Professional Documents
Culture Documents
fields, such as quantum computation and thermodynamics, and we generalize it to the case of mixed
quantum states.
PACS numbers:
ity method” [4], in as diverse fields as learning theory, tion III we will relate these two concepts to each other.
complexity theory, and combinatorics, to name just a This section will also serve to introduce our notation.
few. The notion of Kolmogorov complexity has been
used [3] to solve in a simple and elegant way many oth-
erwise difficult problems, such as, e.g., Gödel’s theorem A. Quantum Kolmogorov Complexity
[7]. Recently it has also been used to establish a connec-
tion between classical information and thermodynamics Many classical quantities, e.g the Shannon entropy,
[8, 9, 10, 11]. How to generalize the concept of Kol- have been successfully generalized to quantum informa-
mogorov complexity to the quantum world has become tion and have become useful and powerful tools to under-
a natural question with the development of the theory stand and further develop quantum information theory.
of quantum information. Not only is a good notion of In the case of Kolmogorov complexity, though, the way to
quantum Kolmogorov complexity of deep fundamental do so is not straightforward. Indeed, recently a number
interest, but one also hopes it will allow to develop a of different definitions for quantum Kolmogorov complex-
theory similarly powerful as its classical counterpart. ity have been introduced by Gàcs [12], Berthiaume et al.
In recent years, a number of different definitions of [13], Vitanyi [14], and Mora and Briegel [17]. Contrary
quantum Kolmogorov complexity have been proposed to the classical case, where different definitions have been
[12, 13, 14, 15], each generalizing the classical quantity shown to be equivalent [4], in the quantum setting they
in a different way. Some measure the number of qubits are not. The definition proposed in [12] differs from the
required to describe a quantum state, while others count others in that it is a generalization of an alternative def-
instead the number of classical bits. The aim of this pa- inition of algorithmic complexity, introduced by Levin
per is to shed new light on the different definitions of [18]. The general idea of such approach is not to search
quantum Kolmogorov complexity. In particular the re- for the shortest program that reproduces the string x (as
sults presented here will allow us to better understand one does in the definition proposed by Kolmogorov), but
the physical meaning of some of the definitions. Further- rather to look at the probability of obtaining x when a
more we will show how Kolmogorov complexity can, also random program (that is, a sequence of 0 and 1 gener-
in the quantum scenario, be used as a proof method. ated, e.g., by coin tosses) is given as input to a Turing
The paper is organized as follows. In Section II we machine. It has been shown that the negative logarithm
review some of the existing definitions of quantum Kol- of such a probability coincides with the Kolmogorov com-
mogorov complexity of pure states. Then we review the plexity of the string x. The probability distribution over
notion of communication complexity. We will focus on the set of binary strings generated by running random
the equality problem and, more precisely, on its solution programs on a Turing machine is called universal proba-
by means of the fingerprinting protocol studied in [16]. In bility, and it can be characterized by precise mathemat-
Section III we relate the number of classical bits required ical properties. In his work, Gàcs proposes a universal
to describe a quantum state (as a general definition) to a density matrix, µ, that can be viewed as a generalization
problem in communication complexity. This allows us to of the universal probability distribution. The definition
prove that there exist quantum states of n qubits whose of the Komogorov complexity of a quantum state follows,
description require exponentially many classical bits. In thus, quite naturally from the classical theory. The two
Section IV we use the relation studied in Section III to possible choices, whose properties are analyzed in detail
derive some implications for any definition of Kolmogorov in [12], are
complexity, including a natural generalization to mixed
states. Furthermore we will provide a non-trivial relation H(|ϕi) = −hϕ| log µ|ϕi and H(|ϕi) = − loghϕ|µ|ϕi .
between the so–called network Komogorov complexity of (2)
a class of pure quantum states [17] and the classical Kol- It is easy to see that, given an n-qubit state |ϕi, both
mogorov complexity of bit strings. In the last section we quantities are upper bounded by n.
apply the notion of quantum Kolmogorov complexity to The other definitions mentioned above are, instead,
a number of problems, including computation complexity based on a generalization of complexity, as introduced by
and thermodynamics. Kolmogorov and Chaitin. In [13] the Kolmogorov com-
plexity of a quantum state ρ is defined as the size of the
smallest quantum program (state) that, given as input to
a quantum Turing machine, yields ρ as output. In this
II. REVIEW OF QUANTUM KOLMOGOROV
work, the “length of a quantum state” is defined as the
COMPLEXITY AND COMMUNICATION
logarithm of the dimension of the smallest Hilbert space
COMPLEXITY
that contains the state itself. One can then define the
complexity of ρ as
In this section we first review the different definitions
ε
of quantum Kolmogorov complexity and their properties. KBLV (ρ) = min{l(π)|F (U (π), ρ) > 1 − ε} , (3)
π
Second, we review the notion of communication complex-
ity. In particular we will describe the fingerprinting pro- where 1 − ε measures the fidelity, F , with which the state
tocol studied in [16] to solve the equality problem. In Sec- ρ must be reproduced. It is clear from the definition, that
3
this quantity is a measure of the quantum information one needs at most 2n log 1ε gates in order to prepare an
contained in a state. One can prove that the complexity n-qubit state, and that the length of ω B,ε (|ϕi) is up-
of an n-qubit state is upper bounded by n. per bounded by the number of gates, one obtains that
In [14], on the contrary, the author proposes a defini- B,ε
KNet (|ϕi) is upper bounded by 2n log 1ε , up to terms of
tion of quantum Kolmogorov complexity based on the smaller order. From now on we will refer to this definition
classical description of the state. More precisely, the as “network” complexity.
complexity of a state |ϕi is defined as Assuming that we restrict the choice of the set of basis
KVit (|ϕi) = min{l(p) − log |hψ|ϕi|2 |U(p) = |ψi}. (4) gates to 1- and 2-qubit gates (differing from the original
p definition presented in [17]), one can show that the net-
The idea here is to compute |ψi which approximates the work complexity is independent (up to a constant) of the
desired state |ϕi, and the term − log |hψ|ϕi|2 penalizes actual choice of basis. Keeping this in mind, we will thus
ε
for a bad approximation of the state. Analogously to the simply write KNet (|ϕi).
definition of Berthiaume et al., the complexity of a state It should be noted here that the network complexity
|ϕi can grow at most linearly with the number of qubits has not only a clear physical meaning, since it gives a
whose state is described by |ϕi. measure of how hard it is to describe a way to prepare the
Another approach, which is related to Vitanyi’s one state, but furthermore it was also shown that there exists
has been introduced by Zurek, first, and later by Caves a connection between this definition of Kolmogorov com-
and Schack [10, 11, 19, 20]. Since we will discuss this plexity and entanglement [15]. In order for a state to have
approach in more detail in the last section of this pa- maximal complexity (that is, exponential in the number
per, we also want to briefly review it here. Suppose of qubits) it is in fact necessary that it is highly entan-
that on is given a set of possible states, S = {|Ψi i} and gled: more precisely, a complex state must have maxi-
an associated probability distribution P = {pi }i , with mum Schmidt measure [21]. On the contrary, a product
pi = P (|Ψi i). In order to specify a particular element state has complexity at most linear in the number of
of this set, one just has to specify the index, which is a qubits.
purely classical task. It is well known that the so–called Another possibility to describe a pure state classi-
Shannon-Fano code can be used to obtain a (almost op- cally is simply to describe the coefficients of the state
timal) description [5]. In order to explain the code, let us in a certain basis, e.g. the computational basis. We
assume that p1 ≥ p2 , . . . and define Fi = ik=1 pk . The
P will refer to this complexity as computational-basis-
codeword ci for index i is r(Fi ), which is Fi rounded off to expansion complexity, or, more briefely, CBE complex-
li = ⌊log2 (1/pi )⌋, i.e. the floor of log2 (1/pi ), bits. Thus, ity.
P The CBE complexity of a n-qubit state |Ψi =
if the information of the set S, also called the background i1 ,...,in ci1 ,...,in |i1 , . . . , in i, with ij ∈ {0, 1} and complex
information [20] is given, then only li bits are addition- coefficients ci1 ,...,in is then
ally required to describe the state |Ψi i [33]. Note that ε
KCBE (|Ψi) = KCl [(c0,...0 , . . . , c1,...1 )]. (6)
the logarithmic term in the definition before, can be un-
derstood in this way [14]. In general, the coefficients ci1 ,...,in are computed only
A completely different approach, based on the identifi- with a finite precision ε, i.e. each of them is approximated
cation of a state with its abstract preparation procedure, by a number c̃i1 ,...,in such that |ci1 ,...,in − c̃i1 ,...,in | ≤ ε.
was investigated in [15, 17]. In this case the Kolmogorov This implies that theP state that has been described in
complexity of a state |ϕi is defined as the amount of the this way, i.e. |Ψ̃i = i1 ,...,in c̃i1 ,...,in |i1 , . . . , in i, fulfills
classical information required to describe a circuit that the inequality k|Ψ̃i − |Ψik2 ≤ 2n ε2 . Thus, 1 − |hΨΨ̃i|2 ≤
prepares |ϕi with the required precision ε from a fixed 2n−1 ε2 , which implies that the precision ε must scale like
⊗n
initial state, e.g. |0i . To be more precise, first a com- 1/2n/2 , similarly as the precision with which one must
plete and finite gate basis B is fixed. Any state |ϕi can describe each gate occurring in the definition of the net-
then be prepared, with arbitrary finite precision, ε, by a work complexity.
circuit, C B,ε (|ϕi), built with gates from such a basis [34]. Yet another possibility is to use a measurement and de-
Furthermore, as the basis is finite, it is possible to define scribe the state by the outcome probability of this mea-
a finite code and to encode any circuit in a classical se- surement. One might for instance choose an “algorith-
quence. Thus it is possible to associate to each state |ϕi mically simple” measurement, i.e. a POVM that can be
a set of strings ω B,ε (|ϕi), each of which encodes a cir- described with only a few bits, and describe the state by
cuit that prepares |ϕi with the required accuracy. The the outcome probability of this measurement.
Kolmogorov complexity of |ϕi is then defined as the mini- As we have already mentioned, in classical information
mum among the classical Kolmogorov complexities of the theory the different definitions of complexity are provably
strings ω(C B,ε )(|ϕi) equivalent, while this does not hold in quantum theory.
B,ε
Indeed some of the definitions described here have qual-
KNet (|ϕi) = min K(ω(C B,ε (|ϕi))). (5) itatively different behaviors, e.g. the scaling with n. In
C B,ε (|ϕi)
this paper we aim to shed new light on the notion of
Here, the minimum is taken over all circuits, C B,ε , such quantum Kolmogorov complexity and to better under-
that |hϕ|C B,ε (|0i⊗n )|2 ≥ 1 − ε. Taking into account that stand its physical meaning.
4
One of the main drawbacks of Kolmogorov complex- are usually solved (see e.g. [23] for an overview). We
ity is the fact that it is not computable. Despite of this, will focus here on the so–called worst case scenario. In
though, it is possible to give upper bounds for it, and this case, one is interested in the amount of communica-
these are sufficient and instrumental for many applica- tion needed to compute f (x, y) correctly with probability
tions. Furthermore, we recall that even in the case of larger or equal to 1 − δ, for any x and y.
computable quantities, such as, e.g., communication or Of course, any function f can be exactly evaluated if
computation complexities, one is often not interested in both Alice and Bob send their full input to the referee,
the exact value, since it might be very hard to compute. who can then compute f (x, y). It follows thus that 2n is
In these cases too, one is mainly concerned with the scal- an upper bound for any communication problem in the
ing of these quantities as a function of the input size, n. simultaneous message passing model [35]. To improve
Owing to this fact, the following notation is widely used. such a protocol, one might consider the case in which
One says that a function, g(n), is O(f (n)) if there exist Alice and Bob send to the referee a compressed version
two constants, c ∈ R and n0 ∈ N such that g(n) ≤ cf (n) of their input. In this case, the communication required
for all n ≥ n0 . That is, for sufficiently large n, g is upper depends on the Kolmogorov complexity of the two inputs.
bounded by f . Similarly one says that a function g(n) is If one considers the worst–case scenario and does not
Ω(f (n)) if there exist two constants, c and n0 such that allow any error, this does not help to improve the upper
g(n) ≥ cf (n) for all n ≥ n0 . If the behavior of two func- bound. By considering the case in which the inputs are
tions, g, f is asymptotically the same, up to a constant incompressible n–bit strings, in fact, it is easy to see that
factor, i.e. if g is both O(f (n)) and Ω(f (n)) then we say 2n bits of communication are still needed [36]. However,
that g is Θ(f (n)). if a certain failure
√ probability is tolerated, one can show
From now on, we will call a bit string (a state) algo- that only O( n) bits of communication are required (see
rithmically simple when its Kolmogorov complexity is at below).
most O(log(n)) (at most O(n)). Even though Holevo’s Theorem states that by sending
m qubits one cannot convey more than m classical bits
of information, it has been shown that there are cases in
B. Communication complexity which the quantum setting is significantly more power-
ful than the classical one (see e.g. [23] for an overview).
In fact, there are functions for which the classical com-
The theory of communication complexity deals with
munication complexity is exponentially larger than the
the following type of problem. Two distant parties, Alice
quantum one. One such instance, which we will con-
and Bob, hold each an input, x and y respectively (usu-
sider in the following, is the equality problem. This is a
ally one has x, y ∈ {0, 1}n). The common goal of the par-
worst–case simultaneous message passing – model where
ties is to compute the value f (x, y) for a given function
the function to be evaluated is
f known to both parties. The communication complex-
ity of a function f , CCl (f ), is defined as the minimum
1, if x = y
number of bits that Alice and Bob need to communicate EQn (x, y) = (7)
0, if x 6= y ,
in order to evaluate f (x, y).
In the quantum setting, the communication between with x, y ∈ {0, 1}n.
Alice and Bob takes place through a quantum channel.
Analogous to the classical case, the quantum communi- It has been proved [24, 25]
√ that, allowing a small prob-
cation complexity of f , CQ (f ), is the minimum number ability of failure, only O( n) bits of classical communi-
of qubits that Alice and Bob need to communicate to cation are needed to compute EQn (x, y). Furthermore,
evaluate f (x, y). this amount of communication has been shown to be both
A slightly different communication model is the so– necessary and sufficient for the solution of the equality
called simultaneous message passing model, first intro- problem.
duced by Yao [22]. In this model there is a third party Let us now briefly recall one optimal classical proto-
–the referee– in addition to Alice and Bob. As before, the col here. It is known that for every n ∈ N and for a
common aim of the parties is to compute f (x, y), where fixed c > 1, and 0 < ∆ < 1, there exists an error cor-
f is a given function and x and y are the input strings recting code E : {0, 1}n → {0, 1}m, with m = cn, such
held by Alice and Bob respectively. In this case, how- that the distance between code words E(x) and E(y) is
ever, no direct communication between Alice and Bob at least (1 − ∆)m [37]. Alice and Bob apply a given er-
is allowed. They can only send messages to the referee, ror correcting code (where the specification of the code
who will then compute f (x, y) based on the information is independent of n) to x, y respectively. Then they
he has received. choose randomly an index i, j ∈ {1, . . . , m}, respectively.
There are many cases in which one does not require Alice (Bob) sends i and the i–th bit value of E(x), i.e.
the function f to be computed with certainty, but al- Ei (x), (j, and Ej (y)) to the referee. If i 6= j the ref-
lows a certain failure probability, δ. The conditions on eree cannot conclude anything and tells Alice and Bob
this error, as well as other factors, distinguish different to restart the protocol. However, if i = j, the referee
scenarios in which communication complexity problems checks whether Ei (x) equals Ej (y). If this is the case
5
he concludes that x = y else he concludes x 6= y. The problem, this consideration leads to a lower bound to the
probability of having a collision, i.e. that the random number of bits required to describe those states, i.e. to
indices i, j coincide, is 1/m. The probability of making their quantum Kolmogorov complexity. We show now
an error is Perror = P (Ei (x) = Ei (y) | x 6= y)/m < ∆. that this lower bound is exponential in the number of
Thus, in order to make the error probability independent qubits.
of m ε
√ (i.e. independent of n), Alice and Bob have to send Let KQ be an arbitrary definition of the quantum Kol-
O( n) bits. Note that the error can be reduced to any mogorov complexity with the following physical meaning:
δ > 0 if Alice and Bob send O(log(1/δ)) random bits of ε
given KQ (|ϕi) classical bits one can prepare the state
the code words, which implies that the communication |ψi with precision ε. Now and in the following, we say
complexity √ of the equality problem, with error probabil- that KQ ε
(|ϕi) bits of information allow to prepare |ϕi
ity δ, is O( n log(1/δ)). with precision ε if it contains all the information that is
In [16], the case in which Alice and Bob’s fingerprints needed to prepare a state ε-close to |ϕi with a perfect
can consist of quantum information is studied and it is apparatus. Or, equivalently, given KQ ε
(|ϕi) bits of infor-
shown that O(log2 n)-qubit fingerprints are sufficient to mation, we can write down a state, ε-close to |ϕi. At the
solve the equality problem (with arbitrarily small error same time we imply that such a task is impossible with
probability). Let us briefly review also this protocol here. a smaller amount of information. We show now that for
Using the same notation as before one defines, for each any such definition there exists a n–qubit state |ψi such
x ∈ {0, 1}n the following (log2 m + 1)-qubit state that KQ ε
(|ψi) is exponential in n, ruling out some of the
m definitions reviewed in Section II.
1 X Let KQ ε
(|hx i) and KQε
(|hy i) be the Kolmogorov com-
|hx i = √ |ii|Ei (x)i , (8)
m i=1 plexities of the two fingerprints used in the protocol.
Then a possible classical communication protocol is the
where, as above, Ei (x) is the i-th bit of the word E(x). following:
Since the two words, E(x) and E(y) can be equal for at
most m∆ positions, for any x 6= y we have |hhx |hy i| ≤ 1. Given x and y, respectively, Alice and Bob send the
∆. The simultaneous message passing protocol for the description of the (log2 m+1)-qubit states |hx i and
ε ε
equality problem is the following. |hy i to the referee, requiring KQ (|hx i) + KQ (|hy i)
bits of communication.
1. Alice and Bob, holding x and y respectively, pre-
pare a quantum system in the states |hx i and |hy i 2. In order to distinguish between the case where
respectively and send it to the referee. |hx i = |hy i or |hx i 6= |hy i, the referee could e.g.
simulate (on a classical computer) the circuit de-
2. The referee uses an auxiliary qubit, R and pre- scribed in the quantum fingerprinting protocol.
pares the state (HR ⊗ IAB )(C-SWAPR,AB )(HR ⊗
IAB )|0iR (|hx i|hy i)AB = 1/2[|0iR (|hx , hy i + The precision with which the referee is able to simulate
|hy , hx i)AB + |1iR (|hx , hy i − |hy , hx i)AB ]. Then the quantum circuit, and thus the probability with which
he measures the auxiliary system in the computa- he can obtain the correct answer, depends on the preci-
tional basis and associates x = y with the outcome sion with which the two quantum states are described. It
“0”, and x 6= y with the outcome “1”. is straightforward to show that a final error probability
(such as in the quantum fingerprinting protocol) ∆, is√ ob-
With this test, the referee can determine the value of
EQn (x, y) with error probability (1 + ∆2 )/2. Such a tained if the states are described with precision ε ≤ ∆
probability can be reduced to any δ > 0 by considering [38].
the fingerprint |hx i⊗k , with k = O(log2 1δ ). In this case The number of bits of communication of this proto-
ε ε
the length of each fingerprint is O(log2 n log2 1/δ). col is KQ (|hx i) + KQ (|hy i). As we know that the op-
timal bound for the classical communication complex-
ity of the equality
√ problem, in the simultaneous passing
III. RELATION BETWEEN CLASSICAL model, is n, it follows that there must be√at least a
ε ε
COMMUNICATION COMPLEXITY AND case for which KQ (|hx i) + KQ (|hy i) ≥ O( n). Con-
QUANTUM KOLMOGOROV COMPLEXITY sidering that m = nc, it follows that there must exist
an M -qubit state |ΨCompl
M i, with M = log2 m such that
ε Compl M/2
In this section we establish a relation between Kol- KQ (|ΨM i) ≥ O(2 /c). This shows that any defini-
mogorov complexity and communication complexity. To tion of the quantum Kolmogorov complexity that counts
this end, we consider a classical protocol that simu- the number of classical bits required to prepare a state
lates the quantum fingerprinting protocol. That is, in- must grow exponentially with the number of qubits.
stead of sending quantum systems (prepared in the states One of the definitions presented in Sec II that fulfills
|hx i , |hy i), Alice and Bob send to the referee a classi- the necessary requirement of growing exponentially with
cal description of these states. Since there exists a lower the number of qubits is the network complexity [15]. Note
bound on the classical communication complexity for this that also the CBE complexity fulfills this condition.
6
B. Quantum complex states from algorithmically generating the state ((|0i + |1i)/2)⊗ log m ⊗ |0i. Then,
complex strings one applies the unitary Ei (such that Ei |0i = |Ei (x)i)
to the last qubit, conditional on the state of the
From the analysis of the fingerprinting protocol, first qubits being |ii, i.e. the total operation being
we have seen that there must exist a complex (and |ii|0i 7→ |ii|Ei (x)i. This operation can be performed
therefore highly entangled) state of the form |hx i = by a circuit composed of single-qubit and CNOT gates,
m n whose description requires at most O((log m)2 ) bits
P
i=1 |ii|Ei (x)i, where x ∈ {0, 1} and Ei (x) is the i-
th bit of the encoding E(x) ∈ {0, 1}m of x ∈ {0, 1}n [39]. The state generated by this circuit is |hx i. The
using code E (of the type described in section II B). complexity of the state |hx i is of course bounded by the
While it is not necessarily true in general, there are complexity of the circuit just described.
cases in which the codes E have the additional property
ε 1 1
of being algorithmically simple, i.e. such that KCl (E) = KNet (|hx i) ≤ 2 log+ (log m)2 log + KCl (E(x))
O(log(n)). This is true, e.g., for Justesen codes [26]. ε ε
When such additional condition is satisfied, it is possible = O(KCl (E(x))),
to prove an important relationship between the classical (12)
Kolmogorov complexities of x, E(x), and the quantum
(network) complexity of the states |hx i. More precisely, when regarded as a function of m. The first term takes
B,ε into account the possibility that the Hadamard and the
one finds that KCl (x) = Θ(KCl (E(x))) = Θ(KNet (|hx i)).
NOT gates might not be included in the gate basis B.
The proof of this statement consists of two parts. In the
Note that this contribution is constant in m. The second
first part (Observation 1) we show that the complexity
term takes into account the description of the subcircuit
of the encoded word E(x) coincides (up to a constant)
in which each gate Ei is applied conditional on the state
with that of x. In the second part (Proposition 1) we
of the first log m qubits. The term KCl (E(x)) is the one
prove that the network complexity of the quantum state
that specifies when Ei = I and when, instead, Ei = NOT.
|hx i is essentially the same as the classical Kolmogorov
Note that this proof remains valid even if the code E is
complexity of E(x).
not algorithmically simple. In the following we see that
Observation 1. If E is an algorithmically simple code, this property of the code is needed in order to prove the
and E(x) ∈ {0, 1}m is the encoding of x ∈ {0, 1}n , then reverse statement.
ε
KCl (E(x)) = Θ(KCl (x)) We show now that KCl (E(x)) = O(KNet (|hx i))
Let ω h be the classical description of the circuit that
ε
Proof. To prove the statement, we show that prepares |hx i, such that KCl (ωh ) = KNet (|hx i), and con-
KCl (E(x)) = O(KCl (x)) and KCl (x) = O(KCl (E(x))). sider the following program.
The first of the two equalities follows immediately
by the fact that the code E is algorithmically sim- 1. Read ωh .
ple. To describe E(x), in fact, it is sufficient to give 2. Simulate the circuit and compute the approximate
the program for E, and a description of x. Thus, state |h̃x i.
KCl (E(x)) ≤ KCl (E) + KCl (x) = O(KCl (x)).
In order to prove the reverse statement, we consider that 3. For i from 1 to m, compute hi|h̃x i .
x can be described as the string whose encoding is E(x).
Thus it is possible to prepare x by allowing a computer This is a purely classical program that, simulating the
to generate all possible strings E(y) (for y ∈ {0, 1}n) (exact) creation of the quantum state |hx i and a num-
and compare each of them with E(x), halting when ber of operations on it, allows to prepare the classical
E(y) = E(x), i.e., when y = x. This implies that bit string E(x). The length of this program can be esti-
KCl (x) ≤ KCl (E(x)) + KCl (E) = O(KCl (E(x))). mated as follows. The first step requires O(KCl (ω h )) +
log KCl (ω h ) bits; the second contributes a quantity that
Proposition 1. Let E be an algorithmically simple is constant in m and depends only on the precision with
error-correcting code, such that d(E(x), E(y)) ≥ (1 − which we require the simulation to be effectuated. The
∆)m, where E(x), E(y) ∈ {0, 1}m are the encodings last step gives a contribution that is O(log m). As this
of x ∈ {0, 1}n, y ∈ {0, 1}n respectively. Then, the will not in general be an optimal program, its length is
Pm
(log m + 1)-qubit state |hx i = i=1 |ii|Ei (x)i, where
an upper bound to the complexity of E(x).
ε
hi|ji = δij , has complexity KNet (|hx i) = Θ(KCl (E(x)))
for sufficiently small ε. KCl (E(x)) ≤ O(KCl (ω h )) + O(log n) + O(1) =
ε (13)
= O(KCl (ω h )) = O(KNet (|hx i)).
Proof. In order to prove the statement we
ε
will show that KNet (|hx i) = O(KCl (E(x))) and In the analysis of the program given above, we have
ε
KCl (E(x)) = O(KNet (|hx i)). Let us first prove that not yet taken into account the effect of the error ε in
ε
KNet (|hx i) = O(KCl (E(x))). Consider the following cir- the preparation of the quantum state |hx i. We do so
cuit, that acts on the initial state |0i⊗(log m+1) . Initially a now, assuming that the circuit described by ωh does
Hadamard gate is applied to each of the first log m qubits, not prepare |hx i, but rather some state |h̃x i such that
8
|hhx |h̃x i|2 ≥ 1 − ε. Without loss of generality we write can clearly see how it follows from basic considerations
√ P
|h̃x i = (1/ m) m i=1 αij |ji|Ẽi (x)i. As a consequence,
related to the known properties of Kolmogorov complex-
the program described above will yield as √ final output ity.
a string Ẽ(x), with hE(x), Ẽ(x)i ≥ m 1 − ε, where
Proposition 2. For any function f : {0, 1}n → {0, 1},
hE(x), Ẽ(x)i is the scalar product between the two m-
the gain in communication complexity that one can obtain
dimensional vectors whose coefficients (in an orthonor-
by using quantum instead of classical communication, if
mal basis) are given by E(x) and Ẽ(x) respectively. only one-way communication is allowed, is at most expo-
Since E is chosen to be algorithmically simple, it is nential, i.e.
possible to verify whether or not the final string Ẽ(x)
is part of the code. If it is not so, then it follows CCl (f ) ≤ O(2CQ (f ) ). (14)
immediately that there has been an error somewhere,
which, if ε is sufficiently small (depending on the er- Proof. The idea of the proof is to simulate the optimal
ror correcting code) can be corrected. The only case quantum protocol classically. Consider an optimal quan-
in which the program described above cannot correct it- tum communication protocol to evaluate f (x, y), and
self is when the output word Ẽ(x) is a codeword, but suppose that a single state of q qubits is sent. Such a
different from E(x). We show now that this event can state can be described by a classical sequence of at most
be excluded by choosing ε small enough. For any two O(2q ) bits, and any operation on the quantum state can
words F, G we have m − hF, Gi ≥ d(F, G). If F, G are be simulated. In this case, thus, CCl (f ) ≤ O(2CQ (f ) ). If
codewords we also have d(F, G) ≥ (1 − ∆)m, which im- instead of a single q-qubit state, k states of q ′ qubits
plies then that hF,√Gi ≤ m∆. As we have seen earlier, are sent (with kq ′ = q), they can be described by
hE(x), Ẽ(x)i ≥ m 1 − ε. Thus, choosing ε ≤ 1 − ∆2
′
classical sequences of at most O(2q ) qubits each, thus
′
ensures that Ẽ(x) cannot be a codeword. CCl (f ) ≤ O(k2q ) ≤ O(2q ) = O(2CQ (f ) ).
This property has two main consequences. The first is Note that a similar argument can be used to consider
that the existence of a complex state of the form |hx i fol- the more general case where also two-way communica-
lows immediately from the existence of a complex string. tion between Alice and Bob is allowed. Similarly as be-
The second is the fact that there exists a “recipe” that fore, one only has to consider the classical simulation of a
allows us to prepare complex states, or would allow us to quantum protocol. However, here it could be that Alice
do so if we could have a classical complex string. does not send all the qubits to Bob which she uses to
solve the problem.
Corollary 1. For any m ≥ 1 there exists Pam (log m + 1)- An exponential quantum/classical gap in a quantum
qubit complex state of the form |hx i = i=1 |ii|Ei (x)i, communication problem, is obtained only when in Eq.
where hi|ji = δij , xi ∈ {0, 1}, and E is an error- (14) we have an equality (an exponential gap, in fact,
correcting code of the type introduced above. requires CQ (f ) ≤ O(log CCl (f ))). This can be obtained
only when at least one of the states that appear in the
quantum protocol is complex (that is, its classical de-
scription grows exponentially with the number of qubits).
V. APPLICATIONS FOR QUANTUM This implies that the power of quantum communication
KOLMOGOROV COMPLEXITY can be explained (mainly) by quantum entanglement. To
see this, let us use for instance the network complexity
In the following we apply the results on quantum Kol- (Eq.(5)). In [15] it has been proved that the entanglement
mogorov complexity to prove some statements from the of a quantum state, in terms of its Schmidt measure, is
theory of communication and computation complexity. an upper bound for its complexity. This implies that in
While some of the results are known, we underline that all communication complexity problems where one sees
the proofs are simpler than those found in literature. Fi- an exponential classical/quantum gap, maximally entan-
nally we see how the definition of network complexity can gled states must be created. This seems to imply that the
be applied to some aspects in thermodynamics, allowing source of the quantum advantage is in fact entanglement.
us to generalize existing results.
B. Computation complexity
A. Communication complexity
Quantum Kolmogorov complexity can also be used as
Let us first show that the exponential upper bound on a powerful tool to gain insight on some issues related to
the Kolmogorov complexity of quantum states implies computation complexity theory. We use it here to de-
that any quantum communication protocol in the SMP termine some properties of the states that appear during
model, requiring q qubits of communication, can be sim- the computation.
ulated using at most O(2q ) classical bits of communica- Let us describe a quantum algorithm by a sequence of
tion. Such result was already known [28], but here we single- and two-qubit operations, Ait ,jt , where it and jt
9
denote the two qubits on which the operation is acting undoubtfully a fundamental resource, it is not sufficient:
during the t-th time step (if it = jt Ait ,jt denotes a single only particular types of entanglement are, in fact, useful.
qubit operation acting on qubit it ). In the t-th step of
the algorithm, the state |Φt−1 i is transformed to |Φt i =
⊗n
Ait ,jt |Φt−1 i, with |Φ0 i = |0i . We have the following C. Kolmogorov complexity and thermodynamics
Caves continued the investigation of the Maxwell‘s de- mation required to fully describe the system [20]. In the
mon in the quantum setting [10, 11]. He studied the example considered above, the background information
change of the total entropy, defined by Zurek as S = S+I, is the list of states S ≡ {cos θk |0i + sin θk |1i | θk =
2m −1
where S denotes the statistical entropy and I the al- k 2πm }k=0 together with the uniform probability distri-
gorithmic information. Here, the algorithmic informa- bution [42].
tion is composed of two parts, the background informa-
tion, which is sufficient to generate a list of all states In the following we shall choose a different point of
(up to a certain precision), and the additional informa- view, for two main reasons. First, while it is clear that a
tion, needed to describe a particular element of this list state can always be described with the background infor-
of states (see Section II). In his considerations, Caves mation and the conditional information, this might not
chooses the background information to be negligible, im- always be the most convenient description. Second, we
plying that only the conditional algorithmic information see that with this approach it is very hard to go from
contributes to the total entropy. one system to composite systems. Consider the example
In order to demonstrate that it is indeed possible to studied above for the case where the demon tries to mea-
conceive situations in which an intelligent demon could sure n photons. A possible generalization of the back-
extract work from a system (i.e. increase the system’s ground information in the case of n systems is the set
total entropy) Caves considered the following situation. S ⊗n . In this case one would still be allowed to consider
A demon, whose system is prepared in some standard the background information negligible. At the same time,
product state denoted by |si = |s, . . . , si, is acting on a though, the demon does not have the possibility to choose
single photon, which is initially prepared in equilibrium, an arbitrary measurement direction with arbitrary good
ρs = 1l/2. The total entropy of the initial state is there- precision, but is instead forced to project the global sys-
fore S in = S(ρs ) + S(|si hs|) = S(ρs ) + I((|si hs|) = 1 tem onto product states. The only situation in which the
[41]. The main idea to increase the total entropy of demon could measure along arbitrary directions would
the closed system is to project the state of the photon be if the set of its available states increased with the di-
onto an algorithmically complex state. To this aim the mension of the system. In fact, as seen before, only if
demon randomizes his register (apart from one bit) to his background information increased exponentially with
obtain a complex string, r, of m bits. He associates the number of qubits the demon could indeed have the
to the bit string r the angle θ, where 0.0r is the bi- freedom to measure an arbitrary state. In this case it
nary representation (first m bits) of θ/π. The preci- seems to be impossible to neglect the contribution due
sion with which he can choose one of the 2m angles, to the background information.
m
{θk = k 2πm }2k=0−1 is therefore εθ = 2−m π. For any choice
of k (that is, r) he sets a polarizing beam splitter to mea- If the demon can only measure each single photon in-
sure the polarization of the photon in the basis {|Ψθk i ≡ dependently from all the others, the increase in total en-
cos(θk ) |0i + sin θk |1i , |Ψ̂θk i ≡ sin(θk ) |0i − cos θk |1i}. tropy is simply n times the one that we have in the case
Thus,P the state describing the system and the demon is of the single photon, i.e. ∆S = nm. However, if the de-
ρ = mon projects the system onto maximally complex (and
k (ρθk ⊗ |0i h0|s ⊗ |Ψθk i hΨθk | + ρ̂θk ⊗ |1i h1|s ⊗
thus necessarily fully entangled) states the increase of
|Ψ̂θk ihΨ̂θk |) ⊗ |ek i hek |d , where we have introduced an
the total entropy can be exponentially larger. The max-
auxiliary system, s that keeps track of information about
imum possible increase in the total entropy is given by
which arm the photon left the beam splitter and is avail- (Max)
able to the system. The auxiliary system d keeps track ∆S = Ifin − N Hin = 2n log 1ε − n, where ε gives the
of all the possible measurement choices. Here, the states precision with which the demon prepares its states.
|ek i denote a orthonormal basis, and ρθk = hΨθk |ρS |Ψθk i
and ρ̂θk = hΨ̂θk |ρS |Ψ̂θk i are not normalized. Thus, the Note that for a general definition of the quantum Kol-
final state describing the photon is given by the reduced mogorov complexity the background information is im-
state of ρ, i.e. by tracing over system D, d. Observing the plicitely accounted for. This is why then it is no problem
measurement outcome the demon stores this additional to consider the projection onto complex states. For in-
bit of information and thus holds the bit string 0.0r (or stance using the network definition, the background in-
0.1r) respectively. From the demons point of view, the formation is a basis consisting of single-qubit and two–
system is after the measurement in a pure state, |Ψθk i, qubit gates, together with an initial tensor-product state.
In this case the background information remains the
or |Ψ̂θk i. The total entropy of the whole system (w.l.o.g.
same, and therefore negligible, when increasing the num-
we assume that 0.0r has been measured) is therefore
ber of considered qubits. In the case of the CBE complex-
S fin = S(|Ψθk i) + I(θk = 0.0r) = I((0.0r)) = m + 1.
ity, instead, the background information consists with the
Therefore, the total entropy is increased by ∆S = m,
specification of the computational basis: which is indeed
which implies that the work W (+) = mkB T ln 2 has been
algorithmically simple. This is why e.g. the network, or
extracted.
the CBE, complexity of a state |Ψi is indeed the number
An important point in these considerations is the pres- of bits required to describe the state and no additional
ence of the background information, which is the infor- information is required.
11
[1] C. E. Shannon, A mathematical theory of communication, [16] H. Buhrman, R. Cleve, J. Watrous and R. de Wolf, Quan-
Bell Sys. Tech. Journal 27, 379 (1948). tum Fingerprinting, Phys. Rev. Lett. 87, 16 (2001).
[2] A. N. Kolmogorov, Three Approaches to the Quantitative [17] C. Mora and H. Briegel, Algorithmic complexity of quan-
Definition of Information, Probl. Inf. Transmission, 1, 1 tum states, quant-ph/0412172, to appear in IJQI.
(1965). [18] L. A. Levin, On the notion of a random sequence, Soviet
[3] G. J. Chaitin, On the length of programs for computing Mathematics Doklady 14, 1413 (1973).
finite binary sequences, J. ACM 13, 547 (1966). [19] W. H. Zurek Algorithmic randomness and physical en-
[4] M. Li, and P. Vitanyi, An Introduction to Kolmogorov tropy, Phys. Rev. A 40, 4731 (1989).
Complexity and Its Applications, Springer (1997). [20] C. M. Caves and R. Schack, Unpredictability, informa-
[5] T. M. Cover, and J. A. Thomas, Elements of Information tion, and chaos, Complexity 3, 4657 (1997).
Theory, John Wiley and Sons (1991). [21] J. Eisert and H. J. Briegel, Phys. Rev. A 64 (2001).
[6] C. H. Bennett, Complexity in the universe, in “Physical [22] A. C.-C. Yao, in Proceedings of the 11th Annual ACM
origins of time asymmetry”, Cambridge university press Symposium on Theory of Computing (ACM, New York,
(1994). 1979), p. 209.
[7] K. Gödel, ber formal unentscheidbare Stze der Principia [23] R. de Wolf Quantum computing and communication com-
Mathematica und verwandter Systeme, I, Monatshefte fr plexity, PhD thesis.
Mathematik und Physik 38, 173 (1931). [24] A. Ambainis, Algorithmica 16, 298 (1996).
[8] C. H. Bennett, The thermodynamics of computation – a [25] I. Newman and M. Szegedy, in Proceedings of the
review, Int. J. Theor. Phys. 21, 305 (1982). 28th Annual ACM Symposium on Theory of Computing
[9] W. H. Zurek, Thermodynamic cost of computation, algo- (ACM, New York, 1996), p. 596.
rithmic complexity and the information metric, Nature [26] J. Justesen, A class of constructive asymptotically good
341 119 (1989). algebraic codes, IEEE Trans. Inform. Theory 18, 652
[10] C. M. Caves, Information and entropy, Phys. Rev. E 47, (1972)
6 (1993). [27] M. A. Nielsen and I. L. Chuang, Quantum computation
[11] C. M. Caves, Information, Entropy and Chaos, in “Phys- and quantum information, Cambridge university press
ical origins of time asymmetry”, Cambridge university (2000).
press (1994). [28] Y. Shi and Y. Zhu, Proc. of the 37th ACM Symposium
[12] P. Gács, Quantum Algorithmic Entropy, arXiv: quant- on Theory of Computing, p.460 (2005)
ph/0011046 v2 (2001). [29] R. Jozsa and N. Linden, Soc. Lond. A, 459, 2011 (2003).
[13] A. Berthiaume, W. van Dam, and S. Laplante, Quantum [30] Maxwell’s demon, 2: entropy, classical and quantum in-
Kolmogorov Complexity, arXiv: quant-ph/005018 (2000). formation, edited by H.S.Leff and A.F. Rex, IOP pub-
[14] P. Vitanyi, Three Approaches to the Quantitative Def- lishing (2003).
inition of Information in an Individual Pure Quantum [31] R. Landauer, Irreversibility and heat generation in the
State, arXiv: quant-ph/9907035 (2000). computing process, IBM Journal of Research and Devel-
[15] C. Mora and H. Briegel, Algorithmic complexity and en- opment 5, 183 (1961).
tanglement of quantum states, Phys. Rev. Lett. 95, 20 [32] Analogously, any definition of the quantum Kolmogorov
(2005). complexity of an n-qubit state will have a log n term that
12
specifies the number of qubits of the states and that can correcting code are almost orthogonal.
be removed by considering the conditional complexity. [39] This second part of the circuit can be described as fol-
[33] Note that this is usually the approach taken in classical lows. For i that goes from 0 to m − 1, repeat the fol-
theory, where the background information of S is always lowing instructions. First apply a NOT gate to all the
small. qubits whose position corresponds to a zero in the binary
[34] As in [17], here and in the following we will say that description of i. Second, apply the controlled-unitary
circuit C B,ε prepares |ϕi with precision (or accuracy) ε if C log m (Ei ) (as defined in [27]): this controlled unitary
hϕ|C B,ε (|0i⊗n )|2 ≥ 1 − ε has a very simple description (see, e.g., [27]). Finally ap-
[35] Note that the same argument holds when direct commu- ply NOT gates to all the qubits that were modified in the
nication between Alice and Bob is allowed. In this case, first step (and reset all work qubits to zero)
though, the bound is n, and not 2n, since only one of the [40] For a comprehensive analysis of the problem, and issues
parties need to communicate his/her input related to it, we refer to [30] and references therein.
[36] Such a compression, though, might be useful if one is [41] Here it is assumed, as mentioned above, that the algo-
interested in other scenarios, or if the domain of x and y rithmic information of the equilibrium state is 0.
is restricted in some way. [42] Equivalently, one might consider as background informa-
[37] An example, for instance, is given by Justesen codes. In tion the two basis states {|0i, |1i} and the set of an-
π
this case one may choose any c > 2 and ∆ < 9/10 + gles {θk = k m }k , where one agrees that the specifica-
1/(15c) tion of an angle θk implies preparing a state of the form
[38] This is clear as all the states coded by the same word cos θk |0i + sin θk |1i.
are almost parallel, while those who appear in the error