make.lik: Observation Process Distribution Function

In CollocInfer: Collocation Inference for Dynamic Systems

Observation Process Distribution Function

Description

Returns a list of functions that calculate the observation process distribution and its derivatives; designed to be used with the collocation inference functions.

Usage

make.SSElik()

make.multinorm()

Details

These functions require more to be a list with elements:

fn

The transform function of the states to observations, or to their derivatives.

dfdx

The derivative of fn with respect to states.

dfdp

The derivative of fn with respect to parameters.

d2fdx2

The second derivative of fn with respect to states.

d2fdxdp

The cross derivative of fn with respect to states and parameters.

make.Multinorm further requires:

var.fn

The variance given in terms of states and parameters.

var.dfdx

The derivative of var.fn with respect to states.

var.dfdp

The derivative of var.fn with respect to parameters.

var.d2fdx2

The second derivative of var.fn with respect to states.

var.d2fdxdp

The cross derivative of var.fn with respect to states and parameters.

make.SSElik further requres weights giving weights to each observation.

Value

A list of functions that calculate the log observation distribution and its derivatives.

make.SSElik	calculates weighted squared error between predictions (given by fn in more) and observations
make.Multinorm	calculates a multivariate normal distribution.

See Also

LS.setup, multinorm.setup

Examples

# Straightforward sum of squares:

lik = make.SSElik()
lik$more = make.id()

# Multivariate normal about an exponentiated state with constant variance

lik = make.multinorm()
lik$more = c(make.exp(),make.cvar())

CollocInfer documentation built on April 4, 2025, 2:21 a.m.