Here’s how you would write a CMake file to compile some C++ code using the Eigen linear algebra library. First, an example C++ source file.

1
2
3
4
5
6
7
8
9
10
11
12
13
#include <Eigen/Dense>

#include <iostream>

int main()
{
    Eigen::MatrixXd m = Eigen::MatrixXd::Random( 5, 5 );
    std::cout << "Original matrix: \n";
    std::cout << m;

    std::cout << "Transpose: \n";
    std::cout << m.transpose();
}

Here’s the corresponding CMake file.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
# Set minimum required version
cmake_minimum_required( VERSION 3.9.1 )

# Name the project
project( C++-Eigen-Example )

# Find the Eigen config file and use it
find_package( Eigen3 3.3.4 REQUIRED CONFIG )

# Add source code to the manifest
set( CPP_SOURCES main.cpp )

# Build executable
add_executable( main ${CPP_SOURCES} )
target_link_libraries( main Eigen3::Eigen )

Note the use of the find_package call with the CONFIG (or equivalently, NO_MODULE) option. This causes CMake to locate and use a config file called Eigen3Config.cmake which it will end up finding in /usr/share/eigen3/cmake/. That file will define a target called Eigen3::Eigen containing a set of instructions to include the Eigen headers in the main target. “Linking” this library to the executable will cause the headers to be included during compilation, making everything work.

By the way, this works for Eigen versions 3 and above, using CMake versions 3 and above. I’ve used Eigen 3.3.4 and CMake 3.9.1 above, just because they are the current versions at the time of writing.