Qubit: the quantum of information | Random physics                                                                             http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.
html
                            Random physics (/)
                            Alberto Verga, research notebook
                            blog (/) | about (/pages/about.html) lectures (/pages/lectures.html)
                                Lectures (/pages/AQ-index.html) on advanced quantum mechanics
                            Qubit: the quantum of information
                            There is a deep difference between the classical unit of information and its quantum counterpart. A qubit
                            corresponds to the state of a quantum system having two levels (a hilbert space of dimension 2); therefore, the
                            qubit is physical. A bit, or binary digit, is related to a logical state, true or false; its physical implementation as a
                            system having two equilibrium states, is not essential in its definition.
                            In 1995, Schumacher1 demonstrated the quantum version of the noiseless coding theorem of Shannon (see the
                            lecture on classical information (/pages/AQ-classical.html)). He showed that the von Neumann entropy is the
                            mean number of qubits necessary to encode the states of an ensemble or composite quantum system. Therefore,
                            a quantum state carries an amount of information that can be measured using the von Neumann entropy.
                            A general qubit, a two dimensional ket |ψ⟩ ∈  (dim = 2), can be written using the spherical polar
                            φ ∈ (0, 2π) and azimutal θ ∈ (0, π) angles:
                                                                                   θ              θ
                                                                      |ψ⟩ = cos      |0⟩ + eiφ sin |1⟩
                                                                                   2              2
                            where {|0⟩, |1⟩} is the canonical basis of  . This representation identifies the qubit state as a direction on the
                            unit sphere, so called the Bloch sphere. The north pole corresponds to |0⟩ and the south pole to |1⟩. Orthogonal
                            vectors are opposite points in the bloch sphere: (θ → π − θ, ϕ → −φ) .
1 of 15                                                                                                                                                         1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                      http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                Bloch sphere: Hilbert space of one qubit.
                            The density matrix of one qubit is ρ = |ψ⟩⟨ψ|, or using the bloch parametrization:
                                                                ( eiφ sin θ/2 ) (
                                                                    cos θ/2
                                                            ρ=                    cos θ/2 e−iφ sin θ/2 )
                                                                2 ( eiφ sin θ 1 − cos θ )
                                                                1 1 + cos θ e−iφ sin θ
                                                              =
                            or equivalently
                                                            1+n⋅σ
                                                       ρ=         ,         n = (sin θ cos φ, sin θ sin φ, cos θ) ,
                                                              2
                            where σ = (X, Y, Z) is the vector of pauli matrices (https://en.wikipedia.org/wiki/Pauli_matrices). This form
                            allows a compact representation of a general mixed state of qubits:
                                                                            1 + rn ⋅ σ
                                                                      ρ=               ,    r ≤ 1,
                                                                                2
2 of 15                                                                                                                                                  1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                         http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                            where r = 1 is the pure state and r < 1 a mixed state, which lies then inside the unit bloch sphere.
                            The hilbert space of a system of N qubits, is the tensor or Kronecker product of one-qubit spaces:
                                                                                        N
                                                                               =              n
                                                                                       ⨂
                                                                                       n=1
                            and
                                                       |s1 ⟩ ⊗ |s2 ⟩ ⊗ … × ⊗|sN ⟩ = |s1 s2 … sN ⟩,           sn = 0, 1 ,
                            is an element of the canonical basis; sometimes it is convenient to write simply |s⟩, where s is the decimal
                            representation of the binary string s1 … sN :
                                                                                      2N −1
                                                                                      ∑
                                                                              |ψ⟩ =           ψs |s⟩ ,
                                                                                      s=0
                            is a general expression of a ket, where
                                                                                    |ψs |2 = 1 .
                                                                               ∑
                                                                                s
                            Important two qubit states are the Bell states:
                                                                                      |00⟩ ± |11⟩
                                                                          |Φ± ⟩ =
                                                                                          √2‾
                                                                                      |01⟩ ± |10⟩
                                                                          |Ψ± ⟩ =                        .
                                                                                              √2‾
                            von Neumann entropy and entanglement
                            One of the deepest facts about the quantum state was discovered by von Neumann in his investigation of the
                            measurement process: one can associate to a quantum state ρ an entropy:
3 of 15                                                                                                                                                     1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                             http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                                                           S(ρ) = − Tr ρ log ρ .
                            This equation defines a fundamental relationship between quantum states and information.
                            Partial trace
                            Consider a bipartite system AB whose hilbert space is  = A ⊗ B , the B partial trace TrB is a linear
                            operation mapping  to A , such that the expected value of an arbitrary operator OA satisfies:
                                                           ⟨OA ⟩ = Tr OA ρA = Tr(OA ⊗ 1B )ρ , ρA = TrB ρ ,
                            where ρ is AB state and ρA is the state of the subsystem A (note that the first trace is over A space, and the
                            second is on the whole space).
                            For instance the partial trace over a fourth order tensor Tkm,ln where the indices k, l = 1, … , L refer to space A
                            and the indices m, n = 1, … , N to space B , corresponds to the contractions over B indices, TrB T = TA with
                            TA (k, l) = Tkn,ln (using the einstein convention), or over the A indices TrA T = TB with TB (n, m) = Tkn,km .
                            Let us apply the partial trace to the Bell state |Ψ− ⟩, which is a state of two qubits A and B :
                                                                        1
                                             ρA = TrB |Ψ− ⟩⟨Ψ− | =        TrB (|01⟩⟨01| − |01⟩⟨10| − |10⟩⟨01| + |10⟩⟨10|)
                                                                        2
                                                  1
                                                =   (|0⟩⟨0|⟨1|1⟩ + |1⟩⟨1|⟨0|0⟩)
                                                  2
                                                  2(0 1)
                                                  1 1 0
                                                =
                            terms with ⟨0|1⟩ vanish; therefore the reduced density matrix is proportional to the identity. ρA corresponds
                            then to a mixed state of equally probable |0⟩ and |1⟩ states. We conclude that in general the state of a
                            subsystem of a pure state system is mixed.
                            We can compute the von Neumann entropy of the whole system and the one of its sybsystems:
                                                         S(AB) = − Tr ρ log ρ = − Tr |ψ⟩⟨ψ|ψ⟩ log(1)⟨ψ| = 0
4 of 15                                                                                                                                                         1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                         http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                            and
                                                                                   1        1
                                                       S(A) = − Tr ρA log ρA =       log 2 + log 2 = log 2 = 1 .
                                                                                   2        2
                            The entropy of the pure state is 0 , while the entropy of its subsystems is S(A) = S(B) = 1 , the maximum
                            possible value of a twodimensional system. If we apply a classical reasoning to understand this result we arrive
                            at a contradiction. The entropy of a given system is zero, meaning that it is perfectly ordered, or analogously
                            that we have a perfect knowledge of its state (the equivalence of the two statements derives from the
                            equivalence of the entropy and information), however, the entropy of its subsystems is maximal, meaning that
                            its subsystems are perfectly disordered. This contradiction was first put forward by Schrödinger in his famous
                            response to Einstein, Podolsky and Rose:
                                  the best possible knowledge of a whole does not necessarily include the best possible knowledge of all
                                  its parts, even though they may be entirely separated and therefore virtually capable of being “best
                                  possibly known” 2
                            Therefore the Bell state contains correlations that cannot be accounted by classical laws (in this respect we
                            discuss later the Bell theorem), this purely quantum information resource is the entanglement inherent of
                            interacting multiparticle quantum states, a concept first introduced by Schrödinger in the previous paper.
                            We observe that the von Neumann entropy can be used to measure the entanglement of quantum states. We will
                            illustrate the entanglement as an information resource using a few simple quantum circuits. Before that we
                            introduce the basic elements of a quantum circuits, the quantum gates.
                            It is also worth mentioning that the reduced density matrix of a pure state corresponds to a mixed state. This is a
                            general fact, one can always enlarge the hilbert space of a composite system in a mixed state ρ̄ , in such a
                            way that
                                                                 ρ̄ = TrA ρ ,   ρ = |ψ⟩⟨ψ| ∈ ̄ ⊗ A ,
                            where the trace is over the ancillary space A .
5 of 15                                                                                                                                                     1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                         http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                            Entangled states
                            The hilbert space of a general quantum system  has the structure of a product of hilbert spaces corresponding
                            to its subsystems
                                                                           = 1 ⊗ 2 ⊗ …
                            For instance,  represents a set of qubits, of atoms in a condensate, or of spins in a lattice, for which each n
                            stands for a qubit, an atom or a spin, respectively. Often the ground state of these systems has a very simple
                            form: it can be factorized in a product of states of its individual components (or subsystems), like the
                            ferromagnetic phase in which all spins are up:
                                                                        |G⟩ = |0⟩ ⊗ |0⟩ ⊗ … ;
                            however, excitations at finite temperature and interactions build much more complex states that cannot be
                            written as a product of basic states belonging to each subsystem. Non factorizable states are entangled states.
                            The bell states are examples of non factorizable two qubit states.
                            Therefore, not entangled pure states of a bipartite system can be expressed as a product,
                                                                        |ψ ⟩AB = |ψ A ⟩ ⊗ |ψ B ⟩ .
                            The von Neumann entropy of the corresponding parties vanishes, S(A) = S(B) = 0 .
                            The density matrix of a two parties system can be decomposed in terms of orthonormal basis vectors, the
                            Schmidt decomposition. It is convenient to write a general pure bipartite state |ψ⟩,
                                                                             ∑
                                                                     |ψ⟩ =         xnA nB |anA ⟩ ⊗ |bnB ⟩
                                                                             A B
                                                                             n n
                            using its matrix representation X ,
                                                                                   |ψ⟩ ≐ X
                            where dimX = NA × NB , whose elements are
6 of 15                                                                                                                                                     1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                             http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                                                            xnA nB = ⟨ψ|anA bnB ⟩ ,
                            with |anA ⟩ and |bnB ⟩ are the basis vectors of the two parties; we assume NA ≤ NB . Consequently, the reduced
                            density matrix of subsystem A is simply the NA × NA matrix,
                                                                                  ρA ≐ X X † ,
                            and similarly for B :
                                                                                  ρB ≐ X † X .
                            Let us diagonalize ρA ,
                                                            ρA ≐ UPU † ,         P = diag(pnA ), dimP = NA ,
                            where
                                                                   ∑
                                                              U=         |unA ⟩⟨anA | ,          ρA |unA ⟩ = pnA |unA ⟩
                                                                   A
                                                                    n
                            is the unitary that transforms the |a⟩ basis to the |uA ⟩ basis in which ρA is diagonal (note that ∑nA pnA = 1).
                            We can proceed similarly with the B subsystem, to obtain the reduced matrices in the form:
                                                                        pn |uAn ⟩⟨uAn | ,                    pn |uBn ⟩⟨uBn |
                                                                   ∑                                    ∑
                                                            ρA =                                 ρB =
                                                                    n                                    n
                            where pn are the nonzero eigenvalues (common to both A and B ), and |uBn ⟩ the corresponding NA eigenvectors
                            of X † X (whose numbers is N ≤ NA ). In conclusion, we can rewrite the pure state in the form:
                                                                                               p |uA ⟩ ⊗ |uBn ⟩ ,
                                                                                            ∑ √‾‾n n
                                                              |ψ⟩ ≐ √P
                                                                     ‾‾,        |ψ⟩ =
                                                                                             n
                            which is just the schmidt decomposition. Note that the trick to represent the state by a matrix does not work for,
                            for example, a tripartite system; the Schmidt decomposition is thus only appropriate for systems divided in
                            two parts.
7 of 15                                                                                                                                                         1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                                 http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                            The schmidt decomposition of a bipartite pure state |ψAB ⟩ gives an explicit expression of the
                            entanglement entropy:
                                                                                     ∑
                                                                              SA =       pn log pn
                                                                                     n
                            where √p  ‾‾n are the schmidt coefficients in the expansion of |ψAB ⟩. Note that the SA contains information about
                            the full spectrum of the corresponding density matrix (in the schmidt basis).
                            Entropy properties
                                  S(ρ) ≥ 0 and is zero for pure states.
                                  The entropy of a uniformly mixed state ρ = 1d /d , where dim = d is the space dimension.
                                  The entropy of the reduced density matrix of a bipartite system AB are equal S(ρA ) = S(ρB ).
                                  Let pn be a probability distribution and |n⟩⟨n| orthogonal states of system A ; suppose ρn are states of B ,
                                  then
                                                                  (∑                     )                   ∑
                                                              S        pn |n⟩⟨n| ⊗ ρn        = H(pn ) +              pn S(ρn ) ,
                                                                   n                                         n
                                  where H is the Shannon entropy.
                                  The entropy is subadditive
                                                                               S(AB) ≤ S(A) + S(B)
                                  The entropy is concave
                                                                          ∑                   (∑             )
                                                                              pn S(ρn ) ≤ S          pn ρn       .
                                                                          n                    n
                            To prove the joint entropy formula we solve the eigenvalue equation ρn |ukn ⟩ = λkn |ukn ⟩, which leads to,
8 of 15                                                                                                                                                             1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                         http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                                                       (∑           )
                                                                                                  pn λkn log pn λkn
                                                                                             ∑
                                                                  S         pn ρn       =−
                                                                        n                    nk
                            from which we obtain
                                                                     (∑             )                ∑
                                                                 S          pn ρn       = H(pn ) +        pn S(ρn )
                                                                      n                               n
                            The subadditivity of the entropy can be proved using the Klein inequality:
                                                                       Tr(ρ log ρ) − Tr(ρ log σ) ≥ 0
                            for arbitrary density matrices ρ and σ .
                            Using the joint probability formula and its subadditivity one can demonstrate the concavity property.
                            Entanglement nonlocality
                            Some very basic properties of quantum states conflict classical physics. Einstein, Podolsky and Rosen
                            noted that
                                In quantum mechanics in the case of two physical quantities the knowledge of one precludes the
                                knowledge of the other.4
                            We verified this statement using the bell state and showing that specifying the reduced state of one qubit (say a
                            physical spin for concreteness) destroys the information on the other one. Thus, in apparent contradiction with
                            the fact that the whole system is in a pure state, and then perfectly known. “Apparent” because the physical
                            reality is quantum, not classical, as results form experiments on bell inequalities. 5
                            In fact, suppose as before that the two spin system is the |Ψ− ⟩ state; a measurement of ZA ⊗ 12 projects |Ψ− ⟩
                            to either |01⟩ or |10⟩ , and the result is 1 or −1 with probability 1/2; therefore, if one measures the second spin
9 of 15                                                                                                                                                     1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                          http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                            12 ⊗ ZB the results are similar. However, if the A spin is found to be in the up state 1 , the measure of the
                            second spin B gives −1. (Note that these results do not depend on the order of the measurement.) Leading thus
                            to a perfect anti-correlation of both measurements. The question posed by Einstein, Podolsky and Rosen is
                            whether this correlation arises from some hidden reality.
                            Bell devised a test to verify if the supposed “uncertainty” of quantum mechanics is only due to the existence of
                            a classical underlying physical mechanism, described by a set of hidden variables. We follow here the version of
                            the Bell theorem proposed by Clauser et al.6
                            We consider two agents Alice A and Bob B . We discuss first the classical approach. Assume a source ot two-
                            particle states. Alice measures the A1 and A2 , two physical quantities related to the first particle and taking the
                            values a1 , a2 = −1, 1; Bob measures the second particle properties B1 and B2 taking the values
                            b1 , b2 = −1, 1. We also assume, and this is the connexion with the hidden variables, the source produces its
                            output following some rule to which one can associate certain probability distribution P(a1 , a2 ; b1 , b2 ).
                            (Remark that a deterministic rule is a special case of probability one.) Therefore, two particle correlations are
                            given by the expectations ⟨ai bj ⟩ with i, j = 1, 2 (over the probability P ). Therefore, the quantity S EX
                                                                 S = ⟨a1 b1 ⟩ + ⟨a1 b2 ⟩ + ⟨a2 b1 ⟩ − ⟨a2 b2 ⟩
                            satisfies the classical inequality
                                                                                   S ≤ 2.
                            We discuss now the quantum approach. The analog quantum setup is a two spin system whose observables are,
                                                                            A1 = XA , A2 = YA
                            and
                                                                            B1 = XB , B2 = YB ,
                            and the quantum state produced by the source is
10 of 15                                                                                                                                                     1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                        http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                                                            1         eiπ/4
                                                                     |ψ⟩ =     |00⟩ +       |11⟩.
                                                                           √2‾        √‾2
                            A simple exercise show that the quantum correlation EX:
                                                                             S = 2√2‾ > 2
                            in contradiction with the classical expectation. The experimental result refutes the hidden variable hypothesis.
                            In addition, quantum correlations are stronger than classical ones, and their nature is nonlocal: independent
                            measurements of spatially separated qubits can be perfectly correlated. The origin of purely quantum
                            correlations, with no classical equivalent, is the entanglement of composite quantum states.
                            As an aside remark, let us note that the impossibility to account for quantum properties in terms of classical
                            physics is often the source of the misleading statements, sometimes presented as “quantum paradoxes”: they are
                            much “classical paradoxes” than quantum ones. What we must retain here is that entanglement is a quantum
                            resource of information, with the potential to overcome classical constraints or limitations.
                            Exercises
                               1. Prove the entropy properties.
                               2. Show that the measurement of the two qubits of a bell state on an arbitrary direction n , leads to perfectly
                                  correlated results.
                            Useful formulas
11 of 15                                                                                                                                                   1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                           http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                Convex function f (x).
                                  Gibbs inequality:
                                                                         ∑                   ∑
                                                                     −       pn log pn ≤ −       pn log qn
                                                                         n                   n
                                  where p and q are two probability distributions.
12 of 15                                                                                                                                      1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                           http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                 Jensen inequality:
                                                                                 f (⟨x⟩) ≤ ⟨f (x)⟩
                                 if f is convex; in general, for positive pn and f convexity implies,
                                                                            (∑            )        ∑
                                                                        f         pn xn        ≤       pn f (xn )
                                                                             n                     n
                                 Remember that a function f is convex if for l ∈ [0, 1],
                                                                f (λx1 + (1 − λ)x) ≤ λf (x1 ) + (1 − λ)f (x2 )
                                 (then −f is concave, see figure above).
                                 Klein inequality:
                                                                                               ρ
                                                                                 Tr ρ log        ≥0
                                                                                               χ
                                 where ρ and χ are two arbitrary density matrices; the quantity on the left is the relative entropy. Indeed,
                                 using well suited orthonormal basis,
                                                                      ∑                                ∑
                                                                 ρ=         pk |pk ⟩⟨pk |,     χ=           ql |ql ⟩⟨ql |,
                                                                       k                                l
                                 we write,
                                                                                               pk |⟨ql |pk ⟩|2 log ql
                                                                                          ∑
                                                                  − Tr ρ log χ = −
                                                                                          kl
                                 or equivalently, noting that Xlk = |⟨ql |pk ⟩|2 is a double stochastic matrix (their marginal sums are
                                 probability distributions),
13 of 15                                                                                                                                                      1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                        http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                                                                                     ∑             (∑                 )
                                                                 − Tr ρ log χ = −             pk         Xlk log ql
                                                                                         k          l
                                 from which, using the jenssen inequality, we obtain,
                                                                                   ∑              ∑       (∑        )
                                                              − Tr ρ log χ ≥ −               pk        log   Xlk ql
                                                                                     k             l       l
                                 We apply now the gibbs inequality to the last term (we can put σk = ∑l Xlk ql , with ∑k σk = 1),
                                                                 ∑        ∑      (∑       )    ∑ k
                                                             −       pk       log   Xlk ql ≥ −   p log pk
                                                                 k        l       l            k
                                 to arrive at
                                                                          − Tr ρ log χ ≥ − Tr ρ log ρ .
                            Notes
                               1. B. Schumacher, Quantum coding, Phys. Rev. A 51, 2738 (1995). .pdf (/pdfs/Schumacher-1995uq.pdf) ↩
                               2. E. Schrödinger, Discussion of probability relations between separated systems, Mathematical Proceedings
                                  of the Cambridge Philosophical Society 31, 555 (1935) .pdf (/pdfs/Schrodinger-1935.pdf) ↩
                               3. A. Barenco et al., Elementary gates for quantum computation, Phys. Rev. A 52, 3457 (1995). Universality
                                  of the set one qubit and controlled not gates. ↩
                               4. A. Einstein, B. Podolsky, and N. Rosen, Can Quantum-Mechanical Description of Physical Reality Be
                                  Considered Complete? Phys. Rev. 47, 777 (1935) ↩
                               5. A. Aspect, Closing the door of Einstein and Bohr quantum debate, Physics (APS) 8, 123 (2015) .pdf
                                  (/pdfs/Aspect-2015.pdf) ↩
14 of 15                                                                                                                                                   1/21/22, 13:26
Qubit: the quantum of information | Random physics                                                                    http://verga.cpt.univ-mrs.fr/pages/AQ-qubit.html
                               6. J. Clauser, M. Horne, A. Shimony, and R. Holt, Proposed Experiment to Test Local Hidden-Variable
                                  Theories, Phys. Rev. Lett. 23, 880 (1969) ↩
                            Archives (/archives.html)                              Pelican (https://github.com/getpelican/pelican) /
                            Categories (/categories.html) Tags (/tags.html)        (https://getbootstrap.com) /
                                                                                   (http://creativecommons.org/licenses/by-nc/4.0/) CC
                                                                                   4.0
15 of 15                                                                                                                                               1/21/22, 13:26