
ACM Collected Algorithms: DCUHRE and DQUAD
This project provides a CMake framework, that automatically downloads the two Fortran integrator libraries DCUHRE and DQUAD from the ACM Collected Algorithms (https://calgo.acm.org/), converts them to C and builds them.
Prerequisites
What you need:
- Git (optional) for getting the code and contributing to the project
- CMake and Ninja for building the project
- C compiler
- GCC for Linux (https://gcc.gnu.org/)
- Clang for Mac OS X (https://clang.llvm.org/)
- Cygwin + GCC for Windows (https://www.cygwin.com/)
- NetlibF2C (https://gitlab.aei.uni-hannover.de/netlib/netlib-f2c)
- LoadStaticSharedTargets (https://gitlab.aei.uni-hannover.de/cmake/loadstaticsharedtargets)
Linux (Ubuntu)
Use your distribution's package manager to install the necessary packages:
$ sudo apt-get install cmake ninja-build gcc
Mac OS X
You will need one of the community package managers for Mac OS X: Homebrew or MacPorts. For installing one of these, please refer to their respective installation instructions.
Homebrew
$ brew install cmake ninja llvm
MacPorts
$ sudo port -v install cmake ninja clang-<XX>
where you have to replace <XX>
with the latest stable version of Clang.
You can discover all available versions of Clang like this:
$ port search --name --line --regex '^clang-?\d*\.?\d*$'
Windows
The easiest thing to do is using the Windows Subsystem for Linux (WSL) and follow the Linux instructions above.
Otherwise, install Git for Windows for version
control and cygwin, which is a large
collection of GNU and Open Source tools which provide functionality similar to a
Linux distribution on Windows. During the cygwin
installation you'll be
prompted to add additional packages. Search and select the following:
-
cmake
,ninja
andgcc-core
for the compiler and the build system
After cygwin
finishes installing, use the cygwin terminal to start the build
process.
How to compile
-
Clone this git repository or download the source code archive and unpack it to an arbitrary directory (e.g.
calgo-integrators
). -
Go to this directory and type
cp ./CMakeUserPresets.template.json ./CMakeUserPresets.json
. -
Next, type
cmake --workflow --list-presets
. A list of available build workflows will be shown to you. -
For configuring, building and installing the project, type
cmake --workflow --preset user-shared-release-install
. This will try to populate the./build
directory with a CMake configuration tree, execute the build and install the project. By default, the configured installation directory is./install
.But it will fail due to unaccepted licenses. Study the license text and proceed to the next step.
-
Open
./CMakeUserPresets.json
in your favorite editor and set"NetlibF2C_LICENSE_ACCEPTED": true
in the first configure preset"user-config"
. Repeat step 4. It'll fail again with the next license. Study the license text and proceed to set"CalgoIntegrators_LICENSE_ACCEPTED": true
. -
Repeat step 4. Configuring, building and installing should be successful now.
-
Repeat step 6. for the other user workflow presets
user-shared-debug-install
,user-static-release-install
anduser-static-debug-install
as needed.
How to use
Now, the integrator code is compiled and you can use it in other projects. The
installed integrator package is discoverable by CMake as CalgoIntegrators.
In the CMakeLists.txt
of the other project, call:
include(FetchContent)
FetchContent_Declare(
CalgoIntegrators
GIT_REPOSITORY "https://gitlab.aei.uni-hannover.de/netlib/calgo-integrators.git"
GIT_TAG v1.11.0
SYSTEM
FIND_PACKAGE_ARGS 1.11.0 CONFIG NAMES CalgoIntegrators
)
set(CalgoIntegrators_INCLUDE_PACKAGING TRUE)
FetchContent_MakeAvailable(CalgoIntegrators)
This discovers an installed version of the CalgoIntegrators package or adds it to the other project's install targets. To help CMake discover the CalgoIntegrators package, call CMake for the other project like this:
$ cmake -B ./build -D "CMAKE_PREFIX_PATH=/path/to/CalgoIntegrators/install/prefix"
Replace the path to the CalgoIntegrators install prefix with the actual path on
your system! If CalgoIntegrators is installed in a proper system-wide location,
the CMAKE_PREFIX_PATH
shouldn't be necessary.
This makes two CMake targets available to you, the Netlib::DCUHRE
and the
Netlib::DQUAD
library targets (static or shared depends on the value of
BUILD_SHARED_LIBS
in the other project). The Netlib::libf2c
is already
contained in these targets.
For more information read the two papers referenced below.
References
Paola Favati, Grazia Lotti, and Francesco Romani. 1991. Algorithm 691: Improving QUADPACK automatic integration routines. ACM Trans. Math. Softw. 17, 2 (June 1991), 218–232. DOI:https://dl.acm.org/doi/10.1145/108556.108580
Jarle Berntsen, Terje O. Espelid, and Alan Genz. 1991. Algorithm 698: DCUHRE: an adaptive multidemensional integration routine for a vector of integrals. ACM Trans. Math. Softw. 17, 4 (Dec. 1991), 452–456. DOI:https://dl.acm.org/doi/10.1145/210232.210234