Stateindependent uncertainty relations and entanglement detection in noisy systems
Abstract
Quantifying quantum mechanical uncertainty is vital for the increasing number of experiments that reach the uncertainty limited regime. We present a method for computing tight variance uncertainty relations, i.e., the optimal stateindependent lower bound for the sum of the variances for any set of two or more measurements. The bounds come with a guaranteed error estimate, so results of preassigned accuracy can be obtained straightforwardly. Our method also works for POVM measurements. Therefore, it can be used for detecting entanglement in noisy environments, even in cases where conventional spin squeezing criteria fail because of detector noise.
pacs:
03.65.Ta, 02.60.Pn, 03.67.MnIntroduction
Uncertainty relations quantitatively express a phenomenon which is ubiquitous in quantum mechanics: Given two observables and , it is usually impossible to prepare a state such that the respective outcome distributions of these observables are both sharp. Of course, for the best known example of this, the position and momentum observables, the relation is in every textbook. It was first established by Kennard [1], who turned Heisenberg’s heuristic ideas [2] into a quantitative statement. In particular, it was his idea to consider the variances
The aim of our paper is to provide an efficient method to obtain the best possible bounds for any given pair of measurements , . This is of direct use in the increasing number of experiments that reach the uncertaintylimited regime. A particular application is the certification of entanglement via steering inequalities [4]; [5]; [6]. In such applications, even if one does not necessarily need an optimal bound, it is crucial to have a correct one, i.e., a bound valid for all states. Any algorithm based on computing the uncertainties “for sufficiently many states” will fail to guarantee this correctness. In particular, in high dimensional Hilbert spaces, typical states will not have uncertainties near the boundary, so it is actually hard to explore the set of uncertainty pairs “from within”. Our method uses instead an “outer” approximation, which has the virtue that in every step it provides a correct bound. The bound is iteratively improved, converging to the optimal one. This feature sets our method apart from several recent works, in which ad hoc methods were used to provide uncertainty bounds.
The problem of getting optimal uncertainty bounds becomes more difficult as the dimension of the Hilbert space increases. Indeed, naively it would seem to be a search problem on the dimensional manifold of pure states, which in bad cases might scale exponentially with . However, we can do much better. We reformulate the problem as a geometric problem in three dimensions, namely of getting a sequence of outer polyhedral approximation of a certain convex set, see Fig. 1. Any such approximation gives a valid uncertainty bound. In the iteration step, i.e., for computing a tighter approximation, one has to compute the lowest eigenvalue of a certain hermitian combination of the operators and . Those eigenvalue problems now determine the scaling of our method as a function of dimension, which will be a low order polynomial in . Moreover, if additional information is available about and , for example, if they are both sparse in the same basis, eigenvalue computations can be speeded up considerably, and our method will speed up by the same factor.
Tight uncertainty bounds have only been obtained for a few specific pairs of observables. One example is angular momentum measurement, where bounds for two or three orthogonal spin components [4]; [7]; [8] are known. In those cases symmetry crucially helps to reduce the problem. Other examples are qubits [9], for which the low dimension allows an analytical solution.
There are also variants, in which the sharpness of a distribution is measured by other quantities than the usual variance [10]; [11]; [12]; [13], for instance entropies [14]; [15], or where more than two observables are considered simultaneously [16]; [17]. Quite different methods [18] are needed for optimal measurement uncertainty relations [10], or informationdisturbance bounds [10], so we will not consider these aspects here.
Methods
Linear state independent bounds
Since we are interested in stateindependent bounds [19] we have no use for the oftencited general relation by Robertson [20] (and its improvements [21]), which have a statedependent expression like , or similar, on the right hand side. Indeed, any relation of product form is useless for stateindependent relations in finite dimension: and have discrete eigenvalues, so the trivial is the best possible bound. We therefore consider bounds of the form
(1) 
Here, is the largest constant for which the above holds on any quantum state . Since our method handles arbitrary and we can also admit factors here, i.e., inequalities of the form . Each of these constrains the set of uncertainty pairs to a halfplane, and together they outline the uncertainty set (or, more precisely its “lower convex hull”, see Fig. 4 and [7]; [9]; [18]).
To see the connection to eigenvalue problems we write the optimal constant in (1) as
(2) 
Here we just wrote the variance as the minimal quadratic deviation, using that the minimum with respect to is attained at the expectation . On the other hand, if we fix and , the minimization with respect to is exactly the ground state problem for the operator in parentheses. This suggested our previous ansatz [7], which we call the seesaw algorithm: One alternatingly minimizes with respect to and . In many practical cases this converges quickly, and with the safeguard of trying out several initial values it seems fairly reliable. However, in general the method of Alternating Minimization may easily fail to find the global minimum, and there is no proof of convergence. Intermediate results of the seesaw algorithm give an upper bound on , but as an upper bound on a lower bound this is useless for applications. Moreover, there are indications that the seesaw algorithm actually may get trapped.
Geometry of outer approximations
In contrast, the method described in this paper is an outer method, in which all intermediate steps give valid lower and upper bounds on . Its geometric core is the joint numerical range
(3) 
where denotes the state space, i.e., the set of density operators. Notice first that this set contains all the information necessary to compute from (2). With the quadratic functional of we find
(4) 
Now the set is clearly convex and compact, because the state space has these properties, and they are preserved by the map taking to the tuple of expectations. The set is therefore completely described by the linear inequalities it satisfies. To get such inequalities, let be a real vector, and consider . Let denote the smallest eigenvalue of this operator. Then, for any state , and hence the corresponding tuple of expectations:
(5) 
Now let be any finite set of vectors, and consider the polytope of those points , which just satisfy the inequalities (5) with . Since these vectors satisfy fewer constraints than , we have , i.e., this is an outer approximation of . Denote by the set of extreme points of , which is also finite. Then
(6) 
Here we have used, firstly, that the minimum over a larger set is smaller, and, secondly, that the functional is concave, so that the minimum over a compact convex set is attained at an extreme point. Hence for every finite set of directions, we get a lower bound on , which is computed as a finite minimum over . On the other hand, for each we get a point , with equality in Eq. (5). Then
(7) 
So for every set , this procedure estimates the optimal constant up to a precision .
.1 Basic algorithm
The idea of the algorithm is now to let the set grow step by step, which shrinks , so increases and decreases (see Fig. 2 and Fig. 3). The algorithm terminates when is below the target accuracy.
Apart from the set it is useful to keep track of the polytope in the form of a list of vertices and edges. To arrive at the next approximation :

Determine a vertex at which becomes minimal, and set
(8) 
Solve the minimumeigenvalue problem for . This provides the bound for the new inequality (5), and an expectation tuple corresponding to the ground state.

Compute and update , if this is smaller than the current value.

Take the new inequality (5), and compute the intersections with all current edges of . This will give some new extreme points for , and corresponding edges.

Evaluate on the new extreme points in and update . Terminate if is as small as desired. Otherwise go to step 1.
All these steps except the choice in step 1 are dictated by the geometry of outer approximation. The rationale of the choice (8) (apart from its flavour of gradient search) is that, whenever possible, it will eliminate the vertex from , and thus strictly increase , unless there are other vertices with the same value of , which have first to be eliminated in a similar manner. A proof of this statement is provided in the appendix. As an application of our method, we derived the uncertainty relations for two nonorthogonal spin components, see the appendix.
.2 Generalization to POVMs
Our method can be applied with minimal modifications to generalized measurements, i.e. observables given by positive operator valued measures (POVMs). In general, a POVM measurement is described by its outcomes and corresponding effects [22]; [23], where the probability of obtaining the outcome is given by . The moments of an outcome distribution are then given by the expectations of the moment operators . The only difference from the “standard” projection valued case is that the identity no longer holds. But this is not required for our method.
We therefore only need to express variances as , and replace in (3) and the definition of : by , by , and analogously for .
Application to entanglement detection
In [4]; [5], it was shown that every stateindependent uncertainty relation like (4) yields a nonlinear entanglement witness, when applied to local measurements in a bipartition. Here the following scenario is considered: Two parties, Alice and Bob, can perform local measurements such as , on an unknown quantum state . Their goal is to decide if is entangled or not. For this, they measure the ’sum observables’ , given by
(9) 
In the POVM case this is generalized to measuring on Alice’s side, on Bob’s, and adding the outcomes, which results in
(10)  
(11) 
Now if is uncorrelated, variances just add up, so
(12) 
where and are the optimal uncertainty constants for the observable pairs and , respectively. Since the variance is concave, this inequality holds also for all convex combinations of uncorrelated states, i.e., for all separable states [4].
Hence if (12) is violated, must be entangled. Of course, there is also an uncertainty bound for the observable pair . So the interesting range allowing the conclusion “ is entangled” is marked by
(13) 
For angular momentum measurements, (12) can be seen [24] as a spinsqueezing criterion. As such, it requires the same experimental data as other spin squeezing criteria, see [25]; [26], namely only a measurement of first and second moments of the total angular momentum. In contrast to entanglement criteria based on single outcomes, this requirement is very advantageous in typical experimental implementation, especially including many particle systems, see [27].
We further sharpen this criterion by applying it to the observable pairs and . In this way we get two convex regions of pairs : A larger one containing the pairs achievable with arbitrary states, given by the bounds of the type , and a smaller one attainable by separable states, given by the bounds of the type . As Fig. 4 shows, this increases the parameter range for which entanglement can be certified. The linear uncertainty bound with equal weights as a function of the local noise, evaluated for measurements and on separable and entangled states is shown by Fig. 8.
Entanglement detection with noisy detectors
The generalization to POVMs increases the possibilities for entanglement detection. Suppose for the sake of discussion that before hitting the detector each subsystem goes through a known noisy channel. This typically increases variance [28], so traditional spin squeezing inequalities would often fail to detect entanglement. Indeed the state after the action of the noisy channels may well fail to be entangled. On the other hand, we might be interested in the presence of entanglement before the action of the noise. This is the appropriate view when the noise is inherent in the detection process. The noise is thus applied in the Heisenberg picture, turning even a standard projection valued measurement into a proper POVM. This might easily find entanglement, which would go undetected by a direct application of the spin squeezing criterion.
These possibilities are shown in Fig. 4 by superimposing the entanglement detection regions for three different noise levels of a partially depolarizing channel , where is the maximally mixed state, and is a noise parameter. Increasing shifts the diagram towards larger variances, but even for a modest noise level of the entanglement detection region lies entirely in the region where traditional spin squeezing (corresponding to ) would never find any entanglement.
Conclusions and Outlook
We provided an algorithm for determining the optimal uncertainty bounds for two arbitrary observables. The precision of the bound is controlled as a duality gap, so terminating the iteration at any step gives a certified lower uncertainty bound together with an error estimate.
The method can, in principle, be extended to more observables, or to variances based not on quadratic but higher order deviations. However, this would increase the dimension of the geometric problem. Thus at every new approximation step one has to determine the intersection of the polytope with the new supporting hyperplane. This requires a better bookkeeping of the topological structure of the polytopes, and a local version of the vertex enumeration problem [29].
The inequalities derived here have an immediate application to entanglement detection by generalized spin squeezing criteria. The possibility to use arbitrary observables (rather than orthogonal angular momentum components) greatly increases the versatility of this method.
It is an apparently open problem how strong the method becomes with arbitrary , , i.e. is every entangled state violating a local uncertainty relation. The problem has been studied carefully for orthogonal spin components [5]; [6], but we do not know of a characterization of the (un)detectable, possibly entangled states.
Acknowledgements.
We gratefully acknowledge inspiring conversations and email exchange with Marcus Cramer, Otfried Gühne, Géza Tóth, Kais Abdelkhalek, David Reeb and Terry Farrelly.We also acknowledge the financial support from the RTG 1991 and CRC 1227 DQmat funded by the DFG and the collaborative research project Q.comQ funded by the BMBF.
Appendix A Appendix
a.1 Strict monotonicity of the gap
We will consider measurements and which could also be represented by two general POVMs. Furthermore, we will assume that we already have an initial outer approximation of the corresponding set by a polyhedron , constructed from initial directions . Such a set of directions can be constructed by taking the face normals of a cube, as in Fig. 3.
Let be a vertex of on which the minimum of is attained, i.e. and take as new direction such as as new set of directions, in every step. Then the bound will either increase after a finite round of such steps or attain a global minimum on .
Proof.
We will show that, by taking as above, the point will be removed from the resulting polyhedron whenever is not in . From this statement we can conclude that: If is removed, and there is no point in which attains the value , the new bound will fail to increase if and only if there is another extremal point in that also attains the same minimal value of . However, because is a finite set, all those points will be removed from it after a finite round of steps. Hence, the bound will increase after a finite round of steps.
In the alternative case, when is in , the upper bound will also be attained on , such that . Therefore, we already would have found the optimal bound.
Consider a fixed and the levelset . The set is convex and obviously contains and as subsets. is a quadratic functional, so its gradient is well defined everywhere. Moreover, the direction is the normal direction of the tangent space of at the point . Due to convexity, is described by linear inequalities corresponding to its tangent spaces, which implies that
(14) 
for all points from , and so, for all from and , as well. More precisely, we have , with a minimum that is attained uniquely on , because is strictly concave. If we now consider the new set of direction and its corresponding inequalities for constructing , see (5), we have
(15) 
for all . Hereby, equality in the last part of (15) holds if and only if . In all other cases the functional separates the set from the point . ∎
a.2 Precision per Step
A crucial property of any numerical method is its performance. For the method provided in this work we measure it by the numerical precision in comparison to the number of steps required. As a benchmarking we computed several random examples and illustrated three of them in Fig. 5. We observe that in the typical case, there are two different kinds of scaling behaviour. During the first part of steps, the precision increases slower than in the second. In the regime of the first steps the algorithm improves the outer approximation at very different points. However, once the outer polyherdron is fine enough, the algorithm generates vertices close to the actual optimum. If this optimum is unique, the algorithm will go to a regime where all improvements are made locally. In Fig. 5 this transition from global to local optimization is marked. After this point the improvement of precision per step, measured in decimal places of the gap , scales linear.
sample  method  size  steps 

(i)  Haar random  
(ii)  \makecellHaar random eigenvectors  
uniform dist. spectrum  
(iii)  Haar random 
This behaviour agrees with the worst case example given in Fig. 6, where we considered two orthogonal angular momentum components, and . Here rotations around the axis, in terms of spin components, impose a rotational degree of freedom on the set of all linear combinations of the operators , and . This results in a region that is rotational symmetric, as well. As the variance sum itself shares the same symmetry the optimum on will be attained on a continuum of points.
This is a worst case scenario, because our method has to improve the outer approximation on a continuum of points. Hence, no localization of the algorithm can be expected, and no transition in a linear scaling regime happens, for this compare Fig. 5 with Fig. 6. Note that, in this highly symetrical case there is no need to performe the algorithm on the whole set . If we take care of the ounderlying symmetry the problem reduces to a fast scaling problem on a two dimensioal subset of again.
a.3 Examples
Nonorthogonal spin components:
We computed the minimal uncertainty for a measurement of two spins components that span an angle . Without loss of generality we can assume one of the components to be given by and the other one to lie in the plane. So we can take
(16) 
The value of the uncertainty bound in dependence of the angle is shown in Fig. 7 and Tab. 2.
spin/angle  

1  0  0.0378  0.1431  0.2910  0.4365 
2  0  0.0743  0.2754  0.5318  0.7478 
3  0  0.1108  0.3984  0.7444  1.0131 
Entanglement detection with local noise
Footnotes
 For hermitian operators we define the variance of a state by .
References
 Earl Kennard. Zur Quantenmechanik einfacher Bewegungstypen. Zeitschr. Phys., 44:326–352, 1927.
 Werner Heisenberg. Über den anschaulichen Inhalt der quantentheoretischen Kinematik und Mechanik. Zeitschrift für Physik, 43(34):172–198, 1927.
 For hermitian operators we define the variance of a state by .
 Holger F. Hofmann and Shigeki Takeuchi. Violation of local uncertainty relations as a signature of entanglement. Phys.Rev. A., 68:032103, 2003. and arXiv:quantph/0212090.
 Otfried Gühne and Géza Tóth. Entanglement detection. Physics Reports, 474(1–6):1 – 75, 2009.
 Otfried Gühne. Detecting quantum entanglement: entanglement witnesses and uncertainty relations. PhD thesis, Universität Hannover, 2004.
 Lars Dammeier, René Schwonnek, and Reinhard F. Werner. Uncertainty relations for angular momentum. New Journal of Physics, 17(9):093046, 2015. and arXiv:1505.00049.
 Qiongyi He, ShiGuo Peng, Peter D. Drummond, and Margaret D. Reid. Planar quantum squeezing and atom interferometry. Phys. Rev. A, 84:022107, Aug 2011. and arXiv:1101.0448.
 Alastair A. Abbott, PierreLouis Alzieu, Michael J. W. Hall, and Cyril Branciard. Tight stateindependent uncertainty relations for qubits. Mathematics, 4(1), 2016. and arXiv:1512.02383.
 Paul Busch, Pekka Lahti, and Reinhard F. Werner. Measurement uncertainty relations. Journal of Mathematical Physics, 55(4):042111, 2014. and arXiv:1312.4392.
 Reinhard F. Werner. Uncertainty relations for general phase spaces. Frontiers of Physics, 11(3):110305, 2016. and arXiv:1601.03843.
 Paul Busch, Pekka Lahti, and Reinhard F. Werner. Sharp uncertainty relations for number and angle. 2016. and arXiv:1604.00566.
 Yichen Huang. Variancebased uncertainty relations. Phys. Rev. A, 86:024101, Aug 2012. and arXiv:1012.3105.
 Hans Maassen and Jos Uffink. Generalized entropic uncertainty relations. Phys. Rev. Lett., 60:1103–1106, 1988.
 Kais Abdelkhalek, René Schwonnek, Hans Maassen, Fabian Furrer, Jörg Duhme, Philippe Raynal, BertholdGeorg Englert, and Reinhard F Werner. Optimality of entropic uncertainty relations. International Journal of Quantum Information, 13(06):1550045, 2015. and arXiv:1509.00398.
 Spiros Kechrimparis and Stefan Weigert. Heisenberg uncertainty relation for three canonical observables. Phys. Rev. A, 90:062118, Dec 2014. and arXiv:1407.0083.
 Spiros Kechrimparis and Stefan Weigert. Preparational uncertainty relations for n continuous variables. Mathematics, 4(3), 2016. and arXiv:1606.09148.
 René Schwonnek, David Reeb, and Reinhard F. Werner. Measurement uncertainty for finite quantum observables. Mathematics, 4(2), 2016. and arXiv:1604.00382.
 David Deutsch. Uncertainty in quantum measurements. Phys. Rev. Lett., 50:631–633, Feb 1983.
 Howard Percy Robertson. The uncertainty principle. Phys. Rev., 34:163–164, 1929.
 Lorenzo Maccone and Arun K. Pati. Stronger uncertainty relations for all incompatible observables. Phys. Rev. Lett., 113:260401, 2014. and arXiv:1407.0338.
 Günther Ludwig. Foundations of Quantum Mechanics I. Springer Berlin Heidelberg, Berlin, Heidelberg, 1983.
 Teiko Heinosaari and Mário Ziman. The mathematical language of quantum theory: from uncertainty to entanglement. Cambridge University Press, 2011.
 Jian Ma, Xiaoguang Wang, ChangPu Sun, and Franco Nori. Quantum spin squeezing. Physics Reports, 509(2–3):89 – 165, 2011. and arXiv:1011.2978.
 Anders S. Sørensen, Luming Duan, Ignacio Cirac, and Peter Zoller. Manyparticle entanglement with bose–einstein condensates. Nature, 409(6816):63–66, 2001. and arXiv:quantph/0006111.
 Anders S. Sørensen and Klaus Mølmer. Entanglement and extreme spin squeezing. Phys. Rev. Lett., 86:4431–4434, May 2001. and arXiv:quantph/0011035.
 Bernd Lücke, Jan Peise, Giuseppe Vitagliano, Jan Arlt, Luis Santos, Géza Tóth, and Carsten Klempt. Detecting multiparticle entanglement of dicke states. Phys. Rev. Lett., 112:155304, Apr 2014. and arXiv:1403.4542.
 Bernd Lücke, Manuel Scherer, Jens Kruse, Luca Pezzé, Frank Deuretzbacher, Phillip Hyllus, Oliver Topic, Jan Peise, Wolfgang Ertmer, Jan Arlt, Luis Santos, Augusto Smerzi, and Carsten Klempt. Twin matter waves for interferometry beyond the classical limit. Science, 2011. and arXiv:1204.4102.
 David Avis and Komei Fukuda. A pivoting algorithm for convex hulls and vertex enumeration of arrangements and polyhedra. Disc. Comput. Geom., 8:295–313, 1992.