|
I check the code in minbleicoptimize. The annotation said it uses the LBFGS to determine the direction, but why the website claims that the algorithm uses nonlinear conjugate gradient? Am I missing something? Attach is the annotation in the source code.
/*
* 1. Calculate search direction D according to L-BFGS algorithm
* using constrained preconditioner to perform inner multiplication.
* 2. Evaluate scaled length of direction D; restart LBFGS if D is zero
* (it may be possible that we found minimum, but it is also possible
* that some constraints need deactivation)
* 3. If D is non-zero, try to use previous scaled step length as initial estimate for new step.
*/
|
Login
Register
FAQ
Search
