You are on page 1of 5

Mutually Unbiased Bases are Complex Projective

2-Designs
Andreas Klappenecker Martin Rötteler
Department of Computer Science NEC Labs America, Inc.
Texas A&M University 4 Independence Way
College Station, TX, 77843–3112, USA Princeton, NJ 08540 U.S.A.
Email: klappi@cs.tamu.edu Email: mroetteler@nec-labs.com

Abstract— Mutually unbiased bases (MUBs) are a primitive that we want to determine the density matrix ρ of an ensemble
used in quantum information processing to capture the principle of quantum systems using as few non-degenerate observables
of complementarity. While constructions of maximal sets of d + 1 as possible. We assume that it is possible toPmake a com-
such bases are known for system of prime power dimension d, it is
unknown whether this bound can be achieved for any non-prime plete measurement of each observable O = b∈B xb |bihb |,
power dimension. In this paper we demonstrate that maximal sets meaning that the statistics tr(ρ |bihb |) = hb| ρ |bi is known for
of MUBs come with a rich combinatorial structure by showing each eigenvalue xb in the spectral decomposition. Ivanović
that they actually are the same objects as the complex projective showed in [12] that complete measurements of at least d + 1
2-designs with angle set {0, 1/d}. We also give a new and observables are needed to reconstruct the density matrix. He
simple proof that symmetric informationally complete POVMs
are complex projective 2-designs with angle set {1/(d+1)}. also showed that this lower bound is attained when d + 1 non-
degenerate pairwise complementary observables are used.
I. I NTRODUCTION A simple example is provided by the Pauli spin matrices σx ,
Two quantum mechanical observables are called comple- σy , σz . A complete measurement of these three observables
mentary if and only if precise knowledge of one of them allows to reconstruct a 2 × 2 density matrix, a fact appar-
implies that all possible outcomes are equally probable when ently known to Schwinger [18]. Nowadays, we know how to
measuring the other, see for example [19, p. 561]. The do this state tomography process—at least in principle—in
principle of complementarity was introduced by Bohr [6] in dimensions d = 3, 4, and 5. It is an open problem whether
1928, and it had a profound impact on the further development it is possible to perform this kind of state tomography in
of quantum mechanics. A recent application is the quantum dimension 6, because the construction of a set of 7 mutually
key exchange protocol by Bennett and Brassard [3] that unbiased bases in dimension d = 6 is elusive.
exploits complementarity to secure the key exchange against
eavesdropping. II. M UTUALLY U NBIASED BASES
We mention a simple mathematical consequence of this
complementarity principle, which motivates some key notion. Definition 1: Two orthonormal bases B and C of Cd are
Suppose that O and O0 are two hermitian d × d matrices called mutually unbiased iff |hb|ci|2 = 1/d holds for all b ∈ B
representing a pair of complementary observables. We assume and c ∈ C.
that the eigenvalues of both matrices are multiplicity free. The goal is to construct d + 1 mutually unbiased bases
It follows that the observables O and O0 respectively have (MUBs) in any dimension d ≥ 2. There are several con-
orthonormal eigenbases B and B 0 with basis vectors uniquely structions known to obtain MUBs. At least for prime power
determined up to a scalar factor. dimension the problem is completely solved. This follows
The complementarity of O and O0 implies that if a quantum from Constructions I-III below. However, in any dimension
system is prepared in an eigenstate b0 of the observable O0 , other than a prime power it is unknown if a maximal set
and O is subsequently measured, then the probability to find of d + 1 MUBs can be found. The best known result is
the system after the measurement in the state b ∈ B is given Construction IV below which only works in dimensions d
by |hb|b0 i|2 = 1/d. Recall that two orthonormal bases B and which are squares and never gives a maximal set of MUBs.
B 0 of Cd are said to be mutually unbiased precisely when Construction I (Wootters and Fields [24]) Let q be an odd
|hb|b0 i|2 = 1/d holds for all b ∈ B and b0 ∈ B 0 . Thus prime power. Define
the eigenbases of non-degenerate complementary observables
2
are mutually unbiased. Conversely, we can associate to a |va,b i = q −1/2 (ωptr(ax +bx)
)x∈Fq ∈ Cq ,
pair of mutually unbiased bases a pair of non-degenerate
complementary observables. with ωp = exp(2πi/p). Then the standard basis together with
There is a fundamental property of mutually unbiased bases the bases Ba = {|va,b i |b ∈ Fq }, a ∈ Fq , form a set of q + 1
that is invaluable in quantum information processing. Suppose mutually unbiased bases of Cq .
Construction II (Galois Rings [13]) Let GR(4, n) be a finite Then we have that:
Galois ring with Teichmüller set Tn . Define r r
• M (p ) = p + 1 for p prime, r ∈ N,
• M (n) ≤ n + 1 for all n ∈ N,
  
−n/2 2πi
|va,b i = 2 exp tr(a + 2b)x . • M (mn) ≥ min{M (m), M (n)} for all m, n ∈ N.
4 x∈Tn 2
• M (d ) ≥ N (d), where N (d) is the number of mutually
Then the standard basis together with the bases Ma = orthogonal Latin squares of size d × d.
{|va,b i |b ∈ Tn }, a ∈ Tn , form a set of 2n + 1 mutually An open problem is to show that lim inf n→∞ M (n) = ∞.
n
unbiased bases of C2 .
Construction III (Bandyopadhyay et al. [1]) Suppose there III. W ELCH ’ S L OWER B OUNDS
exist subsets C1 , . . . , Cm of a unitary error basis B such that Suppose that X is a finite nonempty set of vectors of unit
|Ci | = d, Ci ∩ Cj = {1d } for i 6= j, and the elements of Ci norm in the complex vector space Cd . The vectors in X satisfy
pairwise commute. Let Mi be a matrix which diagonalizes Ci . the inequalities
Then M1 , . . . , Mm are MUBs. 1 X 1
|hx|yi|2k ≥ d+k−1 , (1)
Construction IV (Wocjan and Beth [23]) Suppose there are |X|2
x,y∈X k
w mutually orthogonal Latin squares [4], each of size d × d
over the symbol set S = {1, . . . , d}. Then w + 2 MUBs for all integers k ≥ 0. Welch derived these bounds in [22]
in dimension d2 can be constructed as follows. With each to obtain a lower bound on the maximal cross-correlation of
Latin square L (and additionally the square (1, 2, . . . , n)t ⊗ spreading sequences of synchronous code-division multiple-
(1, . . . , 1)) we can associate vectors of length d over the access systems. Blichfeld [5] and Sidelnikov [20] derived
alphabet {1, . . . , d2 }: for each symbol α ∈ S define a vector similar bounds for real vectors of unit norm.
sL,α ∈ Cd as follows: start with the empty list sL,α = ∅. A set X attaining the Welch bound (1) for k = 1 is
Then traverse the elements of L column-wise starting at the called a WBE-sequence set, a notion popularized by Massey
upper left corner. Whenever α occurs in position (i, j) in L, and Mittelholzer [15] and others. Using equation (1), it is
then append the number i + jd to the list sL,α . The other straightforward to check that the union of d + 1 mutually
ingredient to construct these MUBs is an arbitrary complex unbiased bases of Cd form a WBE-sequence set. These
Hadamard matrix H = (hi,j ) of size d × d. For each Latin extremal sets of mutually unbiased bases are even better, since
square L and each α, ∈ {1, . . . , d} define a normalized they also attain the Welch bound for k = 2. In fact, we
√j P d
vector |vL,α,j i := 1/ d i=1 esL,α [i] hi,j , where ei are the show that a sequence set attains the Welch bounds (1) for all
2
elementary basis vectors in Cd . Then the bases given by k ≤ t if and only if it is a t-design in the complex projective
BL := {|vL,α,j i : α, j = 1, . . . , d} together with the identity space CP d−1 .
matrix 1d2 form a set of w + 2 MUBs. Let us introduce some notation. Let S d−1 denote the sphere
of unit vectors in the complex vector space Cd . We say that
Example 1: In dimension d = 3 Construction I yields the
two vectors u and v of S d−1 are equivalent, in signs u ≡
bases
v, if and only if u = eiθ v for some θ ∈ R. It is easy to
3−1/2 { (1, 1, 1), (1, ω3 , ω32 ), (1, ω32 , ω3 )}, see that ≡ is an equivalence relation. We denote the quotient
3−1/2 {(1, ω3 , ω3 ), (1, ω32 , 1), (1, 1, ω32 )}, manifold S d−1/ ≡ by CS d−1 . Notice that the manifold CS d−1
is isomorphic to the complex projective space CP d−1 , but we
3−1/2 {(1, ω32 , ω32 ), (1, ω3 , 1), (1, 1, ω3 )}, prefer the former notation because normalizing vectors to unit
which together with the standard basis 13 form a maximal length is common practice in quantum computing.
system of four MUBs in C3 . Lemma 1: Let µ be the unique normalized U (d)-invariant
Example 2: In dimension d = 4 Construction II yields the Haar measure on the complex sphere CS d−1 . For any x ∈
bases (where√we have abbreviated “ + ” for 1 and “ − ” for S d−1 , we have
−1 and i = −1): Z
1
1 |hx|yi|2k dµ(y) = d+k−1 .
2 {(+, +, +, +), (+, +, −, −), (+, −, −, +), (+, −, +, −)}, CS d−1 k
1 Proof: The unitary group U (d) acts transitively on the
2 {(+, −, −i, −i), (+, −, i, i), (+, +, i, −i), (+, +, −i, i)},
1 manifold CS d−1 . This means that for any y ∈ CS d−1 there
2 {(+, −i, −i, −), (+, −i, i, +), (+, i, i, −), (+, i, −i, +)}, exists a unitary matrix U mapping y to the first basis vector,
1
2 {(+, −i, −, −i), (+, −i, +, i), (+, i, +, −i), (+, i, −, i)}. U y = e1 . Therefore,
Z Z
These four bases and the standard basis 14 form an extremal
|hx|yi|2k dµ(x) = |hU x|e1 i|2k dµ(x)
set of five MUBs in C4 . CS d−1 CS d−1

A basic question is how many bases can be achieved in


Z
general dimension. To this end, we define the function M : = |hx|e1 i|2k dµ(x),
CS d−1
N → N as follows:
where the last equality holds because of the U (d)-invariance
M (n) := max{|B| : B is a set of MUBs in Cn } of the measure µ. Using Proposition 1.4.9 from Rudin [17],
we obtain We show next that 2) implies 3). We observe that (2) holds
Z Z
1 for all k ≤ t, hence summing over x ∈ X yields (3).
|hx|e1 i|2k dµ(x) = |xk1 |2 dµ(x) = d+k−1
,
CS d−1 CS d−1 d−1
Finally, we show that 3) implies 1). Suppose that equation
(3) holds. For a vector x ∈ Cd , we denote by x⊗k the k-
which proves the claim. k
fold tensor product x⊗k = x ⊗ · · · ⊗ x ∈ Cd . Note that
IV. C OMPLEX P ROJECTIVE t-D ESIGNS hx⊗k |y ⊗k i = hx|yik . Consider the d2k -dimensional vector
Z
We now present some background material on complex 1 X ⊗k
ξ= x ⊗ x⊗k − x⊗k ⊗ x⊗k dµ(x).
projective designs. We will relate those later on to the systems |X| CS d−1
x∈X
of vectors formed by a maximal set of MUBs.
Let us first introduce some notation. We denote Evaluating the inner product of ξ with itself yields
by Hom(k, `) the subset of the polynomial ring 1 X
Z Z
2k
C[x1 , . . . , xd , y1 , . . . , yd ] that consists of all polynomials that |hx|yi| − |hx|yi|2k dµ(y)dµ(x), (4)
|X|2 CS d−1
are homogeneous of degree k in the variables x1 , . . . , xd x,y∈X

and homogeneous of degree ` in the variables y1 , . . . , yd . which is equal to hξ|ξi ≥ 0. The inner integral evaluates to
We associate to each polynomial p in Hom(k, `) a function d+k−1 −1

by Lemma 1, and the double integral has the same
p◦ on the sphere S d−1 by defining p◦ (ξ) = p(ξ, ξ) for k
value, because the measure µ is normalized. It follows from
ξ ∈ S d−1 . The function p◦ is called the “restriction” of p
our assumption that the right hand side vanishes. By con-
onto the complex sphere. It follows from the homogeneity
struction of ξ, we can conclude that averaging over X yields
conditions of the polynomial p that p◦ (eiϑ ξ) = eiϑ(k−`) p◦ (ξ)
an exact cubature formula for all monomials in Hom(k, k)◦ ,
holds for all ϑ ∈ R. Therefore, we obtain a well-defined
hence, by linearity, for all polynomials in Hom(k, k)◦ . This
polynomial function on CS d−1 only if k = `. We define
means that X is a t-design.
Hom(k, k)◦ = {p◦ : p ∈ Hom(k, k)}.
Remark 1: Equation (4) provides a short proof of the Welch
Definition 2: A finite nonempty subset X of CS d−1 is a
inequalities (1). The analogue for real spherical t-designs
t-design in CS d−1 iff the cubature formula
Z of the above result is sketched in [7]. A connection to the
1 X 1 existence of certain isometric Banach space embeddings is
f (x) = f (x)dµ(x)
|X| µ(CS d−1 ) CS d−1 given in [14].
x∈X

holds for all f in Hom(t, t)◦ . V. U NIFORM T IGHT F RAMES


We now show a characterization of t-designs in terms of A finite subset F of nonzero vectors of Cd is called a frame
the inequalities by Welch given in equation (1). if there exist nonzero real constants A and B such that
Theorem 1: Suppose that X is a finite nonempty subset of X
CS d−1 . Then the following statements are equivalent: Akvk2 ≤ |hf |vi|2 ≤ Bkvk2
f ∈F
1) The set X is a t-design in CS d−1 ;
2) for all x ∈ Cd and all k in the range 0 ≤ k ≤ t, we have holds for all v ∈ Cd . The notion of a frame generalizes the
the equality concept of an orthonormal basis. The linear span of the vectors
hx|xik 1 X in F the space Cd , but the vectors in a frame are in general
d+k−1
= |hx|yi|2k ; (2) not linearly independent. A frame is called tight if and only if
k
|X|
y∈X the frame bounds A and B are equal. A tight frame is called
3) the set X satisfies the Welch bounds (1) with equality for isometric (or uniform) if and only if each vector in F has unit
all k in the range 0 ≤ k ≤ t, that is norm.
Theorem 2: Let F be a finite nonempty subset of vectors
1 X 1
|hx|yi|2k = d+k−1 , 0 ≤ k ≤ t. (3) in Cd . The following statements about F are equivalent:
|X| 2
x,y∈X k 1) F is a uniform tight frame;
Proof: We show that 1) implies 2). Fix a vector x ∈ Cd . 2) F is a WBE-sequence set;
Note that p(y) = |hx|yi|2k = hy|xik hx|yik is a polynomial 3) F is a 1-design in CS d−1 .
function in Hom(k, k)◦ . Since X is a t-design, the exact Proof: The frame constants of a uniform tight frame F
cubature formula in Cd are given by A = B = |F |/d, see for example Property
Z
1 X 2k 2.3 in [8]. Therefore, F satisfies equation (2) of Theorem 1
|hx|yi| = |hx|yi|2k dµ(y)
|X| CS d−1 for k = 1. The equivalence of the three statements follow now
y∈X
from Theorem 1.
holds for all k in the range 0 ≤ k ≤ t. By Lemma 1, the latter Corollary 1: Any 1-design in CP d−1 is obtained by pro-
−1
integral evaluates to d+k−1k , which proves that equation jecting an orthogonal basis from a higher-dimensional space
(2) holds for all k ≤ t. (where all basis vectors have the same norm).
VI. E QUIVALENCE OF MUB S AND 2-D ESIGNS We can now evaluate the intersection polynomials gij (0) using
We need a few more notations before we state our main [11, Theorem 5.3] and obtain that |I(x, y)| = d − 2.
results. If B is a subset of CS d−1 , then the set A = Hence we can conclude that each set Bx forms an ortho-
{|hx|yi|2 : x, y ∈ B, x 6= y} is called the “angle” set of B. normal basis of Cd . The sets Bx partition B. If Bx 6= Bz ,
For an element x in B and an “angle” α ∈ A, we define the then the bases are by construction mutually unbiased.
subdegree dα (x) as dα (x) = |{y ∈ B : |hx|yi|2 = α}|. If the Zauner conjectures that if the dimension d is not a prime
subdegree dα of an α ∈ A is independent of x, then B is called power, then a 2-design with angle set {0, 1/d} cannot have
a regular scheme. Note that the union of mutually orthogonal d(d+1) elements [25]. His conjecture can now be reformulated
bases of Cd is a regular scheme with angle set {0, 1/d}. in terms of mutually unbiased bases, which then states that
N (d) < d + 1 for non-prime power d. If Zauner’s conjecture
Theorem 3: The union X of d + 1 mutually unbiased bases
is true, then this would explain the particular role of the finite
in Cd forms a 2-design in CS d−1 with angle set {0, 1/d} and
field construction by Wootters and Fields [24].
d(d + 1) elements.
Remark 2: Theorem 3 was obtained earlier by Zauner as
Proof: We verify that X attains the Welch bound in part of a more general result on combinatorial quantum designs
equation (1) with equality for 0 ≤ k ≤ 2. The statement then using a different terminology, see [25, Theorem 2.19]. The
follows from Theorem 1. Indeed, this is obvious for k = 0. converse direction, our Theorem 4, appears to be new.
We note that |X| = d(d + 1).
If we evaluate the left hand side of the Welch bound for X, VII. SIC-POVM S AND 2-D ESIGNS
then we obtain Finally, to demonstrate the versatility of Theorem 1 we also
1 d(d+1)
  show that another system of vectors used in quantum informa-
21
X
2
|hx|yi| = 1+(d−1)0+d tion theory corresponds to complex projective 2-designs. So-
d2 (d+1)2 d2 (d+1)2 d
x,y∈X called symmetric informationally complete positive operator-
1 valued measures (SIC-POVMs) are systems of d2 vectors
=
d in Cd which have constant inner product, i. e., |hv, wi|2 =
−1 1/(d + 1) for all v, w in the set. Like in case of MUBs it
and this coincides with d+1−1
1 = 1/d; so, X is a 1-design.
Similarly, for k = 2, is a challenging task to construct SIC-POVMs—indeed here
  solutions are known only for a finite number of dimensions
1 X
4 d(d+1) 2 1 [9], [16]. In [16] it was shown that SIC-POVMs actually form
|hx|yi| = 1+(d−1)0+d
d2 (d+1)2 d2 (d+1)2 d2 complex projective 2-designs. The following theorem gives a
x,y∈X
2 new and simple proof of this result.
= , Theorem 5 (SIC-POVMs are 2-designs [16]): Let X be a
d(d + 1)
−1 SIC-POVM X in dimension d. Then X forms a 2-design in
and this coincides with d+2−12 = 2/(d(d + 1)). CS d−1 with angle set {1/(d + 1)} and d2 elements.
Theorem 4: A 2-design B in complex projective space Proof: Again, we only have to verify that the set X of
CS d−1 with angle set {0, 1/d} and |B| = d(d + 1) elements vectors attains the Welch bound with equality for 0 ≤ k ≤ 2.
is the union of d + 1 mutually unbiased bases. The statement then follows from Theorem 1. Indeed, this is
Proof: A complex projective 2-design with s = obvious for k = 0. We note that here |X| = d2 . Evaluating
|{0, 1/d}| = 2 satisfies 2 ≥ s − 1, hence is a regular scheme the left hand side of the Welch bound for X, then we obtain
[10]. For α = 1/d, any x ∈ B has subdegree dα (x) = d2 by 1 X 1

1

2 2 4 2
Theorem 2.5 of [10]. It follows that x is orthogonal to d − 1 |hx|yi| = d · 1+(d − d )
d4 d4 d+1
elements. x,y∈X

Let Bx = {x} ∪ {z ∈ B : hx|zi = 0}. We claim that Bx is 1 1


= (1 + (d − 1)) =
an orthonormal basis of Cd . We may assume that the vectors d2 d
in B are normalized to unit norm. Thus, it suffices to show that −1
and this coincides with d+1−1
1 = 1/d; so, X is a 1-design.
Bx = By for each y ∈ Bx . For x = y this is trivial. We know Similarly, for k = 2,
that x and y are contained in both Bx and By . Therefore, it  
suffices to show that the intersection set 1 X 4 1 2 4 2 1
|hx|yi| = 4 d · 1 + (d − d )
d4 d (d + 1)2
x,y∈X
I(x, y) = {z ∈ B : hx|zi = 0, hy|zi = 0} = Bx ∩ By − {x, y}  
1 d−1 2
contains d − 2 elements. = 2
1+ =
d d+1 d(d + 1)
The number of elements in I(x, y) does not depend on
d+2−1 −1

x, y for a t-design with t ≥ 2s − 2, see [11]. Specializing and this coincides with 2 = 2/(d(d + 1)).
Theorem 5.2 in [11] to the case at hand shows that Remark 3: Zauner pointed out to us that the previous the-
1
X orem can also be obtained in the language of combinatorial
|I(x, y)| = d2 0 0
d(d + 1)gij (0) − 0i − 0j . quantum designs by combining Theorems 2.29 and 2.30 in his

σ1−i σ1−j
i,j=0 dissertation [25].
VIII. C ONCLUSION [4] Th. Beth, D. Jungnickel, and H. Lenz. Design Theory, volume I.
Cambridge University Press, 2. edition, 1999.
We have shown that the seemingly unrelated concepts of [5] H.F. Blichfeld. The minimum value of quadratic forms, and the closest
MUBs on the one hand and complex projective 2-deigns on packing of spheres. Math. Ann., 101:605–608, 1929.
[6] N. Bohr. Das Quantenpostulat und die neueren Entwicklungen der
the other are actually the same objects. This was anticipated Atomistik. Naturwissenschaften, 16:245–257, 1928.
in a paper by Barnum [2] in which it was shown that the [7] J.M. Goethals and J.J. Seidel. Cubature formulae, polytopes, and
union of the (d+1) bases of a particular system of MUBs spherical designs. In C. Davis, B. Grünbaum, and F.A. Sherk, editors,
The Geometric Vein – The Coxeter Festschrift, pages 203–218. Springer-
forms a complex projective 2-design. In the present paper we Verlag, 1981.
have generalized this to arbitrary MUBs and have also shown [8] V.K. Goyal, J. Kovačevič, and J.A. Kelner. Quantized frame expansions
the other direction, i. e., any 2-design in dimension d which with erasures. Appl. Comp. Harm. Analysis, 10:203–233, 2001.
[9] M. Grassl. On SIC-POVMs and MUBs in Dimension 6. pages 60–
consists of d2 + d elements and has angle set {0, 1/d} can be 61. Proceedings ERATO Conference on Quantum Information Science
partitioned into d + 1 sets of MUBs. We have also shown that (EQIS 2004), Tokyo, 2004. See also ArXiv preprint quant-ph/0406175.
these sets meet the Welsh bounds for k = 0, 1, 2 with equality. [10] S.G. Hoggar. Parameters of t-designs in FP n−1 . Europ. J. Combin.,
5:29–36, 1984.
Hence, the present paper can also be seen as a generalization [11] S.G. Hoggar. t-designs with general angle set. Europ. J. Combin.,
of the results of [21] in which the corresponding statement 13:257–271, 1992.
over the real numbers was shown. Finally, we would like to [12] I. D. Ivanovic. Geometrical description of quantal state determination.
Journal of Physics A, 14(12):3241–3245, 1981.
mention that Zauner [25] conjectures that affine 2-designs do [13] A. Klappenecker and M. Rötteler. Constructions of mutually unbiased
not exist in dimensions d having two distinct prime factors. bases. volume 2948 of LNCS, pages 137–144. Springer, 2004.
[14] H. König. Isometric embeddings of Euclidean spaces into finite dimen-
ACKNOWLEDGMENT sional `p -spaces. In B. Jakubczyk, S. Janeczko, and B. Ziemian, editors,
Panoramas of Mathematics, volume 34 of Banach Center Publications,
We are grateful to Emina Soljanin for bringing tight frames pages 79–87, Warsaw, 1995.
[15] J.L. Massey and T. Mittelholzer. Welch’s bound and sequences sets
to our attention, and to Gerhard Zauner for assisting with for code-division multiple-access systems. In Sequences II: Methods
the translation of his results. We also thank Chris Godsil, in Communications, Security and Computer Sciences, pages 63–78,
Markus Grassl, Joe Renes, Aidan Roy for several interesting Heidelberg, 1993. Springer-Verlag.
[16] J. M. Renes, R. Blume-Kohout, A. J. Scott, and C. M. Caves. Sym-
discussions, and Richard Cleve for pointing out a small error metric informationally complete quantum measurements. J. Math. Phys,
in an example of a previous version. 45(6):2171–2180, 2004.
The research of A.K. was supported in part by NSF grant [17] W. Rudin. Function Theory in the Unit Ball of Cn . Springer-Verlag,
New York, 1980.
CCR-0218582, NSF CAREER award CCF-0347310, a TEES [18] J. Schwinger. Unitary operator bases. Proc. Nat. Acad. Sci. U.S.A.,
Select Young Faculty award, and a Texas A&M TITF grant. 46:570–579, 1960.
This work was carried out while M.R. was at the Institute for [19] M.O. Scully and M.S. Zubairy. Quantum Optics. Cambridge University
Press, Cambridge, 1997.
Quantum Computing, University of Waterloo, Canada. [20] V.M. Sidelnikov. New bounds for the density of sphere packings in an
n-dimensional Euclidean space. Mat. Sbornik 95 (1974) = Math. USSR
R EFERENCES Sbornik, 24:147–157, 1974.
[21] S. Waldron. Generalized Welch bound equality sequences are tight
[1] S. Bandyopadhyay, P. O. Boykin, V. Roychowdhury, and F. Vatan. A frames. IEEE Transactions on Information Theory, 49(9):2307–2309,
new proof for the existence of mutually unbiased bases. Algorithmica, 2003.
34:512–528, 2001. [22] L.R. Welch. Lower bounds on the maximum cross correlation of signals.
[2] H. Barnum. Information-disturbance tradeoff in quantum measurement IEEE Transactions on Information Theory, 20(3):397–399, 1974.
on the uniform ensemble and on the mutually unbiased bases. ArXiv [23] P. Wocjan and Th. Beth. New construction of mutually unbiased bases
preprint quant–ph/0205155, 2002. in square dimensions. ArXiv preprint quant–ph/0407081, 2004.
[3] C.H. Bennett and G. Brassard. Quantum cryptography: Public key [24] W. Wootters and B. Fields. Optimal state-determination by mutually
distribution and coin tossing. In Proc. of the IEEE Intl. Conf. Computers, unbiased measurements. Ann. Physics, 191(2):363–381, 1989.
Systems, and Signal Processing, pages 175–179. IEEE, 1984. [25] G. Zauner. Quantendesigns – Grundzüge einer nichtkommutativen
Designtheorie (in German). PhD thesis, Universität Wien, 1999.