• Corpus ID: 222208644

Efficient Sampling from Feasible Sets of SDPs and Volume Approximation

@article{Chalkis2020EfficientSF,
  title={Efficient Sampling from Feasible Sets of SDPs and Volume Approximation},
  author={Apostolos Chalkis and Ioannis Z. Emiris and Vissarion Fisikopoulos and Panagiotis Repouskos and Elias P. Tsigaridas},
  journal={ArXiv},
  year={2020},
  volume={abs/2010.03817}
}
We present algorithmic, complexity, and implementation results on the problem of sampling points from a spectrahedron, that is the feasible region of a semidefinite program. Our main tool is geometric random walks. We analyze the arithmetic and bit complexity of certain primitive geometric operations that are based on the algebraic properties of spectrahedra and the polynomial eigenvalue problem. This study leads to the implementation of a broad collection of random walks for sampling from… 
1 Citations

Figures and Tables from this paper

A Cutting-plane Method for Semidefinite Programming with Potential Applications on Noisy Quantum Devices

It is shown how to leverage quantum speed-up of an eigensolver in speeding up an SDP solver utilizing the cutting-plane method, and that the RCP method is very robust to noise in the boundary oracle, which may make RCP suitable for use even on noisy quantum devices.

References

SHOWING 1-10 OF 61 REFERENCES

Practical Polytope Volume Approximation

This work implements and evaluates randomized polynomial-time algorithms for accurately approximating the polytope’s volume in high dimensions based onhit-and-run random walks based on Monte Carlo algorithms with guaranteed speed and provably high probability of success for arbitrarily high precision.

Faster Polytope Rounding, Sampling, and Volume Computation via a Sub-Linear Ball Walk

A novel method of computing polytope membership, where one avoids checking inequalities which are estimated to have a very low probability of being violated is proposed, likely to be of independent interest for constrained sampling and optimization problems.

A Randomized Cutting Plane Method with Probabilistic Geometric Convergence

A randomized method for general convex optimization problems; namely, the minimization of a linear function over a convex body that can be implemented using hit-and-run versions of Markov-chain Monte Carlo algorithms is proposed.

Fast Algorithms for Logconcave Functions: Sampling, Rounding, Integration and Optimization

  • L. LovászS. Vempala
  • Computer Science
    2006 47th Annual IEEE Symposium on Foundations of Computer Science (FOCS'06)
  • 2006
It is proved that the hit-and-run random walk is rapidly mixing for an arbitrary logconcave distribution starting from any point in the support, and the main conjecture formulated there is settled.

Random walks and an O*(n5) volume algorithm for convex bodies

This work introduces three new ideas: the use of the isotropic position (or at least an approximation of it) for rounding, the separation of global obstructions and local obstructions for fast mixing, and a stepwise interlacing of rounding and sampling.

Random walks for probabilistic robustness

  • G. Calafiore
  • Mathematics
    2004 43rd IEEE Conference on Decision and Control (CDC) (IEEE Cat. No.04CH37601)
  • 2004
The general hit-and-run method for uniform sampling in convex bodies, and several key issues related to the so called mixing rate of this process and to Hoeffding-type inequalities for dependent samples are discussed.

Random walks and an O * ( n 5 ) volume algorithm for convex bodies

This algorithm introduces three new ideas: the use of the isotropic position (or at least an approximation of it) for rounding; the separation of global obstructions and local obstructions for fast mixing; and a stepwise interlacing of rounding and sampling.

Practical Volume Estimation by a New Annealing Schedule for Cooling Convex Bodies

A new, practical algorithm for all representations of convex polytopes, as well as zonotopes, is proposed, which is faster than existing methods and relies on Hit-and-Run sampling, and combines a new simulated annealing method with the Multiphase Monte Carlo approach.

Vaidya walk: A sampling algorithm based on the volumetric barrier

It is proved that for a polytope in Rd defined by n constraints, the Vaidya walk mixes in O (√n/d) fewer steps than the Dikin walk, and hence the speed up is significant for polytopes with n » d.

Random Spectrahedra

It is proved that the average number of singular points on the real variety of singular matrices in $\mathbf{1} + V$ is $n(n-1)$, which is related to the volume of the variety of real symmetric matrices with repeated eigenvalues.
...