bm2: Brownian motion spatPomp generator with shared or...

In kidusasfaw/spatPomp: Inference for Spatiotemporal Partially Observed Markov Processes

Brownian motion spatPomp generator with shared or unit-specific parameters

Description

An extension of bm allowing for shared or unit-specific parameters. Generate a class ‘spatPomp’ object representing a U-dimensional Brownian motion with spatial correlation decaying geometrically with distance around a circle. The model is defined in continuous time though in this case an Euler approximation is exact at the evaluation times.

Usage

bm2(
  U = 5,
  N = 100,
  delta_t = 0.1,
  unit_specific_names = "rho",
  shared_names = NULL,
  unit_params = c(rho = 0.4, sigma = 1, tau = 1, X_0 = 0)
)

Arguments

U	A length-one numeric signifying dimension of the process.
N	A length-one numeric signifying the number of observation time steps to evolve the process.
delta_t	Process simulations are performed every delta_t time units whereas observations occur every one time unit
unit_specific_names	determines which parameters take a different value for each unit. Cannot be specified if shared_names is specified. each unit. Other parameters are considered shared between all units.
shared_names	identifies parameters that have common shared value for all units, which by default is all parameters.
unit_params	parameter values used to build the object, copied across each unit for unit-specific parameters

Value

An object of class ‘spatPomp’ representing a simulation from a U-dimensional Brownian motion

Author(s)

Edward L. Ionides

See Also

Other spatPomp model generators: bm(), gbm(), he10(), lorenz(), measles()

Examples

# Complete examples are provided in the package tests
## Not run: 
b <- bm2(U=4, N=20,shared_names="rho",unit_specific_names=c("sigma","tau"))
# See all the model specifications of the object
spy(b)
# Examples of methodologies applied to this model
# are provided in the tests directory

## End(Not run)

kidusasfaw/spatPomp documentation built on April 27, 2024, 2:20 p.m.