module func More...

Functions/Subroutines
subroutine	deltabox (npar, niter, nind, deltaconst, reduc, arrmain, delta0, delta)
	Updates the size of the hyper-boxes.
real	med (stat, N)
	Computes the mean value.
real	sigm (stat, N)
	Computes the standard deviation.
subroutine	spawn (npar, gensize, nmig, ngaussdev, delta, qpar, qbounds, qfam)
	Asexual reproduction routine.
subroutine	criterion (npar, q, delta, beta, ncrit)
	Evaluates convergence criterion.
subroutine	randdata (ndata, yobs, ysigma)
	Generates new realization from experimental data.
real	gaussdev (idum)
	Generates pseudo-random (Gaussian distributtion)
subroutine	pickbest (npar, nind, nparents, arrmain, fitness, arrpars)
	Select the fittest individuals (nparents number of future parents) from the entire population. Based in the smallest value of the merit function (smaller number=fittest, e.g. $\chi^2$ )
subroutine	gaussian (ngaus, i, arrmain)

Detailed Description

module func

This module contains global functions and subroutines used by AGA-V1

Function/Subroutine Documentation

subroutine func::criterion	(	integer,intent(in)	npar,
		real,dimension(npar),intent(in)	q,
		real,dimension(npar),intent(in)	delta,
		real,intent(in)	beta,
		integer,intent(out)	ncrit
	)

Evaluates convergence criterion.

Parameters:

[in]	npar	: Number of free parameters
[in]	q(npar)	: Individual data
[in]	delta(npar)	: Window size of the individual
[in]	beta	: Tolerance value
[out]	ncrit	: Return value=1 if criterion met, otherwise is not modified

Definition at line 201 of file modules.f90.

subroutine func::deltabox	(	integer,intent(in)	npar,
		integer,intent(in)	niter,
		integer,intent(in)	nind,
		integer,intent(in)	deltaconst,
		real,intent(in)	reduc,
		real,dimension(npar,nind),intent(in)	arrmain,
		real,dimension(npar),intent(in)	delta0,
		real,dimension(npar),intent(out)	delta
	)

Updates the size of the hyper-boxes.

Parameters:

[in]	npar	: number of free parameters (parameter space dimension, denoted by "n" in the paper)
[in]	niter	: number of iteration counter
[in]	nind	: number of individials in the entire population
[in]	deltaconst	: if equal to 1 the reduction of the hyperboxes is at a constant rate (see reduc), otherwise hyperboxes are reduced with the standard deviation of each parameter.
[in]	reduc	: reducing factor, if a constant speed is chosen the hyper-box is reduced according to: $\Delta_{i,niter}=\Delta_{i,niter-1}*reduc^{niter}$
[in]	arrmain(npar,nind)	: Entire population of "nind" individuals, each with "npar" parameters.
[in]	delta0(npar)	: Original size of the hyperboxes, array of $\Delta_{i,1}$ .
[out]	delta(npar)	: array of $\Delta_{i,niter}$ (niter is constant, thus is only a vector w/npar entries). $\Delta_{i,niter}$ is the size of the hyper-box corresponding to the "i-th" parameter

Definition at line 75 of file modules.f90.

References sigm().

Referenced by agav1::Algorithm().

real func::gaussdev ( integer idum )

Generates pseudo-random (Gaussian distributtion)

Modified from "Numerical Recipes in Fortran 77, second ed." (Press et al., Cambridge University Press)

Parameters:

[in] idum : dummy variable (required)

Returns:: Pseudo-random number, drawn from a normal distribution (Gaussian with mean=0 and standard deviation=1)

Definition at line 250 of file modules.f90.

Referenced by spawn().

subroutine func::gaussian	(	integer,intent(in)	ngaus,
		integer,intent(in)	i,
		real,dimension(:,:),intent(inout)	arrmain
	)

Todo:: Preguntarle a ary que chingaos hace esta rutina

Definition at line 410 of file modules.f90.

Referenced by agav1::Algorithm().

real func::med	(	real,dimension(n),intent(in)	stat,
		integer,intent(in)	N
	)

Computes the mean value.

Parameters:

[in]	N	: number of data points
[in]	stat(N)	: vector from which the mean is to be obtained

Returns:: Arithmetic mean value of stat(N)

Definition at line 101 of file modules.f90.

Referenced by agav1::Algorithm(), and sigm().

subroutine func::pickbest	(	integer,intent(in)	npar,
		integer,intent(in)	nind,
		integer,intent(in)	nparents,
		real,dimension(npar,nind),intent(in)	arrmain,
		real,dimension(nind),intent(inout)	fitness,
		real,dimension(npar,nparents),intent(out)	arrpars
	)

Select the fittest individuals (nparents number of future parents) from the entire population. Based in the smallest value of the merit function (smaller number=fittest, e.g. $\chi^2$ )

Parameters:

[in]	npar	: numbr of free parameters
[in]	nind	: number of individuals in the entire population
[in]	nparents,:	number of parents, total of fittest individuals to survive for the next generation
[in]	arrmain(npar,nind)	: entire population
[in,out]	fitness(nind)	: fitness level of each individual, given by the merit function provided b the user in "tarfunc.f90".
[out]	arrpars(npar,nparents)	: array of the fittest individuals, which will be parents in the following generation.

Definition at line 293 of file modules.f90.

References indexx().

Referenced by agav1::Algorithm().

subroutine func::randdata	(	integer,intent(in)	ndata,
		real,dimension(ndata),intent(inout)	yobs,
		real,dimension(ndata),intent(in)	ysigma
	)

Generates new realization from experimental data.

Parameters:

[in]	ndata	: number of experimental data points
[in,out]	yobs(ndata)	: experimental data
[in]	ysigma(ndata)	: associated uncertainties of yobs

Todo:: Hay que implementar dispersion Gaussiana y checar por que no se guarda yobs0

Definition at line 226 of file modules.f90.

Referenced by agav1::Algorithm().

real func::sigm	(	real,dimension(n),intent(in)	stat,
		integer,intent(in)	N
	)

Computes the standard deviation.

Parameters:

[in]	N	: number of data points
[in]	stat(N)	: vector from which the standard deviaiton is to be obtained

Returns:: standard deviation of stat(N)

Definition at line 123 of file modules.f90.

References med().

Referenced by agav1::Algorithm(), and deltabox().

subroutine func::spawn	(	integer,intent(in)	npar,
		integer,intent(in)	gensize,
		integer,intent(in)	nmig,
		integer,intent(in)	ngaussdev,
		real,dimension(npar),intent(in)	delta,
		real,dimension(npar),intent(in)	qpar,
		real,dimension(2,npar),intent(in)	qbounds,
		real,dimension(npar,gensize)	qfam
	)

Asexual reproduction routine.

Parameters:

[in]	npar	: number of free parameters
[in]	gensize	: size of the new generation from each parent. Since the parent is kept gensize= num of sons + 1
[in]	nmig	: turns on/off migration. If its value is 0, migration is disabled, otherwise is enabled
[in]	ngaussdev	: if value equal to 1, offsprinf is obtained with a Gaussian distribution, otherwise they are obtained with uniform random deviates.
[in]	delta(npar)	: Size of the parent hyper-box.
[in]	qpar(npar)	: parent parameters (its location in the npar-dimensional parameter space
[in]	qbounds(npar)	: bounds of the initial universe
[out]	qfam(npar,gensize)	: A new family (progeny+parent)

Definition at line 157 of file modules.f90.

References gaussdev().

Referenced by agav1::Algorithm().

Functions/Subroutines

Detailed Description

Function/Subroutine Documentation