optimize.multiplier {swfdr} | R Documentation |
This implementation relies on the knowledge that a larger lagrange multiplier leads to lower trace(X+lambda*omega).
optimize.multiplier(X, omega, target = 5, interval = c(1e-06, 1, 2), values = c(NA, NA, NA), tol = 1e-04)
X |
matrix |
omega |
matrix |
target |
numeric, target for trace(X+lambda*omega) |
interval |
numeric vector of length 2, current range for lambda (internal use) |
values |
numeric vector of length 2, current target estimates (internal use) |
tol |
numeric, numerical tolerance, does not need to be very small |
The implementation starts in an interval (0, 1, 2) and the expands the interval or zooms in to find an
value of lagrange multiplier that bring criterion close to its target