Dynamically Dimensioned Search (DDS) More...

Functions/Subroutines
real(dp) function, dimension(size(pini)), public	dds (eval, obj_func, pini, prange, r, seed, maxiter, maxit, mask, tmp_file, funcbest, history)
	DDS.

real(dp) function, dimension(size(pini)), public	mdds (eval, obj_func, pini, prange, seed, maxiter, maxit, mask, tmp_file, funcbest, history)
	MDDS.

subroutine	neigh_value (x_cur, x_min, x_max, r, new_value)

Detailed Description

Dynamically Dimensioned Search (DDS)

This module provides routines for Dynamically Dimensioned Search (DDS) of Tolson and Shoemaker (2007). It searches the minimum or maximum of a user-specified function, using an n-dimensional continuous global optimization algorithm (DDS).

Authors: Bryan Tolson, modified by Rohini Kumar, Matthias Cuntz and Juliane Mai.

Date: Jul 2012

Copyright: Copyright 2005-2024, the CHS Developers, Sabine Attinger: All rights reserved. FORCES is released under the LGPLv3+ license

The UFZ (CHS) FORCES code is free software. You can redistribute it and/or modify it under the terms of the GNU General Public License as published by the free Software Foundation either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You received a copy of the GNU Lesser General Public License along with the UFZ CHS FORCES code. It can be found in the files COPYING and COPYING.LESSER provided with this software. The complete GNU license text can also be found at http://www.gnu.org/licenses/.

Function/Subroutine Documentation

◆ dds()

real(dp) function, dimension(size(pini)), public mo_dds::dds	(	procedure(eval_interface), intent(in), pointer	eval,
		procedure(objective_interface), intent(in), pointer	obj_func,
		real(dp), dimension(:), intent(in)	pini,
		real(dp), dimension(:, :), intent(in)	prange,
		real(dp), intent(in), optional	r,
		integer(i8), intent(in), optional	seed,
		integer(i8), intent(in), optional	maxiter,
		logical, intent(in), optional	maxit,
		logical, dimension(:), intent(in), optional	mask,
		character(len = *), intent(in), optional	tmp_file,
		real(dp), intent(out), optional	funcbest,
		real(dp), dimension(:), intent(out), optional, allocatable	history
	)

DDS.

Searches Minimum or Maximum of a user-specified function using Dynamically Dimensioned Search (DDS).
DDS is an n-dimensional continuous global optimization algorithm. It is coded as a minimizer but one can give maxit=True in a maximization problem, so that the algorithm minimizes the negative of the objective function F=(-1*F).

The function to be minimized is the first argument of DDS and must be defined as

function func(p)
  use mo_kind, only: dp
  implicit none
  real(dp), dimension(:), intent(in) :: p
  real(dp) :: func
end function func

Example

dv_range(:,1) = (/ -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0 /)
dv_range(:,2) = (/ 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0 /)
dv_ini        = (/ -.226265e+01, -.130187e+01, -.151219e+01, 0.133983e+00, 0.988159e+00, &
                   -.495074e+01, -.126574e+02, 0.572684e+00, 0.303864e+01, 0.343031e+01 /)
dv_opt = dds(griewank, dv_ini, dv_range)

See also example in test directory.

Literature

Tolson, B. A., and C. A. Shoemaker (2007) Dynamically dimensioned search algorithm for computationally efficient watershed model calibration, Water Resour. Res., 43, W01413, doi:10.1029/2005WR004723.

Parameters

[in]	real(dp) :: obj_func(p)	Function on which to search the minimum
[in]	real(dp) :: pini(:)	inital value of decision variables
[in]	real(dp) :: prange(size(pini),2)	Min/max range of decision variables
[in]	real(dp), optional :: r	DDS perturbation parameter (default: 0.2)
[in]	integer(i8), optional :: seed	User seed to initialise the random number generator (default: None)
[in]	integer(i8), optional :: maxiter	Maximum number of iteration or function evaluation (default: 1000)
[in]	logical, optional :: maxit	Maximization (.True.) or minimization (.False.) of function (default: .False.)
[in]	logical, optional :: mask(size(pini))	parameter to be optimized (true or false) (default: .True.)
[in]	character(len=*) , optional :: tmp_file	file with temporal output
[out]	real(dp), optional :: funcbest	the best value of the function.
[out]	real(dp), optional, allocatable :: history(:)	the history of best function values, history(maxiter)=funcbest allocatable only to be in correspondance with other optimization routines

Return values

real(dp) :: DDS The parameters of the point which is estimated to minimize the function.

Author: Bryan Tolson

Date: Feb 2007

Author: Rohini Kumar

Date: Feb 2008

Author: Matthias Cuntz & Juliane Mai

Date

Jul 2012

module

Author: Juliane Mai

Date

Aug 2012

optional argument funcbest added

Nov 2012

masked parameter

Dec 2012

history output

Author: Pallav Kumar Shrestha

Date

Jun 2019

Added "dds_results.out.current" output file with current iteration (non-updated) parameters

Definition at line 109 of file mo_dds.F90.

References dds(), mo_kind::dp, mo_kind::i4, and mo_kind::i8.

Referenced by dds().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ mdds()

real(dp) function, dimension(size(pini)), public mo_dds::mdds	(	procedure(eval_interface), intent(in), pointer	eval,
		procedure(objective_interface), intent(in), pointer	obj_func,
		real(dp), dimension(:), intent(in)	pini,
		real(dp), dimension(:, :), intent(in)	prange,
		integer(i8), intent(in), optional	seed,
		integer(i8), intent(in), optional	maxiter,
		logical, intent(in), optional	maxit,
		logical, dimension(:), intent(in), optional	mask,
		character(len = *), intent(in), optional	tmp_file,
		real(dp), intent(out), optional	funcbest,
		real(dp), dimension(:), intent(out), optional, allocatable	history
	)

MDDS.

Searches Minimum or Maximum of a user-specified function using the Modified Dynamically Dimensioned Search (DDS).
DDS is an n-dimensional continuous global optimization algorithm. It is coded as a minimizer but one can give maxit=True in a maximization problem, so that the algorithm minimizes the negative of the objective function F=(-1*F).
The function to be minimized is the first argument of DDS and must be defined as

function func(p)
  use mo_kind, only: dp
  implicit none
  real(dp), dimension(:), intent(in) :: p
  real(dp) :: func
end function func

MDDS extents normal DDS by a continuous reduction of the DDS pertubation parameter r from 0.3 to 0.05, and by allowing a larger function value with a certain probablity.

Parameters

[in]	real(dp) :: obj_func(p)	Function on which to search the minimum
[in]	real(dp) :: pini(:)	inital value of decision variables
[in]	real(dp) :: prange(size(pini),2)	Min/max range of decision variables
[in]	integer(i8), optional :: seed	User seed to initialise the random number generator (default: None)
[in]	integer(i8), optional :: maxiter	Maximum number of iteration or function evaluation (default: 1000)
[in]	logical, optional :: maxit	Maximization (.True.) or minimization (.False.) of function (default: .False.)
[in]	logical, optional :: mask(size(pini))	parameter to be optimized (true or false) (default: .True.)
[in]	character(len=*) , optional :: tmp_file	file with temporal output
[out]	real(dp), optional :: funcbest	the best value of the function.
[out]	real(dp), optional, allocatable :: history(:)	the history of best function values, history(maxiter)=funcbest allocatable only to be in correspondance with other optimization routines

Returns: real(dp) :: MDDS — The parameters of the point which is estimated to minimize the function.

Restrictions

None.

Example

dv_range(:,1) = (/ -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0, -600.0 /)
dv_range(:,2) = (/ 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0, 600.0 /)
dv_ini        = (/ -.226265E+01, -.130187E+01, -.151219E+01, 0.133983E+00, 0.988159E+00, &
                   -.495074E+01, -.126574E+02, 0.572684E+00, 0.303864E+01, 0.343031E+01 /)
dv_opt = MDDS(griewank, dv_ini, dv_range)

Literature

Tolson, B. A., and C. A. Shoemaker (2007), Dynamically dimensioned search algorithm for computationally efficient watershed model calibration, Water Resour. Res., 43, W01413, doi:10.1029/2005WR004723.
Huang X-L and Xiong J (2010), Parameter Optimization of Multi-tank Model with Modified Dynamically Dimensioned Search Algorithm, Proceedings of the Third International Symposium on Computer Science and Computational Technology(ISCSCT ''10), Jiaozuo, P. R. China, 14-15 August 2010, pp. 283-288

Author
Written Matthias Cuntz and Juliane Mai

Date
Aug 2012

Definition at line 422 of file mo_dds.F90.

References mo_kind::dp, mo_kind::i4, mo_kind::i8, and mdds().

Referenced by mdds().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ neigh_value()

subroutine mo_dds::neigh_value	(	real(dp), intent(in)	x_cur,
		real(dp), intent(in)	x_min,
		real(dp), intent(in)	x_max,
		real(dp), intent(in)	r,
		real(dp), intent(out)	new_value
	)

private

Definition at line 638 of file mo_dds.F90.

Functions/Subroutines

Detailed Description

Function/Subroutine Documentation

◆ dds()

◆ mdds()

Restrictions

Example

Literature

◆ neigh_value()