cp_lbfgs_optimizer_gopt Module Reference

routines that optimize a functional using the limited memory bfgs quasi-newton method. The process set up so that a master runs the real optimizer and the others help then to calculate the objective function. The arguments for the objective function are physically present in every processor (nedeed in the actual implementation of pao). In the future tha arguments themselves could be distributed. More...

Data Types
type	cp_lbfgs_opt_gopt_type
	info for the optimizer (see the description of this module) More...

Functions/Subroutines
subroutine, public	cp_opt_gopt_create (optimizer, para_env, obj_funct, x0, m, print_every, wanted_relative_f_delta, wanted_projected_gradient, lower_bound, upper_bound, kind_of_bound, master, max_f_per_iter, trust_radius)
	initializes the optimizer
subroutine, public	cp_opt_gopt_release (optimizer)
	releases the optimizer (see doc/ReferenceCounting.html)
logical function, public	cp_opt_gopt_next (optimizer, n_iter, f, last_f, projected_gradient, converged, geo_section, force_env, gopt_param, spgr)
	goes to the next optimal point (after an optimizer iteration) returns true if converged
subroutine, public	cp_opt_gopt_stop (optimizer)
	stops the optimization

Detailed Description

routines that optimize a functional using the limited memory bfgs quasi-newton method. The process set up so that a master runs the real optimizer and the others help then to calculate the objective function. The arguments for the objective function are physically present in every processor (nedeed in the actual implementation of pao). In the future tha arguments themselves could be distributed.

History

09.2003 globenv->para_env, retain/release, better parallel behaviour 01.2020 Space Group Symmetry introduced by Pierre-André Cazade [pcazade]

Author

Fawzi Mohamed

Version

2.2002

Function/Subroutine Documentation

cp_opt_gopt_create()

subroutine, public cp_lbfgs_optimizer_gopt::cp_opt_gopt_create	(	type(cp_lbfgs_opt_gopt_type), intent(out)	optimizer,
		type(mp_para_env_type), pointer	para_env,
		type(gopt_f_type), pointer	obj_funct,
		real(kind=dp), dimension(:), intent(in)	x0,
		integer, intent(in), optional	m,
		integer, intent(in), optional	print_every,
		real(kind=dp), intent(in), optional	wanted_relative_f_delta,
		real(kind=dp), intent(in), optional	wanted_projected_gradient,
		real(kind=dp), dimension(size(x0)), intent(in), optional	lower_bound,
		real(kind=dp), dimension(size(x0)), intent(in), optional	upper_bound,
		integer, dimension(size(x0)), intent(in), optional	kind_of_bound,
		integer, intent(in), optional	master,
		integer, intent(in), optional	max_f_per_iter,
		real(kind=dp), intent(in), optional	trust_radius
	)

initializes the optimizer

Parameters

optimizer	...
para_env	...
obj_funct	...
x0	...
m	...
print_every	...
wanted_relative_f_delta	...
wanted_projected_gradient	...
lower_bound	...
upper_bound	...
kind_of_bound	...
master	...
max_f_per_iter	...
trust_radius	...

History

02.2002 created [fawzi] 09.2003 refactored (retain/release,para_env) [fawzi]

Author

Fawzi Mohamed

Note

redirects the lbfgs output the the default unit

Definition at line 202 of file cp_lbfgs_optimizer_gopt.F.

Here is the call graph for this function:

Here is the caller graph for this function:

cp_opt_gopt_release()

subroutine, public cp_lbfgs_optimizer_gopt::cp_opt_gopt_release

(

type(cp_lbfgs_opt_gopt_type), intent(inout)

optimizer

)

releases the optimizer (see doc/ReferenceCounting.html)

Parameters

optimizer

the object that should be released

History

02.2002 created [fawzi] 09.2003 dealloc_ref->release [fawzi]

Author

Fawzi Mohamed

Definition at line 303 of file cp_lbfgs_optimizer_gopt.F.

Here is the call graph for this function:

Here is the caller graph for this function:

cp_opt_gopt_next()

logical function, public cp_lbfgs_optimizer_gopt::cp_opt_gopt_next	(	type(cp_lbfgs_opt_gopt_type), intent(inout)	optimizer,
		integer, intent(out), optional	n_iter,
		real(kind=dp), intent(out), optional	f,
		real(kind=dp), intent(out), optional	last_f,
		real(kind=dp), intent(out), optional	projected_gradient,
		logical, intent(out)	converged,
		type(section_vals_type), pointer	geo_section,
		type(force_env_type), pointer	force_env,
		type(gopt_param_type), pointer	gopt_param,
		type(spgr_type), optional, pointer	spgr
	)

goes to the next optimal point (after an optimizer iteration) returns true if converged

Parameters

optimizer	the optimizer that goes to the next point
n_iter	...
f	...
last_f	...
projected_gradient	...
converged	...
geo_section	...
force_env	...
gopt_param	...
spgr	...

Returns

...

History

01.2020 modified [pcazade]

Author

Fawzi Mohamed

Version

2.2002

Note

if you deactivate convergence control it returns never false

Definition at line 770 of file cp_lbfgs_optimizer_gopt.F.

Here is the caller graph for this function:

cp_opt_gopt_stop()

subroutine, public cp_lbfgs_optimizer_gopt::cp_opt_gopt_stop

(

type(cp_lbfgs_opt_gopt_type), intent(inout)

optimizer

)

stops the optimization

Parameters

optimizer

...

History

none

Author

Fawzi Mohamed

Version

2.2002

Definition at line 800 of file cp_lbfgs_optimizer_gopt.F.

Here is the call graph for this function:

Here is the caller graph for this function: