1,seed=NULL,CTMM=NULL,VMM=NULL,t=NULL,dt=NULL,res=1,complete=FALSE, precompute=FALSE,...)

object | A |
---|---|

data | Optional |

CTMM | A |

VMM | An optional vertical |

t | Optional array of numeric time values over which the process will be predicted or simulated. |

dt | Timestep to space the prediction or simulation over if |

res | Average number of locations to predict or simulate per |

complete | Additionally calculate timestamps and geographic coordinates. |

nsim | Not yet supported. |

seed | Optional random seed to fix. |

precompute | Precalculate matrices of the Kalman filter (see details). |

... | Unused options. |

The prediction or simulation necessarily requires a `ctmm`

model object.
If a `telemetry`

`data`

object is supplied, the output will be conditional on the `data`

(i.e., simulations that run through the data).
If no `data`

is provided then the output will be purely Gaussian, and times `t`

must be provided.
Details of the movement model parameters can be found in `ctmm.fit`

.

The `t`

argument fixes the output times to a specific array of times.
The `dt`

and `res`

arguments are relative to the sampling schedule present in the optional `telemetry`

object.
The same span of time will be used, while `dt`

will fix the sampling rate absolutely and `res`

will fix the sampling rate relative to that of the data.

The `precompute`

option can speed up calculations of multiple simulations of the same model, data, and *irregular* sampling schedule.
First run `simulate`

with `precompute=TRUE`

to calculate and store all of the necessary matrices of the Kalman filter.
A simulated `telemetry`

object will be produced, as usual, and the precomputed objects are stored in the environment.
Subsequent simulations with `precompute=-1`

will then apply these precomputed matrices for a computational cost savings.
If the sampling schedule is irregular, then this can result in faster simulations.

A simulated animal-tracking `telemetry`

object with components `t`

, `x`

, and `y`

, or a predicted `telemetry`

object that also includes `x`

-`y`

covariances for the location point estimates `x`

and `y`

.

C. H. Fleming, J. M. Calabrese, T. Mueller, K.A. Olson, P. Leimgruber, W. F. Fagan, ``From fine-scale foraging to home ranges: A semi-variance approach to identifying movement modes across spatiotemporal scales'', The American Naturalist, 183:5, E154-E167 (2014) doi: 10.1086/675504 .

C. H. Fleming, D. Sheldon, E. Gurarie, W. F. Fagan, S. LaPoint, J. M. Calabrese, ``Kálmán filters for continuous-time movement models'', Ecological Informatics, 40, 8-21 (2017) doi: 10.1016/j.ecoinf.2017.04.008 .

C. H. Fleming.

Predictions are autocorrelated and should not be treated as data.