Quantum-Enhanced Generalized Pattern Search Optimization
Abstract
:1. Introduction
2. Preliminaries
2.1. Quantum Computing
2.2. QSearch
Algorithm 1 QSearch |
|
- : The expected number of times QSearch will use the operator Q before returning a desired state is in
- : QSearch will fail to terminate.
2.3. Generalized Pattern Search Algorithms
Algorithm 2 Generalized Pattern Search |
|
- 1.
- Ref. [1] (Theorem 3.6). The iterate set contains a refining subsequence, i.e., a subsequence of local mesh optimizers such that , which converges to some .
- 2.
- Ref. [1] (Theorem 3.7). Let be a convergent refining subsequence with limit point and d be an element of a positive spanning set. If the objective function is evaluated at for infinitely many iterates in , and f is Lipschitz in a neighborhood of , then the Clarke generalized directional derivative of f at in the direction of d is non-negative:
- 3.
- Ref. [1] (Theorem 3.9). Let be a convergent refining subsequence with limit . If the objective function f is strictly differentiable at , then .
3. Adapting QSearch for Generalized Pattern Search Algorithms
3.1. The Quantum Representation of Our Optimization Problem
3.2. Quantum Improved Point Search
Algorithm 3 Local Mesh Filter |
|
Algorithm 4 Quantum Improved Point Search |
|
- 1.
- (Correctness) QIPS correctly determines if contains an improved mesh point. More formally, if and only if for all .
- 2.
- (Complexity: Improved Mesh Point) Suppose contains an improved mesh point; more formally, suppose there exists such that . Then, QIPS will return an improved mesh point using an expected number of calls to the oracles F and f in , where t is the number of improved mesh points contained in .
- 3.
- (Complexity: Local Mesh Optimizer) Suppose does not contain an improved mesh point; more formally, suppose for all . Then, QIPS will return x using calls to the oracles F and f.
- Proof.
- 1.
- From lines (7) and (15) of QIPS, we have that if and only if the while loop started on line (4) terminates because . The result follows as line (7) returns if and only if each element originally in was evaluated by the classical oracle f and failed to be an improved mesh point.
- 2.
- Suppose contains improved mesh points. For any iteration that does not return an improved mesh point, the number of calls to f is identical to those of F. Furthermore, if an improved mesh point is found on lines 3, 9, or 11, then no further calls to f are made. Thus, we proceed by bounding the expected number of calls to F. We obtain an upper bound on the expected number of times F is called following a similar approach to that used in [9] (Lemma 2 and Theorem 3) and [3] (Theorem 3). The oracle F is called only when the oracle Q is called. This occurs a single time on line 3, and then again each time lines 9 and 11 are executed. Lines 3 and 11 return an improved mesh point with probability . On the kth iteration of the while loop begun on line 4, we have that line 9 returns an improved mesh point with probability (recall that .) Let denote the probability of obtaining an improved mesh point on the kth iteration of the while loop. Since j is chosen as an integer in uniformly at random, it follows that is given byFor all positive integers k, we can establish a lower bound for as follows:In particular, this implies when .If , then the expected number of times line 11 will be executed is bounded above by 4 and the result follows. We now assume . In this case, we have thatLet , During the kth iteration of the while loop, the total number of calls to F is bounded above by . It follows that the total number calls to F while is then bounded above byHence, if an improved mesh point is returned during iteration , we have that it does so using calls to F. Now suppose an improved mesh point is not found during the first iterations of the while loop. Since the probability of obtaining an improved mesh point after iteration is bounded below by , it follows that the expected number of calls to F needed to obtain an improved mesh point is bounded above byThus, the expected number of calls to the oracle F is in .
- 3.
- Suppose contains no improved mesh points. Then, during each measurement on lines (3), (11), and (9). The while loop terminates if . This occurs only once the LMF algorithm has evaluated each point originally in on line (6) and determined that has no improved mesh points. Line (9) of LMF guarantees that no point in is evaluated by the oracle f twice, ensuring f is called at most N times. Since the number of times the oracle F is called on lines (9) and (11) of QIPS is bounded above by the number of times f is called, it follows that the number of times F is called is also bounded above by N.
4. Conclusions
Author Contributions
Funding
Data Availability Statement
Conflicts of Interest
Abbreviations
MDPI | Multidisciplinary Digital Publishing Institute |
DOAJ | Directory of Open Access Journals |
GPS | Generalized Pattern Search |
QIPS | Quantum Improved Point Search |
DFO | Derivative-Free Optimization |
References
- Audet, C.; Dennis, J.E. Analysis of Generalized Pattern Searches. SIAM J. Optim. 2002, 13, 889–903. [Google Scholar] [CrossRef]
- Grover, L.K. Quantum Mechanics Helps in Searching for a Needle in a Haystack. Phys. Rev. Lett. 1997, 79, 325–328. [Google Scholar] [CrossRef]
- Brassard, G.; Høyer, P.; Mosca, M.; Tapp, A. Quantum Amplitude Amplification and Estimation. Quantum Comput. Inf. 2002, 305, 53–74. [Google Scholar] [CrossRef]
- Gilyén, A.; Arunachalam, S.; Wiebe, N. Optimizing Quantum Optimization Algorithms via Faster Quantum Gradient Computation. arXiv 2017, arXiv:1711.00465. [Google Scholar]
- Jordan, S.P. Fast Quantum Algorithm for Numerical Gradient Estimation. Phys. Rev. Lett. 2005, 95, 050501. [Google Scholar] [CrossRef] [PubMed]
- Bernstein, E.; Vazirani, U. Quantum Complexity Theory. SIAM J. Comput. 1997, 26, 1411–1473. [Google Scholar] [CrossRef]
- Durr, C.; Hoyer, P. A Quantum Algorithm for Finding the Minimum. arXiv 1996, arXiv:quant-ph/9607014. [Google Scholar] [CrossRef]
- Baritompa, W.; Bulger, D.; Wood, G. Grover’s Quantum Algorithm Applied to Global Optimization. SIAM J. Optim. 2005, 15, 1170–1184. [Google Scholar] [CrossRef]
- Boyer, M.; Brassard, G.; Hoeyer, P.; Tapp, A. Tight bounds on quantum searching. Protein Sci. 1996, 46, 493–505. [Google Scholar]
- Arunachalam, S. Quantum Speed-ups for Boolean Satisfiability and Derivative-Free Optimization. Master’s Thesis, University of Waterloo, Waterloo, ON, Canada, 2014. [Google Scholar]
- Torczon, V. On the Convergence of Pattern Search Algorithms. SIAM J. Optim. 1997, 7, 1–25. [Google Scholar] [CrossRef]
- Cortese, J.A.; Braje, T.M. Loading Classical Data into a Quantum Computer. arXiv 2018, arXiv:1803.01958. [Google Scholar]
- Draper, T.G. Addition on a Quantum Computer. arXiv 2000, arXiv:1803.01958. [Google Scholar]
- Audet, C.; Dennis, J. Mesh Adaptive Direct Search Algorithms for Constrained Optimization. SIAM J. Optim. 2006, 17, 188–217. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Mikes, C.; Gutman, D.H.; Howle, V.E. Quantum-Enhanced Generalized Pattern Search Optimization. Quantum Rep. 2024, 6, 509-521. https://doi.org/10.3390/quantum6040034
Mikes C, Gutman DH, Howle VE. Quantum-Enhanced Generalized Pattern Search Optimization. Quantum Reports. 2024; 6(4):509-521. https://doi.org/10.3390/quantum6040034
Chicago/Turabian StyleMikes, Colton, David Huckleberry Gutman, and Victoria E. Howle. 2024. "Quantum-Enhanced Generalized Pattern Search Optimization" Quantum Reports 6, no. 4: 509-521. https://doi.org/10.3390/quantum6040034
APA StyleMikes, C., Gutman, D. H., & Howle, V. E. (2024). Quantum-Enhanced Generalized Pattern Search Optimization. Quantum Reports, 6(4), 509-521. https://doi.org/10.3390/quantum6040034