Simplified Factoring Algorithms for Validating Small-Scale Quantum

Simplified Factoring Algorithms for Validating Small-Scale Quantum

Simplified Factoring Algorithms for Validating Small-Scale Quantum Information Processing Technologies Omar Gamel∗ and Daniel F. V. James Department of Physics, University of Toronto, 60 St. George St., Toronto, Ontario, Canada M5S 1A7. (Dated: November 14, 2013) Tomography has reached its practical limits in characterization of new quantum devices, and there is a need for a new means of characterizing and validating new technological advances in this field. We propose a different verification scheme based on compiled versions of Shor’s factoring algorithm that may be extended to large circuits in the future. The general version Shor’s algorithm has been experimentally elusive due to bottlenecks associated with the modular exponentiation operation. Experiments to date have only been able to execute compiled versions of the latter operation. We provide some new compiled circuits for experimentalists to use in the near future. We also demonstrate that an additional layer of compilation can be added using classical operations, that will reduce the number of qubits and gates needed in a given compiled circuit. PACS numbers: 03.67.Ac I. INTRODUCTION operation. Compilation in this sense uses known infor- mation about the solution to simplify the circuit from Theory remains very much ahead of experiment in its complicated general form to a more manageable size. quantum computing. While complicated quantum algo- Additionally, experimental demonstrations of compiled rithms are developed and studied, experimental capabil- circuits have been carried out using various technologies ities are not advanced enough to implement them. As [11–17]. Experimental implementation of a non-compiled experimental techniques progress on the road to a fully version of Shor’s algorithm may indeed soon be forthcom- functional quantum computer, there is a great need for ing [18]. small scale algorithms and circuit verification techniques In this paper we will use compiled versions of the mod- to verify the functionality of new technology. ular exponentiation operation for the purposes of valida- Traditionally, tomography has been used to character- tion. We demonstrate some techniques to construct the ize and validate quantum computing technology, and test simple compiled circuits, including a “classical layer” of the functionality of new devices [1–3]. However, despite compilation that reduces the number of qubits needed. advances in compressed sensing and other enhancements, It remains to be seen if one can generalize these com- full quantum state tomography beyond more than one or pilation and validation techniques beyond a dozen or so two dozen qubits appears to be intractable due to the qubits, to the regime where tomography has significant amount of measurements required (which grows expo- limitations. nentially with the number of qubits) [4]. Therefore, it is In section II, we provide a brief overview of Shor’s of interest to create new methods of verifying that new algorithm and the steps it involves, distilling it to the quantum devices actually execute their desired function. modular exponentiation operation. Section III demon- strates the process of constructing the compiled modular To this end, it is natural to use well known algorithms, exponentiation circuits for some semiprimes, using cir- such as Shor’s factoring algorithm [5], for verification and cuit synthesis methods similar to those in ref. [19]. In characterization of new technology. However, the com- section IV we briefly discuss the relationship between al- plete version of Shor’s algorithm uses an order finding lowable periods in the modular exponentiation subrou- arXiv:1310.6446v2 [quant-ph] 14 Nov 2013 subroutine that involves a modular exponentiation op- tine for each number to be factored. Section V analyzes eration, which is the bottleneck of the algorithm, need- a specific example of compiled modular exponentiation ing the most quantum gates to implement and the most and order-finding, taking into account the effects of noise time to execute. Creating circuits for a general modular and entanglement in the compiled circuits. exponentiation operation is not possible with currently realizable technology for any nontrivial parameters. There has been much theoretical work on construc- II. SHOR’S ALGORITHM tion of quantum circuits, with efficient techniques devel- oped to simplify circuit construction [6–10]. Among other uses, these techniques have been employed to construct A. Classical Probabilistic Steps “compiled circuits” for the order-finding subroutine in Shor’s algorithm and associated modular exponentiation Shor’s factoring algorithm is held as one of the most promising and useful applications of quantum comput- ing. It allows one to factor large numbers in polyno- mial time, undermining the most common cryptographic ∗Electronic address: [email protected] scheme in use today, RSA cryptography [20]. The algo- 2 rithm is based on an order-finding subroutine, which in and fewer evaluations are needed, leading to an efficient turn makes heavy use of the modular exponentiation op- computation. eration [21, 22]. The modular exponentiation is the bot- However, despite this, the order finding subroutine tleneck of the algorithm, needing the resources in terms comes with its own limitations. Order finding itself is of quantum gates and computation time. broken down to two steps; modular exponentiation and The purpose of Shor’s algorithm is to factorize a num- a quantum Fourier transform. The modular exponenti- ber N. It is assumed N is odd, and that it is not a power ation is what requires the most resources. The quan- of a prime number. The case N = pk for some prime p tum Fourier transform requires few resources, particu- and integer k can be checked by calculating √k N (the kth larly when implemented semi-classically [24]. So Shor’s root of N) for all integer k log3 N, which is an efficient algorithm while efficient, is still very difficult to apply procedure [21, 22]. Since N≤is not the power of a prime, practically given today’s limitations in quantum technol- it can be written as the product of two coprime integers. ogy. To see this, we discuss the order finding subroutine Shor’s algorithm non-deterministically finds a nontriv- separately below. ial factor of N in the following manner. First, a number a<N is chosen arbitrarily. The greatest common divisor of a and N, denoted gcd(a,N), is found using the Eu- B. Quantum Order Finding Subroutine clidean algorithm [23]. If gcd(a,N) = 1, then gcd(a,N) 6 x is a nontrivial factor of N, and no further work is needed. To find the order of the function fa,N (x) = a In case a and N are coprime, i.e. gcd(a,N) = 1, the mod (N) defined in eq.(1), Shor’s algorithm makes heavy algorithm uses the order finding subroutine explained in use of the modular exponentiation operation given by the next subsection to find the order of a modulo N. U(a): x 0 x axmod(N) . (2) Defining the function fa,N (x) as | i| i → | i| i x Clearly the structure of the circuit that implements this fa,N (x) a mod (N), (1) ≡ operation will depend on the random number a chosen. the order of a modulo N is defined as the smallest positive The quantum state is initialized at integer x such that fa,N (x) = 1. We denote the order r. M 1 The order r is also the period of the function fa,N (x), x 0 , (3) √ X | i| i that is fa,N (x) = fa,N (x + r). For Shor’s algorithm to M x=0 succeed, either r has to be even, or a has to be a square of an integer, so that either way ar/2 is an integer. where M = 2m is the smallest power of 2 that satisfies 2 If r is found to be odd, and a is not a square, then M N . We then apply the modular exponentiation ≥ the algorithm trial has failed and a new random number operation in eq.(2) to our initial state, yielding a above must be attempted. If r is even, we compute M M ar/2 mod (N). If ar/2 1 mod (N), once again the 1 1 ≡ − U(a) x 0 = x ax(modN) . (4) algorithm has failed and a new random number a must X X √M =0 | i| i √M =0 | i| i be tried. We note that ar/2 = 1 mod (N), because if x x r/2 6 a was congruent to 1 modulo N then r would not be Applying the inverse quantum Fourier transform will the smallest power that satisfies the order condition. yield a quantum state which when measured, will with Supposing the algorithm succeeded, we have that a high probability lead us to the order r of the function 2 2 ar/ = 1 mod (N) and ar/ = 1 mod (N), yet ar =1 f (x)= ax mod (N). 6 − 6 2 a,N mod (N). These statements imply that N ∤ (ar/ + 1) Since the modular exponentiation operation U(a) de- 2 2 2 and N ∤ (ar/ 1), yet N (ar 1)=(ar/ + 1)(ar/ 1). pends on the arbitrary number a, the general circuit im- − | − − We have used standard number theory notation, where plementing it must have a as a variable input, making b c means b is a factor of c, and b ∤ c means b is not a it very complicated. In other words, it is very difficult | factor of c. to create a single circuit that will execute the modular 2 2 Since N divides the product of (ar/ +1) and (ar/ 1), exponentiation in eq.(2) for arbitrary values of x and a. − but not each of them alone, then the nontrivial factors For this reason, the modular exponentiation is by far of N must be split between the two numbers.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    10 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us