Tutorials#
Getting started#
For a simple walk-through of using Triumvirate, see Quick Guide.
To learn about program parameter configuration, see Parameter Configuration.
Usage guides#
For advanced usage of the Python package, consult the following tutorials for various functionalities:
Although the C++ library is intended as the backend for the Python package, a ‘black-box’ C++ program is provided to offer users an end-to-end clustering measurement pipeline. The guide to running the program can be found in ‘Running the C++ program’.
OpenMP parallelisation
For both Python and C++ programs built with OpenMP enabled
(see Installation), you can set the environmental variable
OMP_NUM_THREADS
for multithreaded execution:
$ export OMP_NUM_THREADS=<num-threads>
In general, the following process affinity settings are found to be performant:
$ export OMP_PLACES=threads
$ export OMP_PROC_BIND=spread
However, the optimal settings may vary depending on the system.
For developers, C++ Library contains the full C++
API reference. To reuse C++ routines, ensure the linker finds libtrv
,
e.g. by providing the path to the static library (built following
Installation) with the flag -L<path-to-libtrv.a>
during compilation.