Optimizer step for estimating the Heligman-Pollard Parameters using the Bayesian Melding with IMIS-opt procedure

Description

Performs the optimizer step in the IMIS procedure for the eight Heligman-Pollard parameters

Usage

1
2
loop.optim(prior, nrisk, ndeath, d = 10, theta.dim = 8, 
age = c(1e-05, 1, seq(5, 100, 5)))

Arguments

prior

A matrix containing the prior

nrisk

A vector containing the number of persons at risk in each age group

ndeath

A vector containing the number of deaths in each age group

d

Number of optimizer iterations

theta.dim

Number of columns of the prior (This should be 8 if estimating all parameters. Functionality for estimation a limited number of parameters does not exist yet.)

age

A vector containing the ages at which each age interval begins

Value

opt.mu.d

A matrix containing the local optimums resulting from the optimizer step. Each local optimum contains a set of 8 parameter values.

opt.cov.d

A array containing the covariance matrix for each of the local optimums

d.keep

The number of local optimums found whose likelihood is greater than the maximum likelihood from the prior

theta.new

The set of parameters from the prior with the greatest weight as calculated with prior.likewts

log.like.0

A vector containing a likelihood for each row of the prior

wts.0

A vector containing an importance weight for each row of the prior

Warning

If the likelihood for the initial local maximum does not exceed the highlest likelihood from the prior, a warning will be issued.

Note

Occasionally, this step fails to produce an initial local maximum that exceeds the highest likelihood of the prior and a warning is issued. Usually drawing a new prior or selecting a different algorithm solves this problem.

References

Poole, David and Adrian Raftery. 2000. "Inference for Deterministic Simulation Models: The Bayesian Melding Approach." Journal of the American Statistical Association 95:1244–1255.

Raftery, Adrian and Le Bao. 2009. "Estimating and Projecting Trends in HIV/AIDS Gen- eralized Epidemics Using Incremental Mixture Importance Sampling." Technical Report 560, Department of Statistics, University of Washington.

See Also

hp.bm.imis

Examples

1
2
3
4
5
6
7
8
#Generate a prior and numbers of death and persons at risk#
## Not run: q0 <- prior.form()
lx <- c(1974, 1906, 1860, 1844, 1834, 1823, 1793, 1700, 1549, 1361, 
1181, 1025, 870, 721, 571, 450, 344, 256, 142, 79, 41, 8)
dx <- c(68, 47, 16, 10, 13, 29, 92, 151, 188, 179, 156, 155, 147, 150, 
122, 106, 88, 113, 63, 38, 32, 8)
opt.result <- loop.optim(prior=q0, nrisk=lx, ndeath=dx)
## End(Not run)