Compiling Dakota

For more experienced users, Dakota can be built from source code. Consider this option if:

• You want to customize Dakota with additional packages, e.g., enable QUESO or an experimental feature

• You are installing to a platform not explcitly listed in the binary executable distribution list (including Linux other than RHEL-compatible)

• You have previously licensed DOT, NPSOL, and/or NLPQL and want to add these packages to the Dakota build.

Compiling Dakota from source requires:

• A development environment that has Dakota’s dependencies.

• Obtaining the source code via download or Git.

• Configuring with CMake to (1) set Dakota options as well as external software locations and options, and (2) generate a development environment-specific build tree.

• Building Dakota, i.e. running the development environment’s build process to compile and link the Dakota executable, e.g., using Unix make, Windows Visual Studio, or nmake.

• Running and assessing tests.

Quickstart Build

This simple example shows the Dakota build process (configure, compile, install) on a Unix-like platform with mostly default options:

mkdir build
cd build

# configure the build tree
cmake -D CMAKE_INSTALL_PREFIX=/path/to/dakota-installation /path/to/dakota-source

# compile/link, specifying the number of processors
make -j 4

# run a minimal set of Dakota acceptance tests (these should all pass)
ctest -j 4 -L Accept

# install to the specified prefix
make install

The remaining pages in this section describe the compilation process and options in more detail.

• Setup Development Environment
• Obtain Source Code
• Configure with CMake
• Build and Install Dakota
• Running Dakota Tests
• End-to-end Compilation Examples