A quantuu algorithu of nding the roots of a polynouial function

Koji Nagata,1 Tadao Nakamura,2 Han Geurdes,3 Josep Batle,4 Soliman

Abdalla,5 Ahmed Farouk,6 Do Ngoc Diep,7, 8 and Santanu Kumar Patro9
1
Department of Physics, Korea Advanced Institute of Science and Technology, Daejeon 34141, Korea
2
Department of Information and Computer Science, Keio University,
3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan
3
Geurdes Datascience, KvK 64522202, C vd Lijnstraat 164, 2593 NN, Den Haag Netherlands
4
Departament de Fsica, Universitat de les Illes Balears,
07122 Palma de Mallorca, Balearic Islands, Europe
5
Department of Physics, Faculty of Science, King Abdulaziz University Jeddah, P.O. Box 80203, Jeddah 21589, Saudi Arabia
6
Computer Sciences Department, Faculty of Computers and Information, Mansoura University
7
Institute of Mathematics, VAST, 18 Hoang Quoc Viet road, 10307, Hanoi Vietnam
8
TIMAS, Thang Long University, Nghiem Xuan Yem, Dai Kim, Hoang Mai, Hanoi, Vietnam
9
Department of Mathematics, Berhampur University, Berhampur-760007, Odisha, INDIA
( Dated: October 14, 2017)
A quantum algorithm of nding the roots of a polynomial function f (x) = xm + am1 xm1 + ... +
a1 x + a0 is proposed. Here all the roots are in the integers Z. The speed of solving the problem is
shown to outperform the classical case by a factor of m.

PACS numbers: 03.67.Lx, 03.67.Ac

Keywords: Quantum computation, Quantum algorithms

Quantum mechanics (cf. [16]) is successful in explain-
ing and predicting many physical phenomena. Therefore
there are reasons to be convinced of the correctness of
quantum mechanics. One of the interesting applications
of quantum principles is the application to information
theory [6] leading to the quantum computer.
Looking at studies of quantum computing, implemen-
tation of a quantum algorithm to solve Deutschs prob-
lem [79] on a nuclear magnetic resonance quantum com-
puter is reported rstly [10]. An implementation of the
Deutsch-Jozsa algorithm on an ion-trap quantum com-
puter is also reported [11]. There are several attempts to
use single-photon two-qubit states for quantum comput-
ing. Oliveira et al. implements Deutschs algorithm with
polarization and transverse spatial modes of the electro-
magnetic eld as qubits [12]. In addition, single-photon
Bell states are prepared and measured [13]. Also the
decoherence-free implementation of Deutschs algorithm
is introduced by using such single-photon and by using
two logical qubits [14]. A one-way based experimental
implementation of Deutschs algorithm is reported [15].
For a number of recent algorithmic developments we
mention the following. In 1993, the Bernstein-Vazirani
algorithm was published [16, 17]. This work can be con-
sidered an extension of the Deutsch-Jozsa algorithm. In
1994, Simons algorithm [18] and Shors algorithm [19]
were discussed. In 1996, Grover [20] provides the high-
est motivation for exploring the computational possi-
bilities oered by quantum mechanics. Implementation
of a quantum algorithm to solve the Bernstein-Vazirani
parity problem without entanglement in an ensemble
quantum computer can be mentioned as an important
quantum algorithm [21]. Fiber-optics implementation of
the Deutsch-Jozsa and Bernstein-Vazirani quantum algo-
rithms with three qubits was also discussed in the recent
past [22]. The question whether or not quantum learning
is robust against noise is a subject of intense study [23].
A quantum algorithm for approximating the inu-
ences of Boolean functions and its applications is re-
cently studied [24]. In addition, quantum computation
with coherent spin states and the close Hadamard prob-
lem [25] is reported. Transport implementation of the
Bernstein-Vazirani algorithm with ion qubits is studied
[26]. Quantum Gauss-Jordan elimination and simulation
of accounting principles on quantum computers are dis-
cussed [27]. We mention that the dynamical analysis of
Grovers search algorithm in arbitrarily high-dimensional
search spaces is studied [28]. A method of computing
many functions simultaneously by using many parallel
quantum systems is reported [29].
On the other hand, we may wonder if we need all the
previously mentioned studies to reach a good quantum
computer. In 2015, it was discussed that the Deutsch-
Jozsa algorithm can be used for quantum key distribution
[30]. In 2017, it was discussed that secure quantum key
distribution based on Deutschs algorithm using an en-
tangled state [31]. Subsequently, a highly speedy secure
quantum cryptography based on the Deutsch-Jozsa algo-
rithm is proposed [32]. The relation between quantum
computer and secret sharing with the use of quantum
principles is discussed [33].
The Bernstein-Vazirani algorithm determines bit
strings. It is extended to determining the values of a
function [34]. By using many parallel quantum systems,
we can discuss many functions [29]. However, the values
of the functions are restricted in {0, 1}. Here, by ex-
tending it more, we present an algorithm of determining
the values of a function that are extended to the natural
 2

numbers N. And we discuss a mathematical problem. In where we have used 0 = 1.

classical case, the ecient algorithm of nding the roots Subsequently let us dene the wave function | as
of a high degree polynomial function does not exist. follows
Motivated by the fact, we propose a quantum algo- 1
rithm of nding the roots of a polynomial function f (x) = | = ( d |0 + d1 |1 + + |d 1 ) (8)
d
xm + am1 xm1 + ... + a1 x + a0 . Here x R and the
roots are in the integers; |r1 | } |r2 | } ... } |rm |, rj Z. where = e2i/d . In the following, we discuss the Fourier
Given 2S(= N ) absolute values of the function transform of |d 1
d1 d1
|f (S)|, ..., |f (2)|, |f (1)|, y(d1) |y ydy |y
|d 1 =
|f (1)|, |f (2)|, ..., |f (S)| (1) y=0 d y=0 d
d1
all the unknown roots of the function shall be found, dy |y
= = | (9)
simultaneously. In the best classical case, we need m d
y=0
steps, whereas in quantum case we need a query. The
speed of nding the roots is shown to outperform the where we have used yd = d = 1.
classical case by a factor of m. Our method is based The Fourier transform of |x1 ...xN is as follows
on the generalized Bernstein-Vazirani algorithm [34] to
|x1 ...xN
qudit systems [35].
d1 d1
Our algorithm combines quantum superposition with z1 x1 |z1 zN xN |zN
...
a property of quantum mechanics known as interference. d d
z1 =0 zN =0
Assume we are given natural numbers
zx |z
= (10)
1, 2, ..., S |a0 | = |r1 ||r2 |...|rm | |rm | (2) dN
zK

where |a0 | is the absolute value of the constant and |rm | where K = {0, 1, 2, ..., d 1}N and z is (z1 , z2 , ..., zN ).
is the largest absolute value of the roots. Here the prob- Hence, for completeness, zK is a shorthand to the
lem is of searching the roots of the function by a query. compound sum
We introduce a positive integer d(> S). Throughout the
paper, we consider the modulo d. Assume the following . (11)
z1 {0,1,..,d1} z2 {0,1,..,d1} zN {0,1,..,d1}
N
After the N + 1 Fourier transforms on the state (6)
|f (S)|, ..., |f (1)|, |f (1)|, ..., |f (S)| } d 1 (3)
N
where f (j) {0, 1, ..., d 1}, and we dene
F |0 F |0 ... F |0 F |d 1 (12)
f (N ) = (|f (S)|, ..., |f (1)|, |f (1)|, ..., |f (S)|) (4) we have
where each entry of f (N ) is a natural number in the |x
|1 = | . (13)
modulo d. Here f (N ) {0, 1, ..., d 1}N . dN
xK
We dene ff (N ) (x) as follows
We introduce SU Mff (N ) (x) gate
ff (N ) (x) = f (N ) x mod d
|x |j |x |(ff (N ) (x) + j) mod d . (14)
= |f (S)|x1 + |f (1 S)|x2 + + |f (S)|xN mod d
Here
(5)
ff (N ) (x) = f (N ) x mod d. (15)
where x {0, 1, ..., d 1}N .
Let us follow the quantum states through the algo- We have
rithm. The input state is SU Mf (N )x |x | = f (N )x |x | . (16)
N
|0 = |0 |d 1 (6) In what follows, we will discuss the reasonable behind
of the above relation (16).
N
Now consider applying the SU Mf (N )x gate to the
N
where |0 means |0, 0, ..., 0 . state |x | . Each term in | is of the form dj |j .
We discuss the Fourier transform of |0 We see
d1 d1
y0 |y |y SU Mf (N )x dj |x |j
|0 = (7)
y=0 d y=0 d dj |x |(j + f (N ) x) mod d . (17)
 3

We introduce k such as f (N ) x + j = k d j = from which

d + f (N ) x k.
Hence (17) becomes |(|f (S)|, ..., |f (1)|, |f (1)|, ..., |f (S)|) (28)

SU Mf (N )x dj |x |j can be obtained. That is to say, if we measure

the rst N qudits state of the state |3 , that is,
f (N )x dk |x |k mod d . (18) |(|f (S)|, ..., |f (1)|, |f (1)|, ..., |f (S)|) , then we can re-
Now, when k < d we have |k mod d = |k and thus, trieve the following values
the terms in | such that k < d are transformed as |f (S)|, ..., |f (1)|, |f (1)|, ..., |f (S)| (29)
follows
using a single query. If |f (3)| = 0, |f (5)| = 0, ..., |f (S
SU Mf (N )x dj |x |j f (N )x dk |x |k . (19) 2)| = 0, then noting that x is again a scalar here, the
Also, as f (N ) x and j are bounded above by d 1, k function can be factorized as follows
is strictly less than 2d. Hence, when d } k < 2d we have f (x) = (x + 3)(x 5)...(x (S 2)). (30)
|k mod d = |k d .
Now, we introduce m such that k d = m then we Therefore we can nd the m roots of the function. All
have we have to do is to perform one quantum measurement.
The speed to determine the m roots improves by a
f (N )x dk |x |k mod d = f (N )x m |x |m
factor of m as compared to the best classical counterpart.
= f (N )x dm |x |m . (20) Notice that we recover the generalized Bernstein-Vazirani
algorithm when d = 2 [29, 34].
Hence the terms in | such that k d are transformed
In conclusion, a quantum algorithm of nding the roots
as follows
of a polynomial function f (x) = xm + am1 xm1 +
SU Mf (N )x dj |x |j f (N )x dm |x |m . (21) ... + a1 x + a0 has been proposed. Here x R and
the roots are in the integers |r1 | } |r2 | } ... }
Hence from (19) and (21) we have
|rm |, rj Z. Given absolute values of the function
SU Mf (N )x |x | = f (N )x |x | . (22) |f (S)|, ..., |f (1)|, |f (1)|, ..., |f (S)|, (|a0 | } S), all the
unknown roots of the function shall have been found,
Therefore, the relation (16) holds. simultaneously. The speed of nding the roots has been
We have |2 by operating SU Mff (N ) (x) to |1 shown to outperform the classical case by a factor of m.
f (N )x |x
SU Mf (N )x |1 = |2 = | . (23)
xK dN
After the Fourier transform on |x , using the previ- [1] J. von Neumann, Mathematical Foundations of Quantum
ous equations (10) and (23) we can now evaluate |3 as Mechanics (Princeton University Press, Princeton, New
follows Jersey, 1955).
[2] R. P. Feynman, R. B. Leighton, and M. Sands, Lectures
()xz+f (N )x |z on Physics, Volume III, Quantum mechanics (Addison-
|3 = | . (24)
dN Wesley Publishing Company, 1965).
zK xK
[3] M. Redhead, Incompleteness, Nonlocality, and Realism
Because we have (Clarendon Press, Oxford, 1989), 2nd ed.
[4] A. Peres, Quantum Theory: Concepts and Methods
()x = 0 (25) (Kluwer Academic, Dordrecht, The Netherlands, 1993).
xK [5] J. J. Sakurai, Modern Quantum Mechanics (Addison-
we may notice Wesley Publishing Company, 1995), Revised ed.
[6] M. A. Nielsen and I. L. Chuang, Quantum Computation
()x(z+f (N )) = dN z+f (N ),0 = dN z,f (N ) . (26) and Quantum Information (Cambridge University Press,
xK 2000).
[7] D. Deutsch, Proc. Roy. Soc. London Ser. A 400, 97
Therefore, the above summation is zero if z = f (N ) (1985).
and the above summation is dN if z = f (N ). Thus [8] D. Deutsch and R. Jozsa, Proc. Roy. Soc. London Ser. A
434, 553 (1992).
()xz+f (N )x |z [9] R. Cleve, A. Ekert, C. Macchiavello, and M. Mosca, Proc.
|3 = |
dN Roy. Soc. London Ser. A 454, 339 (1998).
zK xK
[10] J. A. Jones and M. Mosca, J. Chem. Phys. 104, 1648
dN z,f (N ) |z (1998).
= | [11] S. Gulde, M. Riebe, G. P. T. Lancaster, C. Becher, J.
dN
zK Eschner, H. Haner, F. Schmidt-Kaler, I. L. Chuang, and
= |(|f (S)|, ..., |f (1)|, |f (1)|, ..., |f (S)|) | (27) R. Blatt, Nature (London) 421, 48 (2003).
 4

[12] A. N. de Oliveira, S. P. Walborn, and C. H. Monken, J. (2015).

Opt. B: Quantum Semiclass. Opt. 3, 288-292 (2005).
[13] Y.-H. Kim, Phys. Rev. A 63, 040301(R) (2003).
[14] M. Mohseni, J. S. Lundeen, K. J. Resch, and A. M. Stein-
berg, Phys. Rev. Lett. 41, 187903 (2003).
[15] M. S. Tame, R. Prevedel, M. Paternostro, P. Bohi, M.
S. Kim, and A. Zeilinger, Phys. Rev. Lett. 48, 140501
(2007).
[16] E. Bernstein and U. Vazirani, Proceedings of the
Twenty-Fifth Annual ACM Symposium on The-
ory of Computing (STOC 93), pp. 11-20 (1993),
doi:10.1145/167088.167097.
[17] E. Bernstein and U. Vazirani, SIAM J. Comput. 26-5,
pp. 1411-1473 (1997).
[18] D. R. Simon, Foundations of Computer Science, (1994)
Proceedings., 35th Annual Symposium on: 116-123, re-
trieved 2011-06-06.
[19] P. W. Shor, Proceedings of the 35th IEEE Symposium
on Foundations of Computer Science. 124 (1994).
[20] L. K. Grover, Proceedings of the Twenty-Eighth Annual
ACM Symposium on Theory of Computing. 212 (1996).
[21] J. Du, M. Shi, X. Zhou, Y. Fan, B. J. Ye, R. Han, and
J. Wu, Phys. Rev. A 64, 042306 (2001).
[22] E. Brainis, L.-P. Lamoureux, N. J. Cerf, Ph. Emplit, M.
Haelterman, and S. Massar, Phys. Rev. Lett. 40, 157902
(2003).
[23] A. W. Cross, G. Smith, and J. A. Smolin, Phys. Rev. A
42, 012327 (2015).
[24] H. Li and L. Yang, Quantum Inf. Process. 14, 1787
[25] M. R. A. Adcock, P. Hoyer, and B. C. Sanders, Quantum
Inf. Process. 15, 1361 (2016).
[26] S. D. Fallek, C. D. Herold, B. J. McMahon, K. M. Maller,
K. R. Brown, and J. M. Amini, New J. Phys. 18, 083030
(2016).
[27] D. N. Diep, D. H. Giang, and N. Van Minh, Int J Theor
Phys (2017) 56: 1948. https://doi.org/10.1007/s10773-
017-3340-8.
[28] W. Jin, Quantum Inf. Process. 15, 65 (2016).
[29] K. Nagata, G. Resconi, T. Nakamura, J. Batle, S. Ab-
dalla, A. Farouk, and H. Geurdes, Asian J. Math. Phys.
1 (1) (2017), 14.
[30] K. Nagata and T. Nakamura, Open Ac-
cess Library Journal, 2: e1798 (2015).
http://dx.doi.org/10.4236/oalib.1101798.
[31] K. Nagata and T. Nakamura, Int J Theor Phys (2017)
56: 2086. https://doi.org/10.1007/s10773-017-3352-4
[32] K. Nagata, T. Nakamura, and A. Farouk, Int J Theor
Phys (2017) 56: 2887. https://doi.org/10.1007/s10773-
017-3456-x.
[33] D. N. Diep and D. H. Giang, Int J Theor Phys (2017)
56: 2797. https://doi.org/10.1007/s10773-017-3444-1.
[34] K. Nagata, G. Resconi, T. Nakamura, J. Batle, S. Ab-
dalla, and A. Farouk, MOJ Ecology and Environmental
Science, Volume 2, Issue 1, 00010, (2017).
[35] R. Krishna, V. Makwana, A. P. Suresh, arXiv:1609.03185
[quant-ph] (2016).