Quantum principal component analysis only achieves an exponential speedup because of its state preparation assumptions

PDFHTML

A central roadblock to analyzing quantum algorithms on quantum states is the lack of a comparable input model for classical algorithms. Inspired by recent work of the author [E. Tang, STOC'19], we introduce such a model, where we assume we can efficiently perform $\ell^2$-norm samples of input data, a natural analogue to quantum algorithms that assume efficient state preparation of classical data. Though this model produces less practical algorithms than the (stronger) standard model of classical computation, it captures versions of many of the features and nuances of quantum linear algebra algorithms. With this model, we describe classical analogues to Lloyd, Mohseni, and Rebentrost's quantum algorithms for principal component analysis [Nat. Phys. 10, 631 (2014)] and nearest-centroid clustering [arXiv:1307.0411]. Since they are only polynomially slower, these algorithms suggest that the exponential speedups of their quantum counterparts are simply an artifact of state preparation assumptions.
Submitted 31 Oct 2018 to Data Structures and Algorithms [cs.DS]
Published 02 Nov 2018
Updated 06 Aug 2021
Author comments: 6 pages + 7 pages (supplemental material). Title used to be "Quantum-inspired classical algorithms for principal component analysis and supervised clustering"
Journal ref: Phys. Rev. Lett. 127, 060503 (2021)
Doi: 10.1103/PhysRevLett.127.060503
https://arxiv.org/abs/1811.00414
https://arxiv.org/pdf/1811.00414.pdf
https://arxiv-vanity.com/papers/1811.00414
Scited by: Abel Molina, Abhijith J, Abhinav Deshpande, Adam Bouland, Alessandro, Alexander Poremba, Alex Wozniakowski, Andrea Mari, Andrea Rocchetto, Andrew Childs, Andrew Guo, Andrey Rakhubovsky, Andru Gheorghiu, Anirban N. Chowdhury, Anurag Anshu, Anurag Saha Roy, Aram Harrow, Arthur Pesah, Ben, Ben Criger, Carlo Sparaciari, Changpeng Shao, Chao-Hua Yu, Christopher Chamberland, Christophe Vuillot, Christoph Hirche, cookie, Corey Ostrove, Danial Dervovic, Daniel Suess, David Elkouss, David Gosset, Dawei Ding, Dripto Debroy, Eleanor Rieffel, Elizabeth Crosson, Ernesto Galvão, Frédéric Grosshans, Gael Sentís, Guillaume Verdon, Han-Hsuan Lin, Hsin-Yuan Huang, Jalex Stark, James D. Watson, Jianhao He, João Bravo, John Smolin, Juani Bermejo-Vega, Kejia, Kunal Sharma, Leonard Wossnig, Luca Innocenti, Man-Hong Yung, Marco Tomamichel, Marcus P da Silva, Mario Berta, Māris Ozols, Martin Schwarz, Mathys Rennela, Matthias C. Caro, Michal Oszmaniec, Mithuna Yoganathan, Nana Liu, Nayeli A. Rodriguez Briones, Nicolas Delfosse, Nicolò Ginatta, Nikolas Breuckmann, Ninnat Dangniam, Noon van der Silk, Oscar Higgott, Patrick Rall, Ramis Movassagh, Roberto Ferrara, Rudy Raymond, Sanketh Menda, Shantanav, Shih-Han Hung, Shijie Pan, Sirui Lu, Srinivasan, Stephen Jordan, Stuart Hadfield, sultana, Supartha Podder, Thomas D. Galley, Tongyang Li, Unpil Baek, Vishal Katariya, Will Kirby, Xin Wang, Xun Gao, Yehua Liu, Yichen Huang, Yuanjia, Yu Luo, Yupan Liu, Zhang Danbo, Евгений Вихров


View this paper on arXiv.wiki:
https://arxiv.wiki/abs/1811.00414

0 comments