CasADi

CasAdi is an open-source tool for nonlinear optimization and algorithmic differentiation. It facilitates rapid — yet efficient — implementation of different methods for numerical optimal control, both in an offline context and for nonlinear model predictive control (NMPC).

CasADi for MATLAB

Using CasADi for MATLAB interactive
(Graphical User Interface)

Use your browser to connect to Umbrella On Demand and start MATLAB session that includes CasADi.

Using CasADi for MATLAB in SLURM batch jobs
(Command Line Interface)

A CasADi module for MATLAB needs to be loaded so MATLAB can use CasADi.

#!/bin/bash
#SBATCH --job-name=test_casadi
#SBATCH --output=test_casadi-%j.log
#SBATCH --partition=tue.default.q
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem-per-cpu=2gb
#SBATCH --time=00:05:00

module purge
module load CasADi/3.7.2-linux64-matlab
module load MATLAB/2025b

matlab -batch test_casadi

test_casadi.m

import casadi.*

% Create scalar/matrix symbols
x = MX.sym('x',5);

% Compose into expressions
y = norm(x,2);

% Sensitivity of expression -> new expression
grad_y = gradient(y,x)

% Create a Function to evaluate expression
f = Function('f',{x},{grad_y});

% Evaluate numerically
grad_y_num = f([1;2;3;4;5])

Using CasADi for Python in SLURM batch jobs

When loading the CasADi module the correct version of Python and it's dependencies are automaticaly loaded.

#!/bin/bash
#SBATCH --job-name=test_casadi
#SBATCH --output=test_casadi-%j.log
#SBATCH --partition=tue.default.q
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem-per-cpu=2gb
#SBATCH --time=00:05:00

module purge
module load CasADi/3.7.2-foss-2024a

python test_casadi.py

test_casadi.py

from casadi import *

# Create scalar/matrix symbols
x = MX.sym('x',5)

# Compose into expressions
y = norm_2(x)

# Sensitivity of expression -> new expression
grad_y = gradient(y,x);

# Create a Function to evaluate expression
f = Function('f',[x],[grad_y])

# Evaluate numerically
grad_y_num = f([1,2,3,4,5]);

print(grad_y_num);

Including and using the CoinHSL solvers

Load the module for CoinHSL

MATLAB

ml purge
ml CoinHSL/2024.05.15-foss-2024a
ml CasADi/3.7.2-linux64-matlab

Python

ml purge
ml CoinHSL/2024.05.15-foss-2024a
ml CasADi/3.7.2-foss-2024a

Using solvers (e.q. MUMPS) that request MPI in Python

If the following ERROR is thrown by CasADi for python "The MPI_Comm_f2c() function was called before MPI_INIT was invoked." this indicates a missing MPI module. Some solvers are compiled to use MPI and need the mpi4py module to be loaded and the python package mpy4py.MPI needs to be imported.

ml purge
ml mpi4py/4.0.1-gompi-2024a
ml CasADi/3.7.2-foss-2024a

from casadi import *
from mpi4py import MPI

# Symbols/expressions
x = MX.sym('x')
y = MX.sym('y')
z = MX.sym('z')
f = x**2+100*z**2
g = z+(1-x)**2-y

nlp = {}                 # NLP declaration
nlp['x']= vertcat(x,y,z) # decision vars
nlp['f'] = f             # objective
nlp['g'] = g             # constraints

# Create solver instance
F = nlpsol('F','ipopt',nlp);

# Solve the problem using a guess
F(x0=[2.5,3.0,0.75],ubg=0,lbg=0)