
Certification of the maximally entangled state using non-projective measurements
Shubhayan Sarkar1, ∗
1Center for Theoretical Physics, Polish Academy of Sciences, Aleja Lotników 32/46, 02-668 Warsaw, Poland
In recent times, device-independent certification of quantum states has been one of the inten-
sively studied areas in quantum information. However, all such schemes utilise projective measure-
ments which are practically difficult to generate. In this work, we consider the one-sided device-
independent (1SDI) scenario and propose a self-testing scheme for the two-qubit maximally entan-
gled state using non-projective measurements, in particular, three three-outcome extremal POVM’s.
We also analyse the robustness of our scheme against white noise.
Introduction.— The existence of nonlocal correlations,
as was first realised by Einstein, Podolski and Rosen
in 1935 [1] as a paradox and then subsequently by
Schr¨
odinger in the same year [2], is one of the most in-
triguing features of quantum theory. Consequently, Bell
[3,4] proposed a mathematical formulation to detect
whether quantum theory is inherently nonlocal or not,
and thus it is commonly referred to as Bell nonlocality.
Apart from its relevance in the foundations of physics,
Bell nonlocality has given rise to an enormous number
of applications in computation, communication and in-
formation theory [5].
A recent well profound application of nonlocality
is device-independent (DI) certification where assum-
ing quantum theory and some other physically well-
motivated assumptions, the statistics obtained from a
black box are enough to validate the underlying mecha-
nism inside it. The strongest DI certification is referred
to as self-testing. First introduced in [6], self-testing al-
lows one to certify the underlying quantum states and
the measurements, upto some freedom based on the
maximal violation of a Bell inequality [7]. In recent
times, there has been increased interest to find proto-
cols to self-test various quantum systems due to their
applicability in various quantum information tasks. De-
spite the progress in designing schemes to self-test vari-
ous quantum states using projective measurements, for
instance, Refs. [8–18], there is no scheme that utilises
non-projective measurements.
The precision to experimentally generate sets of pro-
jective measurements, which are pre-requisite to ob-
serve any form of non-locality, reduces as the dimen-
sion of the system grows [for instance see [?]]. Thus,
a natural question arises whether noisy measurements
or non-projective measurements can also be used to self-
test quantum states. Further on to self-test any state or
measurement, one needs to observe the maximal viola-
tion of an inequality which touches the set of quantum
correlations, or simply the quantum set, at a particu-
lar point. It is also an open question in quantum foun-
dations whether a point on the boundary of the quan-
∗sarkar@cft.edu.pl
tum set in some scenarios can be saturated by only non-
projective measurements.
Self-testing quantum states using non-projective mea-
surements is not possible in the standard Bell scenario.
The reason being that the maximal violation of Bell in-
equalities can always be achieved by projective mea-
surements. Consequently, we consider another form of
nonlocality, known as quantum steering [19–21]. To wit-
ness quantum steering, one needs to consider the Bell
scenario with an additional assumption that one of the
parties is trusted. In the DI regime, this is referred to
as one-sided device-independent (1SDI) scenario. Cer-
tification of quantum states and measurements in 1SDI
scenario has gained recent interest [22–28] as they are
more robust to noise and require detectors with lower
efficiencies when compared to fully DI scenarios [29,30].
In this work, we provide the first scheme to certify the
two-qubit maximally entangled state
ϕ+=1
√2(|00⟩+|11⟩)(1)
using three three-outcome non-projective extremal mea-
surements in the 1SDI scenario. For this purpose, we
first construct a steering inequality with two parties
such that each of them chooses three inputs and gets
three outputs. We then use the maximal violation of this
steering inequality to obtain the self-testing result. We
finally show that our scheme is highly robust when the
states and measurements are mixed with white noise.
Preliminaries—Before proceeding to the results, let us
first describe the scenario and notions used throughout
this work.
Extremal positive operator valued measure (POVM). Any
measurement in quantum theory, usually referred to as
a POVM, is represented as M={Ma}where Maare the
measurement elements corresponding to the a-th out-
come of M. These elements are positive semi-definite
operators and PaMa=. Now, a POVM that can not
be expressed as a convex combination of other POVM’s
is defined as an extremal POVM. As shown in [31] the
elements Maof any rank-one extremal POVM can be
expressed as Ma=λa|νa⟩⟨νa|where λa≥0 and the
elements are linearly independent.
Quantum steering scenario. In this work, we consider a
simple scenario to witness quantum steering, consisting
arXiv:2210.14099v2 [quant-ph] 5 Jun 2023