==== MATLAB serial job ====

This approach has to be used if you do not need any interactivity and you do use MATLAB on the HPC cluster.
In this case MATLAB does not need to be installed on your local computer.

=== Example of execution of a MATLAB serial program ===

<code>
cp -p /hpc/share/samples/matlab/pi_greco.m .
</code>

Script ''pi_greco.m'':

<code matlab pi_greco.m>
tic
N_int=100;
N_int_0=1e2;
N_int_f=1e7;

passo=ceil((N_int_f-N_int_0)/N_int);
res=zeros(N_int,1);
i=1;
for N=N_int_0:passo:N_int_f
    Z=rand(2,N);
    res(i)=sum((Z(1,:)-0.5).^2+(Z(2,:)-0.5).^2<0.5^2)/(N*0.5^2);
    i=i+1;
end

disp(res(end))
toc
</code>

Once you have written your MATLAB code you need to supply a shell script (that is a file with extension ''.sh'') where you specify the resources you need to run your MATLAB code on the HPC servers.

[[https://researchcomputing.princeton.edu/support/knowledge-base/scaling-analysis| See this small introduction on resource management]]

See the example below:

Script ''my_matlab_script.sh'':

<code bash my_matlab_script.sh>
#!/bin/sh
#SBATCH --job-name=matlab
#SBATCH --output=%x.o%j
#SBATCH --error=%x.e%j
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=1
#SBATCH --partition=vrt
#SBATCH --mem=8G
#SBATCH --time=00:30:00
#SBATCH --account=<account>

module load matlab/R2022a

matlab -nodisplay -r pi_greco
</code>

To submit the shell script ''my_matlab_script.sh'' and start the computation, type:

<code>
sbatch my_matlab_script.sh
</code>

FIXME add a section for retrieving and display the output

[[calcoloscientifico:userguide:matlab|go back one level]]