LikelihoodAscent.optimize

Performs optimization.

void optimize(bool delegate(T likelihoodValuePrev, T likelihoodValue, in T[] weightsPrev, in T[] weights) tolerance, bool delegate(T a, T b) @(nogc) nothrow findRootTolerance)
interface LikelihoodAscent(T)
void
optimize
(
scope bool delegate
(
T likelihoodValuePrev
,
T likelihoodValue
,
in T[] weightsPrev
,
in T[] weights
)
tolerance
,
scope bool delegate
(
T a
,
T b
)
@nogc nothrow findRootTolerance = null
)
if (
isFloatingPoint!T
)
void optimize(bool delegate(T likelihoodValuePrev, T likelihoodValue) tolerance, bool delegate(T a, T b) @(nogc) nothrow findRootTolerance)

Parameters

tolerance bool delegate ( T likelihoodValuePrev , T likelihoodValue , in T[] weightsPrev , in T[] weights ): Defines an early termination condition. Receives the current and previous versions of normalized log-likelihood and weights. The delegate must return true when likelihood and weights are acceptable.
findRootTolerance bool delegate ( T a , T b ) @nogc nothrow: Tolerance for inner optimization.