Abstract
We develop priors for Bayes estimation of quantum states that provide minimax state estimation. The relative entropy from the true density operator to a predictive density operator is adopted as a loss function. The proposed prior maximizes the conditional Holevo mutual information, and it is a quantum version of the latent information prior in classical statistics. For one qubit system, we provide a class of measurements that is optimal from the viewpoint of minimax state estimation.
  1. Introduction
In quantum mechanics, the outcome of a measurement is subject to a probability distribution determined from the quantum state of the measured system and the measurement performed. The task of estimating the quantum state from the outcome of measurement is called the quantum estimation and it is a fundamental problem in quantum statistics [,,]. Tanaka and Komaki [] and Tanaka [] discussed quantum estimation using the framework of statistical decision theory and showed that Bayesian methods provide better estimation than the maximum likelihood method. In Bayesian methods, we need to specify a prior distribution on the unknown parameters of the quantum states. However, the problem of prior selection has not been fully discussed for quantum estimation [].
The quantum state estimation problem is related to the predictive density estimation problem in classical statistics []. This is a problem of predicting the distribution of an unobserved variable y based on an observed variable x. Suppose  where  denotes an unknown parameter. Based on the observed x, we predict the distribution  of y using a predictive density . The plug-in predictive density is defined as , where  is some estimate of  from x. The Bayesian predictive density with respect to a prior distribution  is defined as
      
      
        
      
      
      
      
    
      where  is the posterior distribution. We compare predictive densities using the framework of statistical decision theory. Specifically, a loss function  is introduced that evaluates the difference between the true density q and the predictive density p. Then, the risk function  is defined as the average loss when the true value of the parameter is :    
      
        
      
      
      
      
    
A predictive density  is called minimax if it minimizes the maximum risk among all predictive densities:
      
        
      
      
      
      
    
We adopt the Kullback–Leibler divergence
      
      
        
      
      
      
      
    
      as a loss function, since it satisfies many desirable properties compared to other loss functions such as the Hellinger distance and the total variation distance []. Under this setting, Aitchison [] proved
      
      
        
      
      
      
      
    
      where
      
      
        
      
      
      
      
    
      is called the Bayes risk. Namely, the Bayesian predictive density  minimizes the Bayes risk. We provide the proof of Equation (4) in the Appendix A. Therefore, it is sufficient to consider only Bayesian predictive densities from the viewpoint of Kullback–Leibler risk, and the selection of the prior  becomes important.
For the predictive density estimation problem above, Komaki [] developed a class of priors called the latent information priors. The latent information prior  is defined as a prior that maximizes the conditional mutual information  between the parameter  and the unobserved variable y given the observed variable x. Namely,
      
      
        
      
      
      
      
    
      where
      
      
        
      
      
      
      
    
      is the conditional mutual information between y and  given x. Here,
      
      
        
      
      
      
      
    
      are marginal densities. The Bayesian predictive densities based on the latent information priors are minimax under the Kullback–Leibler risk:
      
        
      
      
      
      
    
The latent information prior is a generalization of the reference prior [] that is a prior maximizing the unconditional mutual information  between  and y.
Now, we consider the problem of estimating the quantum state of a system  based on the outcome of a measurement on a system . Suppose the quantum state of the composed system  be  where  denotes an unknown parameter. We perform a measurement on the system  and obtain the outcome x. Based on the measurement outcome x, we estimate the state of the system  by a predictive density operator . Similarly to the Bayesian predictive density (1), the Bayesian predictive density operator with respect to the prior  is defined as
      
      
        
      
      
      
      
    
      where  is the posterior distribution. Like the predictive density estimation problem discussed above, we compare predictive density operators using the framework of statistical decision theory. There are several possibilities for the loss function  in quantum estimation such as the fidelity and the trace norm []. In this paper, we adopt the quantum relative entropy
      
      
        
      
      
      
      
    
      as a loss function, since it is a quantum analogue of the Kullback–Leibler divergence (3). Note that the fidelity and the trace norm correspond to the Hellinger distance and the total variation distance in the classical statistics, respectively. Under this setting, Tanaka and Komaki [] proved that the Bayesian predictive density operators minimize the Bayes risk:
      
        
      
      
      
      
    
This is a quantum version of Equation (4).
From Tanaka and Komaki [], the selection of the prior becomes important also in quantum estimation. However, this problem has not been fully discussed []. In this paper, we provide a quantum version of the latent information priors and prove that they provide minimax predictive density operators. Whereas the latent information prior in the classical case maximizes the conditional Shannon mutual information, the proposed prior maximizes the conditional Holevo mutual information. The Holevo mutual information, which is a quantum version of the Shannon mutual information, is a fundamental quantity in the classical-quantum communication []. Our result shows that the conditional Holevo mutual information also has a natural meaning in terms of quantum estimation.
Unlike the classical statistics, the measurement is not unique in quantum statistics. Therefore, selection of the measurement also becomes important. From the viewpoint of minimax state estimation, measurements that minimize the minimax risk are considered to be optimal. We provide a class of optimal measurements for one qubit system. This class includes the symmetric informationally complete measurement [,]. These measurements and latent information priors provide robust quantum estimation.
2. Preliminaries
2.1. Quantum States and Measurements
We briefly summarize several notations of quantum states and measurements. Let  be a separable Hilbert space of a quantum system. A Hermitian operator  on  is called a density operator if it satisfies
        
      
        
      
      
      
      
    
The state of a quantum system is described by a density operator. We denote the set of all density operators on  as .
Denote the set of all linear operators on Hilbert space  by  and the set of all positive linear operators by . Let  be a measurable space of all possible outcomes of a measurement and  be a -algebra of . A map  is called a positive operator-valued measure (POVM) if it satisfies , and  where . Any quantum measurement is represented by a POVM on . In this paper, we mainly assume  is finite. In such case, we denote  and any POVM is represented by a set of positive Hermitian operators  such that .
The outcome of a measurement E on a quantum system with the state  is distributed with a probability measure
        
      
        
      
      
      
      
    
Let ,  be quantum systems with Hilbert spaces  and . The Hilbert space of the composed system  is given by the tensor product . Suppose the state of this composed system is . Then, the states of two subsystems can be yielded by the partial trace:
      
        
      
      
      
      
    
If a measurement  is performed on the system  and the measurement outcome is x, then the state of the system  becomes
        
      
        
      
      
      
      
    
        where the normalization constant
        
      
        
      
      
      
      
    
        is the probability of the outcome x. Here,  is the identity operator on the space . We call the operator  the conditional density operator.
2.2. Quantum State Estimation
We formulate the quantum state estimation problem using the framework of statistical decision theory. Let  and  be quantum systems with finite-dimensional Hilbert spaces  and , where  and .
Suppose the state of the composed system  be , where  denotes unknown parameters. We perform a measurement  on , observe the outcome , and estimate the conditional density operator  of  by a predictive density operator . As discussed in the introduction (1) and (7), the Bayesian predictive density operator based on a prior  is defined by
        
      
        
      
      
      
      
    
        where  is the posterior distribution.
To evaluate predictive density operators, we introduce a loss function  that evaluates the difference between the true conditional density operator  and the predictive density operator . In this paper, we adopt the quantum relative entropy (8) since it is a quantum analogue of the Kullback–Leibler divergence (3). Then, the risk function  of a predictive density operator  is defined by
        
      
        
      
      
      
      
    
        where
        
      
        
      
      
      
      
    
        is the probability of the outcome x. Similarly to the classical case (2), a predictive density operator  is called minimax if it minimizes the maximum risk among all predictive density operators [,]:
      
        
      
      
      
      
    
Tanaka and Komaki [] showed
        
      
        
      
      
      
      
    
        where
        
      
        
      
      
      
      
    
        is called the Bayes risk. Namely, the Bayesian predictive density operator minimizes the Bayes risk. This result is a quantum version of Equation (4). Although Tanaka and Komaki [] considered separable models (), the relation (9) holds also for non-separable models as shown in the Appendix A. Therefore, it is sufficient to consider only Bayesian predictive density operators and the problem of prior selection becomes crucial.
2.3. Notations
For a quantum state family , we define another quantum state family
        
      
        
      
      
      
      
    
        where
        
      
        
      
      
      
      
    
        is a density operator in . Since , the state family  can be regarded as a subset of the Euclidean space . By identifying  with , the parameter space  is endowed with the induced topology as a subset of .
Any measurement on the system  is represented by a projective measurement , where  is an orthonormal basis of . For every , we define  as
        
      
        
      
      
      
      
    
        which is the unnormalized state of  conditional on the measurement outcome x. We also define
        
      
        
      
      
      
      
    
3. Minimax Estimation of Quantum States
In this section, we develop the latent information prior for quantum state estimation and show that this prior provides a minimax predictive density operator.
In the following, we assume the following conditions:
- is compact.
- For every , .
- For every , there exists such that .
The third assumption is achieved by adopting sufficiently small Hilbert space. Namely, if there exists  such that  for every , then we redefine the state space  as the orthogonal complement of .
Let  be the set of all probability measures on  endowed with the weak convergence topology and the corresponding Borel algebra. By the Prohorov theorem [] and the first assumption,  is compact.
When x is fixed, the function  is bounded and continuous. Thus, for every fixed , the function
      
      
        
      
      
      
      
    
      is continuous because  is endowed with the weak convergence topology and . Let  be the eigenvalues and the normalized eigenvectors of the predictive density operator . For every predictive density operator , consider the function from  to  defined by
      
      
        
      
      
      
      
    
The last term in (10) is lower semicontinuous under the definition  [], since each summand takes either zero or infinity and so the set of  such that this term takes zero is closed. In addition, the other terms in (10) are continuous since the von Neumann entropy is continuous []. Therefore, the function  in (10) is lower-semicontinuous.
Now, we prove that the class of predictive density operators that are limits of Bayesian predictive density operators is an essentially complete class. We prepare three lemmas. Lemma 1 is useful for differentiation of quantum relative entropy (see Hiai and Petz []). Lemmas 2 and 3 are from Komaki [].
Lemma 1. 
Let  be n-dimensional self-adjoint matrices and t be a real number. Assume that  is a continuously differentiable function defined on an interval and assume that the eigenvalues of  are in  if t is sufficiently close to . Then,
      
        
      
      
      
      
    
Lemma 2 
([]). Let μ be a probability measure on Θ. Then,
      
        
      
      
      
      
    is a closed subset of  for .
Lemma 3 
([]). Let  be continuous, and let μ be a probability measure on Θ such that  for every . Then, there is a probability measure  in
      
        
      
      
      
      
    for every n, such that . Furthermore, there exists a convergent subsequence  of  and the equality  holds, where .
By using these results, we obtain the following theorem, which is a quantum version of Theorem 1 of Komaki [].
Theorem 1. 
- (1)
- Let be a predictive density operator. If there exists a prior such that and for every , then for every .
- (2)
- For every predictive density operator ρ, there exists a convergent prior sequence such that , exists, and for every .
Next, we develop priors that provide minimax predictive density operators. Let x be a random variable, which represents the outcome of the measurement, i.e., . Then, as a quantum analogue of the conditional mutual information (5), we define the conditional Holevo mutual information [] between the quantum state  of Y and the parameter  given the measurement outcome x as
      
      
        
      
      
      
      
    
      which is a function of . Here, we used
      
      
        
      
      
      
      
    
      and
      
      
        
      
      
      
      
    
The conditional Holevo mutual information provides an upper bound on the conditional mutual information as follows.
Proposition 1. 
Let  be the state of the composed system . Suppose that a measurement is performed on X with the measurement outcome x and then another measurement is performed on Y with the measurement outcome y. Then,
      
        
      
      
      
      
    
Proof.  
Since any measurement is a trace-preserving completely positive map, inequality (12) follows from the monotonicity of the quantum relative entropy []. ☐
Analogous with the latent information priors [] in classical statistics, we define latent information priors as priors that maximize the conditional Holevo mutual information. It is expected that the Bayesian predictive density operator  based on a latent information prior is a minimax predictive density operator. This is true from the following theorem, which is a quantum version of Theorem 2 of Komaki [].
Theorem 2. 
- (1)
- Let be a prior maximizing . If for all ; then, is a minimax predictive density operator.
- (2)
- There exists a convergent prior sequence such that is a minimax predictive density operator and the equality holds.
The proof of Theorems 1 and 2 are deferred to the Appendix A.
We note that the minimax risk  depends on the measurement E on . Therefore, the measurement E with minimum minimax risk is desirable from the viewpoint of minimaxity. We define a POVM  to be a minimax POVM if it satisfies
      
      
        
      
      
      
      
    
In the next section, we provide a class of minimax POVMs for one qubit system.
4. One Qubit System
In this section, we consider one qubit system and derive a class of minimax POVMs satisfying (13).
Qubit is a quantum system with a two-dimensional Hilbert space. It is the fundamental system in the quantum information theory. A general state of one qubit system is described by a density matrix
      
      
        
      
      
      
      
    
      where . The parameter space  for pure states is called the Bloch sphere.
Let  be a separable state. We consider the estimation of  from the outcome of a measurement on . Here, we assume that the state  is separable, since the state of Y changes according to the outcome of the measurement on X and so the estimation problem is not well-defined if the state  is not separable.
Let  and  be Borel sets. From Haapasalo et al. [], it is sufficient to consider POVMs on . For every probability measure  on  that satisfies
      
      
        
      
      
      
      
    
      we define a POVM  by
      
      
        
      
      
      
      
    
In the following, we identify E with .
Let  be a class of POVMs on  represented by measures that satisfy the conditions
      
      
        
      
      
      
      
    
      where  is the expectation with respect to a measure . We provide two examples of POVMs in .
Proposition 2. 
The POVM corresponding to
      
        
      
      
      
      
    where  is surface element, is in .
Proof.  
From the symmetry of , . Moreover, from  and the symmetry of , . ☐
Proposition 3. 
Suppose that  satisfies . Let μ be a four point discrete measure on Ω defined by
      
        
      
      
      
      
    
Then, the POVM corresponding to μ belongs to .
Proof.  
Let  and . From the assumption on ,
        
      
        
      
      
      
      
    
        where  is the identity matrix and  is a matrix whose elements are all one. From (16), we have . Therefore,  and it implies .
In addition, from (16),
        
      
        
      
      
      
      
    
Therefore, . Since , it implies . Then,  and . ☐
We note that the POVM (15) is a special case of the SIC-POVM (symmetric, informationally complete, positive operator valued measure) [,].
Let  be a class of priors on  that satisfies the conditions
      
      
        
      
      
      
      
    
      where  is the expectation with respect to a prior .
Proposition 4. 
The uniform prior
      
        
      
      
      
      
    where  is the surface element on the Bloch sphere, belongs to .
Proof.  
Same as Proposition 2. ☐
Proposition 5. 
Suppose that  satisfies . Then, the four point discrete prior
      
        
      
      
      
      
    belongs to .
Proof.  
Same as Proposition 3. ☐
We obtain the following result.
Lemma 4. 
Suppose . Then, for general measurement E, the risk function of the Bayesian predictive density operator  is
      
        
      
      
      
      
    
Proof.  
The distribution of the measurement outcome  is
        
      
        
      
      
      
      
    
Then, since , the marginal distribution of the measurement outcome is
        
      
        
      
      
      
      
    
Therefore, the posterior distribution of  is
        
      
        
      
      
      
      
    
The posterior mean of  and  are  and , respectively.
Thus, the Bayesian predictive density operator based on prior  is
        
      
        
      
      
      
      
    
        and we have
        
      
        
      
      
      
      
    
Therefore, the quantum relative entropy loss is
        
      
        
      
      
      
      
    
Hence, the risk function is
        
      
        
      
      
      
      
     ☐
Theorem 3. 
For a measurement , every  is a latent information prior:
      
        
      
      
      
      
    
In addition, the risk of the Bayesian predictive density operator based on  is
      
        
      
      
      
      
    where h is the binary entropy function .
Proof.  
From Lemma 4 and ,
        
      
        
      
      
      
      
    
Therefore, the risk depends only on  and we have
        
      
        
      
      
      
      
    
Since
        
      
        
      
      
      
      
    
      
        
      
      
      
      
    
        the function  is convex. In addition, we have . Therefore,  takes the maximum at .
In other words,  takes maximum on the Bloch sphere. In addition, since , the support of  is included in the Bloch sphere . Therefore,  and it implies that  is a latent information prior. ☐
We note that the Bayesian predictive density operator is identical for every . In fact, every  also provides the minimax estimation of density operator  when there is no observation system X. Figure 1 shows the risk function  in (17) and also the minimax risk function  when there is no observation:    
      
        
      
      
      
      
    
 
      
    
    Figure 1.
      Risk functions of predictive density operators. solid line: , dashed line: .
  
Whereas  around , we can see that  around . Both risk functions take the maximum at  and
      
      
        
      
      
      
      
    
The decrease  in the maximum risk corresponds to the gain from the observation X.
Now, we consider the selection of the measurement E. As we discussed in the previous section, we define a POVM  to be a minimax POVM if it satisfies (13). We provide a sufficient condition on a POVM to be minimax. Let  be a minimax predictive density operator for the measurement E.
Lemma 5. 
Suppose  is a latent information prior for the measurement . If
      
        
      
      
      
      
    then  is a minimax POVM.
Proof.  
For every , we have
        
      
        
      
      
      
      
    
The last equality is from the minimaxity of . Therefore,  is a minimax POVM. ☐
Theorem 4. 
Every  is a minimax POVM.
Proof.  
Let . From Theorem 6,  is a latent information prior for .
For general measurement E, from Lemma 4, the risk function of the Bayesian predictive density operator  is
        
      
        
      
      
      
      
    
Hence, the Bayes risk of  with respect to  is
        
      
        
      
      
      
      
    
Now, since the Bayesian predictive density operator  minimizes the Bayes risk with respect to  among all predictive density operators [],
        
      
        
      
      
      
      
    
        for every E. Therefore,
        
      
        
      
      
      
      
    
On the other hand,
        
      
        
      
      
      
      
    
        is obvious.
Hence,
        
      
        
      
      
      
      
    
From Lemma 5,  is minimax. ☐
Whereas Theorems 1 and 2 are valid even when  is not separable, Theorems 3 and 4 assume the separability .
From Theorem 4, the POVM (15) is a minimax POVM. Since this POVM is identical to the SIC-POVM [,], it is an interesting problem whether the SIC-POVM is a minimax POVM also in higher dimensions. This is a future work.  
Acknowledgments
We thank the referees for many helpful comments. This work was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Numbers 26280005 and 14J09148.
Author Contributions
All authors contributed significantly to the study and approved the final version of the manuscript.
Conflicts of Interest
The authors declare no conflict of interest.
Appendix A. Proofs
Proof of (4). 
Therefore, for arbitrary ,
          
      
        
      
      
      
      
    
          which is nonnegative since the Kullback–Leibler divergence  in (3) is always nonnegative. ☐
Proof of (9). 
Therefore, for arbitrary ,
          
      
        
      
      
      
      
    
          which is nonnegative since the quantum relative entropy  in (8) is always nonnegative. ☐
Proof of Theorem 1. 
(1) Let  be the orthogonal projection matrix onto the eigenspace of  corresponding to eigenvalue 0,  and  be the set of all probability measures on .
If , the assertion is obvious because  for . Therefore, we assume  in the following. In this case, . Since  if and only if , we have .
Define
          
      
        
      
      
      
      
    
          for  and , where  is the probability measure satisfying . Then, , and we have
          
      
        
      
      
      
      
    
Thus, if ,
          
      
        
      
      
      
      
    
          If , . Therefore, for every , the inequality  holds.
(2) We note that  and  are compact subsets of  and , respectively.
If , the assertion is obvious, because  for every . Therefore, we assume  in the following. Let  and  be a probability measure on  such that  for every .
Because  is continuous as a function of , there exists  such that . From Lemma 3, there exists a convergent subsequence  of  such that  where .
Let  be the integer satisfying . We can make the subsequence  satisfy  for some positive constant c.
Since
          
      
        
      
      
      
      
    
          for every , we have
          
      
        
      
      
      
      
    
          for every  and . Thus,
          
      
        
      
      
      
      
    
Hence,
          
      
        
      
      
      
      
    
          where  is the orthogonal projection matrix onto the eigenspace of  corresponding to the eigenvalue 0. Here, we have
          
      
        
      
      
      
      
    
          and
          
      
        
      
      
      
      
    
By taking an appropriate subsequence  of , we can make the subsequence of density operators  converge for all  because   and .
Then, from (A4), if ,
          
      
        
      
      
      
      
    
If ,  because .
Hence, the risk of the predictive density operator defined by
          
      
        
      
      
      
      
    
          where  is an arbitrary predictive density, is not greater than that of  for every .
Therefore, by taking a sequence  that converges rapidly enough to 0, we can construct a predictive density operator
          
      
        
      
      
      
      
    
          as a limit of Bayesian predictive density operators based on priors , where  is a measure on  such that  for every .
Hence, the risk of the predictive density operator (A5) is not greater than that of  for every . ☐
Proof of Theorem 2. 
(1) Define  for all  and . Then,
          
      
        
      
      
      
      
    
Since  for every  and  if , we have
          
      
        
      
      
      
      
    
          for every .
On the other hand, we have
          
      
        
      
      
      
      
    
Here, the first equality is from the fact [] that the Bayes risk with respect to 
      
        
      
      
      
      
    
          is minimized when
          
      
        
      
      
      
      
    
Therefore, the predictive density operator  is minimax.
(2) Let  be a probability measure on  such that  for every , and let  be a prior satisfying . From Lemma 3, there exists a convergent subsequence  of  and  where . Let  be the integer satisfying . As in the proof of Theorem 1, we can make the subsequence  satisfy  for some positive constant c.
Then, for every ,
          
      
        
      
      
      
      
    
          belongs to  for  because  and .
Thus,
          
      
        
      
      
      
      
    
Since  for every m and  if , we have
          
      
        
      
      
      
      
    
Hence,
          
      
        
      
      
      
      
    
          where  is the orthogonal projection matrix onto the eigenspace of  corresponding to the eigenvalue 0. Here, we used two equalities
          
      
        
      
      
      
      
    
          and
          
      
        
      
      
      
      
    
          since  is a bounded continuous function of .
By taking an appropriate subsequence  of , we can make  converge for every x. Then, for every ,
          
      
        
      
      
      
      
    
      
        
      
      
      
      
    
          since  for x with .
On the other hand, we have
          
      
        
      
      
      
      
    
Here, the first equality is from the fact [] that the Bayes risk
          
      
        
      
      
      
      
    
          is minimized when . Although  is not uniquely determined for x with , the Bayes risk does not depend on the choice of  for such x.
Therefore, the predictive density operator  is minimax. ☐
References
- Barndorff-Nielsen, O.E.; Gill, R.D.; Jupp, P.E. On quantum statistical inference. J. R. Stat. Soc. B 2003, 65, 775–804. [Google Scholar] [CrossRef]
- Holevo, A.S. Probabilistic and Statistical Aspects of Quantum Theory; Elsevier: Amsterdam, The Netherlands, 1982. [Google Scholar]
- Paris, M.; Rehacek, J. Quantum State Estimation; Springer: Berlin, Germany, 2004. [Google Scholar]
- Tanaka, F.; Komaki, F. Bayesian predictive density operators for exchangeable quantum-statistical models. Phys. Rev. A 2005, 71, 052323. [Google Scholar] [CrossRef]
- Tanaka, F. Bayesian estimation of the wave function. Phys. Lett. A 2012, 376, 2471–2476. [Google Scholar] [CrossRef]
- Tanaka, F. Noninformative prior in the quantum statistical model of pure states. Phys. Rev. A 2012, 85, 062305. [Google Scholar] [CrossRef]
- Geisser, S. Predictive Inference: An Introduction; Chapman & Hall: London, UK, 1993. [Google Scholar]
- Csiszar, I. Axiomatic characterizations of information measures. Entropy 2008, 10, 261–273. [Google Scholar] [CrossRef]
- Aitchison, J. Goodness of prediction fit. Biometrika 1975, 62, 547–554. [Google Scholar] [CrossRef]
- Komaki, F. Bayesian predictive densities based on latent information priors. J. Stat. Plan. Inference 2011, 141, 3705–3715. [Google Scholar] [CrossRef]
- Bernardo, J.M. Reference posterior distributions for Bayesian inference. J. R. Stat. Soc. B 1979, 41, 113–147. [Google Scholar]
- Petz, D. Quantum Information and Quantum Statistics; Springer: New York, NY, USA, 2008. [Google Scholar]
- Holevo, A.S. Quantum Systems, Channels, Information: A Mathematical Introduction; Walter de Gruyter: Berlin, Germany, 2013. [Google Scholar]
- Appleby, D.M. SIC-POVMs and the extended Clifford group. J. Math. Phys. 2004, 46, 547–554. [Google Scholar]
- Renes, J.M.; Blume-Kohout, R.; Scott, A.J.; Caves, C.M. Symmetric informationally complete quantum measurements. J. Math. Phys. 2004, 45, 2171–2180. [Google Scholar] [CrossRef]
- Ferrie, C.; Blume-Kohout, R. Minimax quantum tomography: Estimators and relative entropy bounds. Phys. Rev. Lett. 2016, 116, 090407. [Google Scholar] [CrossRef] [PubMed]
- Tanaka, F. Quantum minimax theorem. arXiv 2014, arXiv:1410:3639. [Google Scholar]
- Billingsley, P. Convergence of Probability Measures; Wiley: New York, NY, USA, 1999. [Google Scholar]
- Hiai, F.; Petz, D. Introduction to Matrix Analysis and Applications; Springer: New York, NY, USA, 2014. [Google Scholar]
- Haapasalo, E.; Heinosaari, T.; Pellonpää, J.P. Quantum measurements on finite dimensional systems: Relabeling and mixing. Quantum Inf. Process. 2012, 11, 1751–1763. [Google Scholar] [CrossRef]
© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
