Systems and methods for hybrid algorithms using cluster contraction
Systems and methods are described for operating a hybrid computing system using cluster contraction for converting large, dense input to reduced input that can be easily mapped into a quantum processor. The reduced input represents the global structure of the problem. Techniques involve partitioning the input variables into clusters and contracting each cluster. The input variables can be partitioned using an Unweighted Pair Group Method with Arithmetic Mean algorithm. The quantum processor returns samples based on the reduced input and the samples are expanded to correspond to the original input.
1. A method of operation of a hybrid computing system, the hybrid computing system comprising a quantum processor and a classical processor, the method comprising:
applying an algorithm to a problem having an input with n population variables via the classical processor;
converting the input with n population variables to a reduced input with k population variables, where k≤n, via the classical processor, wherein converting the input with n population variables to a reduced input with k population variables, where k≤n, includes partitioning the n population variables into k clusters and contracting each of the k clusters to produce a reduced input with k population variables;
obtaining samples S r from the reduced input via the quantum processor; and
expanding the samples S r to obtain a set of modified states to the problem via the classical processor.
2. The method of claim 1 wherein applying an algorithm to a problem having an input with n population variables via the classical processor includes applying a parallel tempering algorithm to a problem having an input with n population variables via the classical processor.
3. The method of claim 1 wherein contracting each of the k clusters to produce a reduced input with k population variables includes contracting each of the k clusters relative to a low-energy state s to produce a reduced input with k population variables and expanding the samples S r to obtain a set of modified states to the problem includes expanding the samples S r relative to the low-energy state s to obtain a set of modified states to the problem.
4. The method of claim 3 wherein contracting each of the k clusters relative to a low-energy state s includes one of: clamping all population variables in each cluster relative to each other based on their value in the state s and using a heuristic solver to create a new state s based on the clustering.
5. The method of claim 1 wherein partitioning the n population variables into k clusters includes partitioning the n population variables into k clusters using a low-energy subset of states and contracting each of the k clusters includes averaging the values of all the population variable relative the low-energy subset of states in each cluster.
6. The method of claim 5 wherein partitioning the n population variables into k clusters using a low-energy subset of states includes partitioning the n population variables into k clusters via an Unweighted Pair Group Method with Arithmetic Mean algorithm using pairwise entropy as a distance metric.
7. The method of claim 1 wherein obtaining samples S r from the reduced input via the quantum processor includes applying an embedding algorithm to the problem with the reduced input via the classical processor before obtaining samples S r from the reduced input via the quantum processor.
8. The method of claim 1 , further comprising:
iteratively repeating until an exit condition is met:
applying an algorithm to the problem via the classical processor;
converting the input with n population variables to a reduced input with k population variables, where k≤n, via the classical processor;
obtaining samples S r from the reduced input via the quantum processor;
expanding the samples S r to obtain a set of modified states to the problem via the classical processor; and
integrating the modified states into the problem.
9. The method of claim 8 , further comprising determining whether an exit condition has been met.
10. The method of claim 9 wherein determining whether an exit condition has been met includes determining whether a measure of the quality of the population variables is satisfied.
11. A hybrid computing system, the hybrid computing system comprising a quantum processor and a classical processor, the classical processor operable to:
apply an algorithm to a problem having an input with n population variables;
convert the input with n population variables to a reduced input with k population variables, where k≤n, wherein the classical processor is operable to partition the repopulation variables into k clusters and contract each of the k clusters to produce a reduced input with k variables;
obtain samples S r from the reduced input from the quantum processor; and
expand the samples S r to obtain a set of modified states to the problem.
12. The system of claim 11 wherein the algorithm is a parallel tempering algorithm.
13. The system of claim 11 wherein each of the k cluster is contracted relative to a low-energy state s and the samples S r are expanded relative to the low-energy state s.
14. The system of claim 13 wherein all population variables in each cluster are contracted via at least one of: clamping all population variables in each cluster relative to each other based on their value in the state s and creating a new state s based on the clustering via a heuristic solver.
15. The system of claim 11 wherein the n population variables are partitioned into k clusters using a low-energy subset of states and the samples S r are expanded relative to the low-energy subset of states.
16. The system of claim 15 wherein the n population variables are partitioned into k clusters via an Unweighted Pair Group Method with Arithmetic Mean algorithm using pairwise entropy as a distance metric.
17. The system of claim 11 , wherein the classical processor is operable to embed the problem with the reduced input via an embedding algorithm before obtaining samples S r from the reduced input from the quantum processor.
18. The system of claim 11 , wherein the classical processor is further operable to iteratively repeat until an exit condition is met:
applying an algorithm to the problem;
converting the input with n population variables to a reduced input with k population variables, where k≤n;
obtaining samples S r from the reduced input from the quantum processor;
expanding the samples S r to obtain a set of modified states to the problem; and
integrating the modified states into the problem.
19. A method of operation of a hybrid computing system, the hybrid computing system comprising a quantum processor and a classical processor, the method comprising:
applying an algorithm to a problem having an input with n population variables via the classical processor;
converting the input with n population variables to a reduced input with k population variables, where k≤n, via the classical processor;
obtaining samples S r from the reduced input via the quantum processor, wherein obtaining samples S r from the reduced input via the quantum processor includes applying an embedding algorithm to the problem with the reduced input via the classical processor before obtaining samples S r from the reduced input via the quantum processor; and
expanding the samples S r to obtain a set of modified states to the problem via the classical processor.
20. The method of claim 19 wherein applying an algorithm to a problem having an input with n population variables via the classical processor includes applying a parallel tempering algorithm to a problem having an input with n population variables via the classical processor.