Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
144 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
46 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Low-depth Hamiltonian Simulation by Adaptive Product Formula (2011.05283v3)

Published 10 Nov 2020 in quant-ph

Abstract: Various Hamiltonian simulation algorithms have been proposed to efficiently study the dynamics of quantum systems on a quantum computer. The existing algorithms generally approximate the time evolution operators, which may need a deep quantum circuit that is beyond the capability of near-term noisy quantum devices. Here, focusing on the time evolution of a fixed input quantum state, we propose an adaptive approach to construct a low-depth time evolution circuit. By introducing a measurable quantifier that characterizes the simulation error, we use an adaptive strategy to learn the shallow quantum circuit that minimizes that error. We numerically test the adaptive method with electronic Hamiltonians of the $\mathrm{H_2O}$ and $\mathrm{H_4}$ molecules, and the transverse field Ising model with random coefficients. Compared to the first-order Suzuki-Trotter product formula, our method can significantly reduce the circuit depth (specifically the number of two-qubit gates) by around two orders while maintaining the simulation accuracy. We show applications of the method in simulating many-body dynamics and solving energy spectra with the quantum Krylov algorithm. Our work sheds light on practical Hamiltonian simulation with noisy-intermediate-scale-quantum devices.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (39)
  1. R. P. Feynman, Int. J. Theor. Phys 21 (1982).
  2. I. M. Georgescu, S. Ashhab, and F. Nori, Reviews of Modern Physics 86, 153 (2014).
  3. I. Buluta and F. Nori, Science 326, 108 (2009).
  4. R. Blatt and C. F. Roos, Nature Physics 8, 277 (2012).
  5. A. Y. Kitaev, arXiv preprint quant-ph/9511026  (1995).
  6. N. H. Stair, R. Huang, and F. A. Evangelista, Journal of Chemical Theory and Computation 16, 2236 (2020).
  7. R. M. Parrish and P. L. McMahon, arXiv preprint arXiv:1909.08925  (2019).
  8. R. D. Somma, New Journal of Physics 21, 123025 (2019).
  9. O. Kyriienko, npj Quantum Information 6, 1 (2020).
  10. P. Zeng, J. Sun, and X. Yuan, arXiv e-prints , arXiv:2109.15304 (2021), arXiv:2109.15304 [quant-ph] .
  11. M. Suzuki, Journal of mathematical physics 26, 601 (1985).
  12. M. Suzuki, Journal of Mathematical Physics 32, 400 (1991).
  13. D. W. Berry, A. M. Childs, and R. Kothari, in 2015 IEEE 56th Annual Symposium on Foundations of Computer Science (IEEE, 2015) pp. 792–809.
  14. G. H. Low and I. L. Chuang, Quantum 3, 163 (2019).
  15. G. H. Low and I. L. Chuang, Physical review letters 118, 010501 (2017).
  16. A. M. Childs and Y. Su, Physical review letters 123, 050503 (2019).
  17. A. M. Childs, A. Ostrander, and Y. Su, Quantum 3, 182 (2019b).
  18. E. Campbell, Phys. Rev. Lett. 123, 070503 (2019).
  19. M. Heyl, P. Hauke, and P. Zoller, Science advances 5, eaau8342 (2019).
  20. J. Preskill, Quantum 2, 79 (2018).
  21. This property only holds when we consider the first-order error ΔΔ\Deltaroman_Δ, which may fail for higher-order errors.
  22. The CNOT gate count is computed by assuming an explicit decomposition of e−i⁢P⁢xsuperscript𝑒𝑖𝑃𝑥e^{-iPx}italic_e start_POSTSUPERSCRIPT - italic_i italic_P italic_x end_POSTSUPERSCRIPT using 2⁢nq−22subscript𝑛𝑞22n_{q}-22 italic_n start_POSTSUBSCRIPT italic_q end_POSTSUBSCRIPT - 2 CNOT gates, where the Pauli word P𝑃Pitalic_P applies nontrivially on nqsubscript𝑛𝑞n_{q}italic_n start_POSTSUBSCRIPT italic_q end_POSTSUBSCRIPT qubits.
  23. S. B. Bravyi and A. Y. Kitaev, Ann. Phys. 298, 210 (2002).
  24. We choose the STO-3g basis and use the Bravyi-Kitaev transformation.
  25. There are n⁢(n−1)/2+n=78𝑛𝑛12𝑛78n(n-1)/2+n=78italic_n ( italic_n - 1 ) / 2 + italic_n = 78 terms in the Hamiltonian and the coefficients are then normalized so that the sum of their absolute value is 78×0.5=39.0780.539.078\times 0.5=39.078 × 0.5 = 39.0.
  26. G. E. Crooks, arXiv preprint arXiv:1811.08419  (2018).
  27. E. Farhi, J. Goldstone, and S. Gutmann, arXiv preprint arXiv:1411.4028  (2014).
  28. Here, ⟨H2⟩delimited-⟨⟩superscript𝐻2\langle H^{2}\rangle⟨ italic_H start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ⟩ is assumed to be accurately estimated and the variance of ⟨H2⟩delimited-⟨⟩superscript𝐻2\langle H^{2}\rangle⟨ italic_H start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ⟩ is negligible because ⟨H2⟩delimited-⟨⟩superscript𝐻2\langle H^{2}\rangle⟨ italic_H start_POSTSUPERSCRIPT 2 end_POSTSUPERSCRIPT ⟩ is invariant under the ideal time evolution e−i⁢H⁢tsuperscript𝑒𝑖𝐻𝑡e^{-iHt}italic_e start_POSTSUPERSCRIPT - italic_i italic_H italic_t end_POSTSUPERSCRIPT.
  29. Here, β𝛽\betaitalic_β is a positive number to be set. Note that Δreg′⁣2⁢(λ→)superscriptsubscriptΔreg′2→𝜆\Delta_{\mathrm{reg}}^{\prime 2}(\vec{\lambda})roman_Δ start_POSTSUBSCRIPT roman_reg end_POSTSUBSCRIPT start_POSTSUPERSCRIPT ′ 2 end_POSTSUPERSCRIPT ( over→ start_ARG italic_λ end_ARG ) can be minimized by solving linear equations. See Supplemental Materials Not for detail.
  30. T. Albash and D. A. Lidar, Rev. Mod. Phys. 90, 015002 (2018).
  31. R. Babbush, P. J. Love, and A. Aspuru-Guzik, Scientific reports 4, 1 (2014).
  32. V. Verteletskyi, T.-C. Yen, and A. F. Izmaylov, The Journal of Chemical Physics 152, 124114 (2020).
  33. H.-Y. Huang, R. Kueng, and J. Preskill, arXiv preprint arXiv:2002.08953  (2020).
  34. X. Bonet-Monroig, R. Babbush, and T. E. O’Brien, Physical Review X 10, 031064 (2020).
  35. Y. Li and S. C. Benjamin, Physical Review X 7, 021050 (2017).
  36. S. Endo, S. C. Benjamin, and Y. Li, Physical Review X 8, 031027 (2018).
  37. K. Temme, S. Bravyi, and J. M. Gambetta, Physical review letters 119, 180509 (2017).
  38. M. Otten and S. K. Gray, Npj Quantum Inf. 5, 11 (2019).
  39. Q. Sun, Journal of Computational Chemistry 36, 1664 (2015).
Citations (26)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets