| step size in constrained_min? [message #73189] |
Wed, 27 October 2010 12:18 |
rdh
Messages: 6
Registered: October 2009
|
Junior Member |
|
|
All -
Does anyone know what determines the step size used in
constrained_min? I'm attempting to use it to minimize a single
function, and although the variable bounds are wide (~100), it doesn't
seem to want to vary any variable by more than +/-0.000001. What am I
doing wrong?
Here's a snippet:
; reformat scale (bounds) for this optimization function
prec = MACHAR()
ftol = sqrt(prec.EPS)*2.^downsample
intial_guess = [3.3, 10.5, 35.4, 61.23, 78.435]
bounds = [-0.01, 10.0 , 5.25 , 50.2, 10.52, 100.5, 52.3, 142.36,
61.2 , 155.03]
func_bounds = [-1.0e30,1.0e30] ;unbounded (also tried loosely
bounding)
CONSTRAINED_MIN, initial_guess, bounds, func_bounds, 0, 'my_func',
inform, $ NSTOP=5, REPORT=new_dir + 'IDL_constr_min.txt', TITLE='BHC
Spline Pts', $ EPSTOP=ftol
Thanks,
Romy
------------------------------------------------------------ ---
Romy Hanna
Research Scientist Associate
High Resolution X-ray CT Facility
University of Texas, Austin
romy@jsg.utexas.edu
------------------------------------------------------------ ---
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
