An SDP solver exploiting structures of clusters of constraints and low-rank of the constraint matrices. This is a more general version of SDPB (https://github.com/davidsd/sdpb), and part of the MSc thesis (http://resolver.tudelft.nl/uuid:53e114a8-61cd-4f48-8151-76abb0159408).
(TODO: add program it solves)
Functions include
solvempmp, which converts multivariate polynomial programs to a clustered low-rank SDP and calls the solver.prepareabc, which does the conversion for a single constraint.- functions to create sample points and bases
Using multiple threads is supported. For small problems, this might not give a speed up due to the overhead. To use multiple cores, start Julia with julia -t [number of threads].
Arblib, for arbitrary precision arithmeticBlockDiagonalsGenericSVDGenericLinearAlgebra