mirror: Generates weights using mirror algorithm. Fulfills equality...

In davidkane9/kmatching: Creates random vectors fulfilling constraints

Description

To project a vector u onto vector v we use the operation proj(u)_v = v * fracu*vv*v. We do this several times with an operation that looks like u * u

Usage

mirror(Amat, x0, n, verbose = FALSE, numjump = 20,
  includeInfeasible = FALSE)

Arguments

Amat	This is the matrix of the equality constraint coefficients
x0	An original solution to the constraints
n	Number of random solutions to output
verbose	Give verbose output describing the progress of the function
numjump	The number of jumps to scatter around the direction given by the difference from zero
includeInfeasible	TRUE to include all "bad" points in the output

Value

Gives back a matrix with 'n' columns corresponding to 'n' positive solutions to Ax = b. If includeInfeasible is TRUE then infeasible points will also be included in the output, so there will be some negatives. This is to be able to track the walk of mirror.

Author(s)

Mike Flynn <[e-mail]>

References

Van Den Meershe, Karel, Karline Soetaert, and Dick Van Oevelen. "Xsample(): An R Function for Sampling Linear Inverse Problems." Journal of Statistical Software 30 (2009): 1–15. http://cran.cermin.lipi.go.id/web/packages/limSolve/vignettes/xsample.pdf

Examples

Amat <- matrix(1, ncol = 3, nrow = 1)
x0 <- c(.3, .3, .4)
mirror(Amat, x0, 1)

davidkane9/kmatching documentation built on May 15, 2019, 1:14 a.m.