Hi users
I would like your help with converting this Matlab code to Fortran code.Im not familiar with Matlab.

thanks in advance.

[]
% make sure that the array with data that you have is a long one, correlation functions are need a lot of trajectories to become smooth.
file = fopen(DATA....);
Pxy = fscanf(file,'%f',[1 inf]);

% make an array with the final correlation function, choose NTimeSteps to be much smaller than the length of Pxy
NTimeSteps = round(tmax/dt);
ACF = zeros(1,NTimeSteps);

% start various evaluations of the correlation function from different (uncorrelated) times.
Shift = round(0.5/dt); % where '0.5' is approximately the time is takes for starting points to become uncorrelated, depending on your system
Ntraj = round((length(Pxy) - NTimeSteps)/Shift);

for j = 1 : Ntraj
for i = 1 : NTimeSteps
ACF(i) = ACF(i) + Pxy(1 + (j-1)*shift)*Pxy(i + (j-1)*shift); % (j-1) because Matlab won't allow starting a loop with j=0. But you might use a different language.
end
end

% ensemble averaging
ACF = ACF / Ntraj;

% integral over the correlation function
intACf = cumsum(ACF) / (V*k_B*T) * dt;

cumsum is not the most elegant integration method, you can also use simpson rule or something like that. I believe that would be 'quad' in Matlab.


[/]

Homework?

Hi astrogeek
its not homework! any help will be appreciated.

Ok, I guess the place to start would be to ask what you have done so far.

I do not know matlab, but the expressions look understandable enough (although they do not look to be complete). Maybe it would help to know the source and target contexts. Where is this used?

I remember someone I know doing this conversion for some reason. Take a look at these two links

1 - http://sourceforge.net/projects/matlab2fmex/
2 - http://emmanuel.branlard.free.fr/con...ode-to-fortran

The second link is one where there is a matlab script that converts a MATLAB code to FORTRAN.

Of course, it is necessary that you understand the MATLAB code completely in order to be able to use the Fortran converted one.

This is what I have done so far.What I dont understand is matlab functions and syntax


[]
program integrate_dataset
implicit none
real:: pxy,NTtimesteps,ACF,tmax,dt,Ntraj,intACF
integer :: i,j

!% make sure that the array with data that you have is a long one,
!correlation functions are need a lot of trajectories to become smooth.

open(7,file='DATA_input.txt')
!open (10,file = fname,form='formatted')


Pxy = fscanf(file,'%f',[1 inf])

!% make an array with the final correlation function,
!choose NTimeSteps to be much smaller than the length of Pxy

NTimeSteps = round(tmax/dt)
ACF = zeros(1,NTimeSteps)

!% start various evaluations of the correlation function from different (uncorrelated) times.

!Shift = round(0.5/dt)

!% where '0.5' is approximately the time is takes for starting points
!to become uncorrelated, depending on your system

Shift = round(0.5/dt)
Ntraj = round((length(Pxy) - NTimeSteps)/Shift)

do j=1,Ntraj
do i=1,NTimesteps
!for j = 1 : Ntraj

!for i = 1 : NTimeSteps

!% (j-1) because Matlab won't allow starting a loop with j=0. But you might use a different language.
ACF(i) = ACF(i) + Pxy(1 + (j-1)*shift)*Pxy(i + (j-1)*shift)

enddo
enddo

!% ensemble averaging
ACF = ACF / Ntraj

!% integral over the correlation function

intACf = trap(ACF) / (V*k_B*T) * dt
end


[/code]

I have a matlab code that needs to use this command

Find(A>3,5)

the equivalent if this command in fortran is which command?


Thank you alls