Quantum Computing

Sanjay S, Manish P, & Shanmuga Prasanth

23bsds152sanjays@skacas.ac.in, 23bsds130manishp@skacas.ac.in, 23bsds155shanmugaprasanth@skacas.ac.in

 II. ELEMENTS OF QUANTUM COMPUTING
Abstract— Changing the model underlying information and A. Bits and Qubits
computation from a classical mechanical to a quantum The state space of a physical system consists of all possible
mechanical one yields faster algorithms, novel cryptographic states of the system. Any quantum mechanical system that can
mechanisms, and alternative methods of communication. be modelled by a two dimensional complex vector space can
Quantum algorithms can perform a select set of tasks vastly
more efficiently than any classical algorithm, but for many tasks be viewed as a qubit. Such systems include photon
it has been proven that quantum algorithms provide no polarization, electron spin, and a ground state and an excited
advantage. The breadth of quantum computing applications is state of an atom. A key difference between classical and
still being explored. Major application areas include security and quantum systems is the way in which component systems
the many fields that would benefit from efficient quantum combine. The state of a classical system can be completely
simulation. The quantum information processing viewpoint characterized by the state of each of its component pieces. A
provides insight into classical algorithmic issues as well as a surprising and unintuitive aspect of quantum systems is that
deeper understanding of entanglement and other non-classical
aspects of quantum physics. This text describes some of the
most states cannot be described in terms of the states of the
introductory aspects of quantum computing. system’s components.
Such states are called entangled states. Another key property
is quantum measurement. In spite of there being a continuum
Keywords— quantum, information processing, qubit, classical, of possible states, any measurement of a system of qubits has
protocol.
only a discrete set of possible outcomes; for n qubits, there are
at most 2n possible outcomes. After measurement, the system
will be in one of the possible outcome states. Which outcome
I. INTRODUCTION
is obtained is probabilistic; outcomes closest to the measured
In the last two decades of the twentieth century, researchers state are most probable. Unless the state is already in one of
recognized that the standard model of computation placed the possible outcome states, measurement changes the state; it
unnecessary limits on computation. Our world is inherently is not possible to reliably measure an unknown state without
quantum mechanical. By placing computation on a quantum disturbing it. Just as each measurement has a discrete set of
mechanical foundation faster algorithms, novel cryptographic possible outcomes, any mechanism for copying quantum
mechanisms, and alternative methods of communication have states can only correctly copy a discrete set of quantum states.
been found. Quantum information processing, a field that For an n qubit system, the largest number of quantum states a
includes quantum computing, quantum cryptography, copying mechanism can copy correctly is 2n. For any state
quantum communication, and quantum games, examines the there is a mechanism that can correctly copy it, but if the state
implications of using a quantum mechanical model for is unknown, there is no way to determine which mechanism
information and its processing. Quantum information should be used. For this reason, it is impossible to copy
processing changes not only the physical processes used for reliably an unknown state, an aspect of quantum mechanics
computation and communication, but the very notions of called the no cloning principle.
information and computation themselves. A qubit has two arbitrarily chosen distinguished states,
In quantum computers we exploit quantum effects to compute labelled |0i and |1i, which are the possible outcomes of a
in ways that are faster or more efficient than, or even single measurement. Every single qubit state can be
impossible, on conventional computers. Quantum computing represented as a linear combination, or superposition, of these
does not provide efficient solutions to all problems. Nor does two states. In quantum information processing, classical bit
it provide a universal way of circumventing the slowing of values of 0 and 1 are encoded in the distinguished states |0i
Moore’s law as fundamental limits to miniaturization are and |1i. This encoding enables a direct comparison between
reached. Quantum computation enables certain problems to be bits and qubits: bits can only take on two values, 0 and 1,
solved efficiently; some problems which on a classical while qubits can take on any superposition of these values, a|
computer would take more than the age of the universe, a 0i+b|1i, where a and b are complex numbers such that |a|2+|b|
quantum computer could solve in a couple of days. But for 2 = 1.
Any transformation of an n qubit system can be obtained by
performing a sequence of one and two qubit operations. Most
transformations cannot be performed efficiently in this
other problems it has been proven that quantum computation manner. Figuring out an efficient sequence of quantum
cannot improve on classical methods, and for yet another transformations that can solve a useful problem is the heart of
class, that the improvement is small. quantum algorithm design.
Quantum computing combines quantum mechanics, B. Entangled States
information theory, and aspects of computer science. The field Subatomic particles can be entangled, this means that they are
is a relatively new one that promises secure data transfer, connected, regardless of distance. Their effect on each other
dramatic computing speed increases, and may take component upon measurement is instantaneous. This can be useful for
miniaturisation to its fundamental limit.
computational purposes. Measuring entangled states accounts algorithm. At around the same time the quantum complexity
for the correlations between them. classes were developed and the quantum Turing machine was
described.
C. Quantum Circuits Then in 1996 Lov Grover developed a fast database search
If we take a quantum state, representing one or more qubits, algorithm (known as Grover’s algorithm). The first prototypes
and apply a sequence of unitary operators (quantum gates) the of quantum computers were also Elements of Quantum
result is a quantum circuit. We now take a register and let Computing built in 1996. In 1997 quantum error correction
gates act on qubits, in analogy to a conventional circuit. techniques were developed at Bell labs and IBM. Physical
implementations of quantum computers improved with a three
qubit machine in 1999 and a seven qubit machine in 2000.
A. What classical computers can and cannot do
Computer scientists categorize problems according to how
many computational steps it would take to solve a large
FIGURE 1. SIMPLE QUANTUM CIRCUIT example of the problem using the best algorithm known. The
This circuit above is a series of operations and measurements problems are grouped into broad, overlapping classes based
on the state of n-qubits. Each operation is unitary and can be on their difficulty. Three of the most important classes are
described by an 2n £ 2n matrix. Each of the lines is an abstract listed below. Contrary to myth, quantum computers are not
wire, the boxes containing Un are quantum logic gates (or a known to be able to solve efficiently the very hard class called
series of gates) and the meter symbol is a measurement. NP-complete problems.
Together, the gates, wires, input, and output mechanisms 1) P Problems: Ones computers can solve efficiently, in
implement quantum algorithms. polynomial time.
Unlike classical circuits which can contain loops, quantum Example: Given a road map showing n towns, can you get
circuits are “one shot circuits” that just run once from left to from any town to every other town? For a large value of n, the
right (and are special purpose: i.e. we have a different circuit number of steps a computer needs to solve this problem
for each algorithm). increases in proportion to n2, a polynomial. Because
It should be noted that it is always possible to rearrange polynomials increase relatively slowly as n increases,
quantum circuits so that all the measurements are done at the computers can solve even very large P problems within a
end of the circuit. Quantum circuit diagrams have the reasonable length of time.
following constraints which make them different from 2) NP Problems: Ones whose solutions are easy to
classical diagrams. verify.
1. They are acyclic (no loops). Example: You know an n-digit number is the product of two
2. No FANIN, as FANIN implies that the circuit is NOT large prime numbers, and you want to find those prime
reversible, and therefore factors. If you are given the factors, you can verify that they
not unitary. are the answer in polynomial time by multiplying them.
3. No FANOUT, as we can’t copy a qubit’s state during the Every P problem is also an NP problem, so the class NP
computational phase because of the no-cloning theorem. contains the class P within it. The factoring problem is in NP
All of the above can be simulated with the use of ancilla and but conjectured to be outside of P, because no known
garbage bits if we assume that no qubits will be in a algorithm for a standard computer can solve it in only a
superposition. Garbage bits are useless qubits left over after polynomial number of steps. Instead the number of steps
computation and ancilla bits are extra qubits needed for increases exponentially as n gets bigger.
temporary calculations. 3) NP-complete problems: An efficient solution to one
The fundamental unit of quantum computation, the qubit, can would provide an efficient solution to all NP challenges.
take on a continuum of values, but a discrete version of Example: Given a map, can you colour it using only three
quantum computation can be constructed that preserves the colours so that no neighbouring countries are the same colour?
features of standard quantum computation. If you had an algorithm to solve this problem, you could adapt
the algorithm to solve any other NP problem (such as the
III. WHY QUANTUM COMPUTING? factoring problem above or determining if you can pack n
boxes of various sizes into a trunk of a certain size) in about
0. History the same number of steps. In that sense, NP-complete
In 1982 Richard Feynman theorised that classic computation problems are the hardest of the NP problems. No known
could be dramatically improved by quantum effects, building algorithm can solve an NP-complete problem efficiently.
on this, David Deutsch developed the basis for quantum
computing between 1984 and 1985. The next major B. Where quantum computing fits in
breakthrough came in 1994 when Peter Shor described a
method to factor large numbers in quantum poly-time (which
breaks RSA encryption). This became known as Shor’s
 While “quantum cryptography” is often used as a synonym for
“quantum key distribution,” quantum approaches to a wide
variety of other cryptographic tasks have been developed.
Some of these protocols use quantum means to secure
classical information. Others secure quantum information.
Many are “unconditionally” secure in that their security is
based entirely on properties of quantum mechanics. Others are
only quantum computationally secure in that their security
depends on a problem being computationally intractable for a
quantum computer.
Closely related to quantum key distribution schemes are
protocols for unclonable encryption, a symmetric key
encryption scheme that guarantees that an eavesdropper
cannot copy an encrypted message without being detected.
Unclonable encryption has strong ties with quantum
authentication. One type of authentication is digital signatures.
Quantum digital signature schemes have been developed, but
FIGURE 1. VARIOUS CLASSES OF COMPUTATIONAL PROBLEMS
the keys can be used only a limited number of times. In this
respect they resemble classical schemes such as Merkle’s one-
The map above depicts how the class of problems that time signature scheme.
quantum computers would solve efficiently (BQP) might
relate to other fundamental classes of computational problems. B. Broader Implications
(The irregular border signifies that BQP does not seem to fit Quantum information theory has led to insights into
neatly with the other classes.) fundamental aspects of quantum mechanics, particularly
The BQP class (the letters stand for bounded-error, quantum, entanglement. Efforts to build quantum information
polynomial time) includes all the P problems and also a few processing devices have resulted in the creation of highly
other NP problems, such as factoring and the so-called entangled states that have enabled deeper experimental
discrete logarithm problem. Most other NP and all NP- exploration of quantum mechanics. These entangled states,
complete problems are believed to be outside BQP, meaning and the improvements in quantum control, have been used in
that even a quantum computer would require more than a quantum microlithography to affect matter at scales below the
polynomial number of steps to solve them. wavelength limit and in quantum metrology to achieve
In addition, BQP might protrude beyond NP, meaning that extremely accurate sensors. Applications include clock
quantum computers could solve certain problems faster than accuracy beyond that of current atomic clocks, which are
classical computers could even check the answer. (Recall that limited by the quantum noise of atoms, optical resolution
a conventional computer can efficiently verify the answer of beyond the wavelength limit, ultra-high resolution
an NP problem but can efficiently solve only the P problems.) spectroscopy, and ultra-weak absorption spectroscopy.
To date, however, no convincing example of such a problem The quantum information processing viewpoint has also
is known. provided a new way of viewing complexity issues in classical
Computer scientists do know that BQP cannot extend outside computer science, and has yielded novel classical algorithmic
the class known as PSPACE, which also contains all the NP results and methods. Classical algorithmic results stemming
problems. PSPACE problems are those that a conventional from the insights of quantum information processing include
computer can solve using only a polynomial amount of lower bounds for problems involving locally decodable codes,
memory but possibly requiring an exponential number of local search, lattices, reversible circuits, and matrix rigidity.
steps. The usefulness of the complex perspective for evaluating real
valued integrals is often used as an analogy to explain this
IV. IMPLICATIONS AND APPLICATIONS phenomenon.
A. Quantum Protocols Cryptographic protocols usually rely on the empirical
hardness of a problem for their security; it is rare to be able to
Applications of quantum information processing include a prove complete, information theoretic security. When a
number of communication and cryptographic protocols. The cryptographic protocol is designed based on a new problem,
two most famous communication protocols are quantum the difficulty of the problem must be established before the
teleportation and dense coding. Both use entanglement shared security of the protocol can be understood. Empirical testing
between the two parties that are communicating. of a problem takes a long time. Instead, whenever possible,
Quantum key distribution schemes were the first examples of “reduction” proofs are given that show that if the new problem
quantum protocols. Quantum key distribution protocols were solved it would imply a solution to a known hard
establish a secret symmetric key between both parties, but problem.
their security rests on properties of quantum mechanics.
C. Impact on security
Electronic commerce relies on secure public key encryption  If quantum computers ever become a reality, the
and digital signature schemes, as does secure electronic “killer app” for them will most likely not be code
communication. Without secure public key encryption, breaking but rather something so obvious it is rarely
authentication and the distribution of symmetric session keys even mentioned: simulating quantum physics. This is
become unwieldy. a fundamental problem for chemistry,
Both factoring and the discrete logarithm problem are nanotechnology and other fields, important enough
candidate NP intermediate problems. Hope for alternative that Nobel Prizes have been awarded even for partial
public key encryption protocols centers on using other NP progress.
intermediate problems. The leading candidates are certain
 As transistors in microchips approach the atomic
lattice based problems. Some of these schemes have
scale, ideas from quantum computing are likely to
impractically large keys, while for others their security
become relevant for classical computing as well.
remains in question. Also, Regev showed that lattice based
problems are closely related to the dihedral hidden subgroup  Quantum computing experiments focus attention
problem. The close relationship of the dihedral hidden directly on the most mystifying features of quantum
subgroup problem with problems solved by Shor’s algorithm mechanics—and hopefully, the less we can sweep
makes many people nervous, though so far the dihedral hidden those puzzles under the rug, the more we will be
subgroup problem has resisted attack. forced to understand them.
Given the historic difficulty of creating practical public key  Quantum computing can be seen as the most
encryption systems based on problems other than factoring or stringent test to which quantum mechanics itself has
discrete log, it is unclear which will come first, a large scale ever been subjected. In my opinion, the most exciting
quantum computer or a practical public key encryption system possible outcome of quantum computing research
secure against quantum and classical attacks. If the building of would be to discover a fundamental reason why
quantum computers wins the race, the security of electronic quantum computers are not possible. Such a failure
commerce and communication around the world will be would overturn our current picture of the physical
compromised. world, whereas success would merely confirm it.
V. LIMITATIONS VI. CONCLUSION
Beals et al. proved that, for a broad class of problems, Will scalable quantum computers ever be built? Yes. Will
quantum computation cannot provide any speed-up. Their quantum computers eventually replace desktop computers?
methods were used by others to provide lower bounds for No. Quantum computers will always be harder to build and
other types of problems. Ambainis found another powerful maintain than classical computers, so they will not be used for
method for establishing lower bounds. In 2002, Aaronson the many tasks that classical computers do equally efficiently.
showed that quantum approaches could not be used to Quantum computers will be useful for a number of specialized
efficiently solve collision problems. This result means there is tasks. The extent of these tasks is still being explored.
no generic quantum attack on cryptographic hash functions. However long it takes to build a scalable quantum computer
Shor’s algorithms break some cryptographic hash functions, and whatever the breadth of applications turns out to be,
and quantum attacks on others may still be discovered, but quantum information processing has changed forever the way
Aaronson’s result says that any attack must use specific in which quantum physics is taught and understood. The
properties of the hash function under consideration. quantum information processing view of quantum mechanics
Grover’s search algorithm is optimal; it is not possible to clarifies key aspects of quantum mechanics such as quantum
search an unstructured list of N elements more rapidly than measurement and entangled states. The practical
O(√N). This bound was known before Grover found his consequences of this increased understanding of nature are
algorithm. Childs et al. showed that for ordered data, quantum hard to predict, but they can hardly fail to profoundly affect
computation can give no more that a constant factor technological and intellectual developments in the coming
improvement over optimal classical algorithms. Grigni et al. decades.
showed in 2001 that for most non-abelian groups and their
subgroups, the standard Fourier sampling method, used by
Shor and successors, yields exponentially little information REFERENCES
about a hidden subgroup. [1] Eleanor Rieffel, “Quantum Computing,” April 29, 2011.
[2] Riley T. Perry, “The Temple of Quantum Computing,” April 29, 2006.
[3] Scott Aaronson, “The Limits of Quantum,” Scientific American, p. 62-
If a large, ideal quantum computer would face most of the 69, March 2008.
[4] Wikipedia-The free encyclopedia [Online]. Available:
same limitations as our present-day classical computers do, http://www.wikipedia.org/
should the physicists working on the extraordinarily hard task [5] TheFreeDictionary.com [Online]- Available:
of building even rudimentary quantum computers pack up and http://encyclopedia.thefreedictionary.com/
go home? The answer is no, for four reasons. [6] Wolfram, A New Kind of Science, 1st edition, Wolfram Media, USA,
2002.
[7] Science Blogs [Online]- Available: http://scienceblogs.com/
[8] R. Feynman. Feynman Lectures on Computation. Addison-Wesley,
Reading, MA, 1996.