C. Rigetti

Unsupervised Machine Learning on a Hybrid Quantum Computer

  1. J. S. Otterbach,
  2. R. Manenti,
  3. N. Alidoust,
  4. A. Bestwick,
  5. M. Block,
  6. B. Bloom,
  7. S. Caldwell,
  8. N. Didier,
  9. E. Schuyler Fried,
  10. S. Hong,
  11. P. Karalekas,
  12. C. B. Osborn,
  13. A. Papageorge,
  14. E. C. Peterson,
  15. G. Prawiroatmodjo,
  16. N. Rubin,
  17. Colm A. Ryan,
  18. D. Scarabelli,
  19. M. Scheer,
  20. E. A. Sete,
  21. P. Sivarajah,
  22. Robert S. Smith,
  23. A. Staley,
  24. N. Tezak,
  25. W. J. Zeng,
  26. A. Hudson,
  27. Blake R. Johnson,
  28. M. Reagor,
  29. M. P. da Silva,
  30. and C. Rigetti
Machine learning techniques have led to broad adoption of a statistical model of computing. The statistical distributions natively available on quantum processors are a superset of
those available classically. Harnessing this attribute has the potential to accelerate or otherwise improve machine learning relative to purely classical performance. A key challenge toward that goal is learning to hybridize classical computing resources and traditional learning techniques with the emerging capabilities of general purpose quantum processors. Here, we demonstrate such hybridization by training a 19-qubit gate model processor to solve a clustering problem, a foundational challenge in unsupervised learning. We use the quantum approximate optimization algorithm in conjunction with a gradient-free Bayesian optimization to train the quantum machine. This quantum/classical hybrid algorithm shows robustness to realistic noise, and we find evidence that classical optimization can be used to train around both coherent and incoherent imperfections.
arxiv pdf

Parametrically-Activated Entangling Gates Using Transmon Qubits

  1. S. Caldwell,
  2. N. Didier,
  3. C. A. Ryan,
  4. E. A. Sete,
  5. A. Hudson,
  6. P. Karalekas,
  7. R. Manenti,
  8. M. Reagor,
  9. M. P. da Silva,
  10. R. Sinclair,
  11. E. Acala,
  12. N. Alidoust,
  13. J. Angeles,
  14. A. Bestwick,
  15. M. Block,
  16. B. Bloom,
  17. A. Bradley,
  18. C. Bui,
  19. L. Capelluto,
  20. R. Chilcott,
  21. J. Cordova,
  22. G. Crossman,
  23. M. Curtis,
  24. S. Deshpande,
  25. T. El Bouayadi,
  26. D. Girshovich,
  27. S. Hong,
  28. K. Kuang,
  29. M. Lenihan,
  30. T. Manning,
  31. J. Marshall,
  32. Y. Mohan,
  33. W. O'Brien,
  34. C. Osborn,
  35. J. Otterbach,
  36. A. Papageorge,
  37. J.-P. Paquette,
  38. M. Pelstring,
  39. A. Polloreno,
  40. G. Prawiroatmodjo,
  41. V. Rawat,
  42. R. Renzas,
  43. N. Rubin,
  44. D. Russell,
  45. M. Rust,
  46. D. Scarabelli,
  47. M. Scheer,
  48. M. Selvanayagam,
  49. R. Smith,
  50. A. Staley,
  51. M. Suska,
  52. N. Tezak,
  53. T.-W. To,
  54. M. Vahidpour,
  55. N. Vodrahalli,
  56. T. Whyland,
  57. K. Yadav,
  58. W. Zeng,
  59. and C. Rigetti
We propose and implement a family of entangling qubit operations activated by radio-frequency flux pulses. By parametrically modulating the frequency of a tunable transmon, these operations
selectively actuate resonant exchange of excitations with a statically coupled, but otherwise off-resonant, neighboring transmon. This direct exchange of excitations between qubits obviates the need for mediator qubits or resonator modes, and it allows for the full utilization of all qubits in a scalable architecture. Moreover, we are able to activate three highly-selective resonances, corresponding to two different classes of entangling gates that enable universal quantum computation: an iSWAP and a controlled-Z rotation. This selectivity is enabled by resonance conditions that depend both on frequency and amplitude, and is helpful in avoiding frequency crowding in a scalable architecture. We report average process fidelities of F = 0.93 for a 135 ns iSWAP, and F = 0.92 for 175 ns and 270 ns controlled-Z operations.
arxiv pdf