OCBA: Low Level OCBA

In bartzbeielstein/SPOT: Sequential Parameter Optimization Toolbox

Description

This computes the Optimal Computing Budget Allocation.

Usage

OCBA(sMean, sVar, n, addBudget)

Arguments

sMean	vector of sample means of candidate solutions.
sVar	vector of sample variances of candidate solutions. Note, that these should be non-zero. If the solutions with lowest or second lowest mean have zero variance, This function will not distribute any additional evaluations, and return a vector of zeros. If any other solution has zero observed variance, the respective solution will be ignored, and never be assigned additional evaluations.
n	number of already performed evaluations for each candidate solutions.
addBudget	the number of additional evaluations (replications) that can be distributed to the candidate solutions.

Value

A vector that specifies how often each solution should be evaluated.

References

Chun-hung Chen and Loo Hay Lee. 2010. Stochastic Simulation Optimization: An Optimal Computing Budget Allocation (1st ed.). World Scientific Publishing Co., Inc., River Edge, NJ, USA.

See Also

repeatsOCBA is based on this function.

Examples

## Simple test
OCBA(1:10,1:10,rep(4,10),3)
## Example from the referenced book:
res <- OCBA(c(1.2,2.1,3.4,4.87,6.05),c(3.3,2.0,4.5,5.3,6.9),c(12,6,5,5,4),100)
##expected result: 
res == c(48,38,11,2,1)

bartzbeielstein/SPOT documentation built on June 13, 2020, 5:58 p.m.