Einstein-Podolsky-Rosen steering paradox “2=1” for $N$ qubits

Abstract

Einstein-Podolsky-Rosen (EPR) paradox highlights the absence of a local realistic explanation for quantum mechanics, and shows the incompatibility of the local-hidden-state models with quantum theory. For $N$-qubit states, or more importantly, the $N$-qubit mixed states, we present the EPR steering paradox in the form of the contradictory equality “2=1”. We show that the contradiction holds for any $N$-qubit state as long as both “the pure state requirement” and “the measurement requirement” are satisfied. This also indicates that the EPR steering paradox exists in more general cases. Finally, we give specific examples to demonstrate and analyze our arguments.

Keywords: Einstein-Podolsky-Rosen steering; quantum paradox; $N$ qubits states.

I Introduction

The quantum paradox serves as a powerful tool in elucidating the fundamental distinction between the quantum theory and the classical theory. Quantum correlations play a central role in the study of quantum information and quantum mechanics. Among the quantum correlations, quantum entanglement and Bell’s nonlocality are the first to be proposed and studied. In 1935, Einstein, Podolsky, and Rosen (EPR) published their famous article “Can Quantum-Mechanical Description of Physical Reality Be Considered Complete?” EPR1935 , which questioned the completeness of quantum mechanics under the assumptions of locality and reality. This is nowadays well-known as the EPR paradox. Soon after the publication of the EPR paper, Schrödinger introduced two important concepts, namely, quantum entanglement and quantum steering Schr1935 ; Schr1936 . Quantum entanglement distinguishes quantum theory from classical theory. And quantum steering is closely related to “the spooky action at a distance”. However, the idea of steering did not receive considerable attention or advancement until the year 2007, at which point Wiseman et al. presented a meticulous definition by utilizing quantum information concepts Wiseman2007 ; jones2007 . So far, quantum steering has been widely applied in various fields preskill1998 ; Galindo2002 ; Ekert1996 ; Childs2010 .

Steering is a quantum correlation between entanglement Quantum entanglement2009 and Bell nonlocality Bell nonlocality2014 ; Quantum machine learning2017 ; bell1964 . Steering is used to describe the situation in a bipartite system. When people use different observables to detect one of the particles, it will cause the corresponding other particle to collapse to a certain state. In practice, Alice prepares a bipartite quantum state, and she sends one of the particles to Bob. They each measure the particles in their hands and communicate over a classical channel. It is Alice’s task to convince Bob that Alice has prepared a pair of entangled state. In the process, Bob needs to assess the correctness of the assumptions of quantum mechanics and acknowledge the measurements as described by quantum mechanics. Specifically, Bob can disbelieve Alice’s equipment and measurements. However, in this case, Bob needs to rule out the influence of hidden variables on the measurement results by the measurements he has in hand. Bob can fully trust his own equipment and results. If Bob cannot explain the measurement results on his side with the local-hidden states (LHS), he must recognize that Alice has prepared an entangled bipartite state. Only EPR steering states can accomplish this task. And quantum steering is an asymmetric quantum nonlocality. That is, in some cases Alice can steer Bob, but in turn Bob cannot steer Alice Asymmetric2010 ; Bowles2014 ; Quintino2015 ; Wollmann2016 ; HV2012 . Based on some properties of quantum steering, Chen et al. proposed the EPR steering paradox “2=1” chenjl2016 , where “2” is the quantum result and “1” is the corresponding result of LHS models. They verified the EPR steering state by the contradiction between quantum mechanics and classical theory. In the 2-setting EPR steering protocol, they found that any 2-qubit entangled pure state possesses the contradiction. Thereafter, Liu et al. found that such a contradiction was also valid for a specific 4-qubit entangled mixed state Liu2021 . In other words, the discussion of the EPR steering paradox “2=1” has been limited to arbitrary 2-qubit pure state and a special 4-qubit mixed state.

The purpose of this paper is to study the EPR steering paradox “2=1” for $N$ qubits. Based on the 2-setting steering protocol, we have obtained such EPR steering paradox “2=1” for $N$-qubit entangled states. In this work, we demonstrate that any $N$-qubit state can lead to the contradiction, provided that both “the pure state requirement” and “the measurement requirement” are fulfilled simultaneously. The paper is organized as follows. In Sec. II, we propose a theorem for $N$-qubit quantum states that contains two requirements : “the pure state requirement” and “the measurement requirement”. In Sec. III, We obtain the EPR steering paradox “2=1” for the $N$-qubit states under Bell-like basis measurement. In Sec. IV, we apply the results to the 2-qubit mixed states and obtain a corollary that there is no EPR steering paradox “2=1” for 2-qubit mixed states. Finally, we conclude with a summary in Sec. V. Some detailed proofs are given in the Appendix A and B.

II EPR steering paradox “2=1” for $N$-qubit states

II.1 “2=1” for the 2-qubit pure state and the 4-qubit mixed state

To make the paper be self-contained, in this subsection let us make a brief review. In 2016, Chen et al. first simplified the EPR steering paradox as a contradiction “2=1” for any 2-qubit pure entangled state chenjl2016 . They analyzed a 2-qubit pure entangled state given by

$$\rho_{AB}=\left|\Psi\left(\theta\right)\right\rangle\left\langle\Psi\left(\theta\right)\right|,$$ (1)

where

$$\left|\Psi\left(\theta\right)\right\rangle=\cos\theta\left|0\right\rangle\left|0\right\rangle+\sin\theta\left|1\right\rangle\left|1\right\rangle,$$ (2)

with $\theta\in\left(0,\pi/2\right)$. They chose the 2-setting protocol as $\left\{\hat{z},\hat{x}\right\}$. In the protocol, Bob asks Alice to carry out either one of two possible projective measurements on her qubit along the $z$-direction and the $x$-direction, i.e.,

$$\begin{array}[c]{l}{{P}_{0}^{\hat{z}}=}\left|0\right\rangle\left\langle 0\right|,\\ {{P}_{1}^{\hat{z}}=}\left|1\right\rangle\left\langle 1\right|,\\ {{P}_{0}^{\hat{x}}=}\left|+\right\rangle\left\langle+\right|,\\ {{P}_{1}^{\hat{x}}=}\left|-\right\rangle\left\langle-\right|,\end{array}$$ (3)

where $\left|\pm\right\rangle=(1/\sqrt{2})(\left|0\right\rangle\pm\left|1\right\rangle)$, and to inform him of the measurement results of “$a$” (where $a=0,1$). Then Bob’s four unnormalized conditional states are

$$\begin{array}[c]{l}\tilde{\rho}{{}_{0}^{\hat{z}}=\cos}^{2}\theta\left|0\right\rangle\left\langle 0\right|,\quad\tilde{\rho}{{}_{1}^{\hat{z}}=\sin}^{2}{\theta}\left|1\right\rangle\left\langle 1\right|,\\ \tilde{\rho}{{}_{0}^{\hat{x}}=}\dfrac{1}{2}\left(\cos\theta\left|0\right\rangle+{\sin\theta}\left|1\right\rangle\right)\left(\cos\theta\left\langle 0\right|+{\sin\theta}\left\langle 1\right|\right),\\[9.0pt] \tilde{\rho}{{}_{1}^{\hat{x}}=}\dfrac{1}{2}\left(\cos\theta\left|0\right\rangle-{\sin\theta}\left|1\right\rangle\right)\left(\cos\theta\left\langle 0\right|-{\sin\theta}\left\langle 1\right|\right).\end{array}$$ (4)

If Bob’s states have a LHS description, then there exists an ensemble $\{\wp_{\xi}\rho_{\xi}\}$ and a stochastic map $\wp(a|\hat{n},\xi)$ satisfying

$$\tilde{\rho}_{a}^{\hat{n}}={\displaystyle\sum\limits_{\xi}}\wp\left(a|\hat{n},\xi\right)\wp_{\xi}\rho_{\xi},$$ (5)

and

$${\displaystyle\sum\limits_{\xi}}\wp_{\xi}\rho_{\xi}=\rho_{B}.$$ (6)

Here $\wp_{\xi}$ and $\wp\left(a|\hat{n},\xi\right)$ are probabilities satisfying

$${\sum\limits_{\xi}}\wp_{\xi}=1,$$ (7)

and

$${\sum\limits_{a}}\wp\left(a|\hat{n},\xi\right)=1,$$ (8)

for a fixed $\xi$, and $\rho_{B}=\mathrm{tr}_{A}\left(\rho_{AB}\right)$ is Bob’s reduced density matrix Wiseman2007 ; jones2007 .

Then Bob’s four unnormalized conditional states satisfy Eq. (5), and one has

$$\begin{array}[c]{c}\widetilde{\rho}_{0}^{\hat{z}}={\displaystyle\sum\limits_{\xi}}\wp\left(0|\hat{z},\xi\right)\wp_{\xi}\rho_{\xi},\\ \widetilde{\rho}_{1}^{\hat{z}}={\displaystyle\sum\limits_{\xi}}\wp\left(1|\hat{z},\xi\right)\wp_{\xi}\rho_{\xi},\\ \widetilde{\rho}_{0}^{\hat{x}}={\displaystyle\sum\limits_{\xi}}\wp\left(0|\hat{x},\xi\right)\wp_{\xi}\rho_{\xi},\\ \widetilde{\rho}_{1}^{\hat{x}}={\displaystyle\sum\limits_{\xi}}\wp\left(1|\hat{x},\xi\right)\wp_{\xi}\rho_{\xi}.\end{array}$$ (9)

Because the four states of Eq. (4) are pure states, it is sufficient to take $\xi$ from $1$ to $4$. Eq. (9) can be write as

$$\begin{array}[c]{c}\widetilde{\rho}_{0}^{\hat{z}}=\wp\left(0|\hat{z},1\right)\wp_{1}\rho_{1}+\wp\left(0|\hat{z},2\right)\wp_{2}\rho_{2}+\wp\left(0|\hat{z},3\right)\wp_{3}\rho_{3}+\wp\left(0|\hat{z},4\right)\wp_{4}\rho_{4},\\ \widetilde{\rho}_{1}^{\hat{z}}=\wp\left(1|\hat{z},1\right)\wp_{1}\rho_{1}+\wp\left(1|\hat{z},2\right)\wp_{2}\rho_{2}+\wp\left(1|\hat{z},3\right)\wp_{3}\rho_{3}+\wp\left(1|\hat{z},4\right)\wp_{4}\rho_{4},\\ \widetilde{\rho}_{0}^{\hat{x}}=\wp\left(0|\hat{x},1\right)\wp_{1}\rho_{1}+\wp\left(0|\hat{x},2\right)\wp_{2}\rho_{2}+\wp\left(0|\hat{x},3\right)\wp_{3}\rho_{3}+\wp\left(0|\hat{x},4\right)\wp_{4}\rho_{4},\\ \widetilde{\rho}_{1}^{\hat{x}}=\wp\left(1|\hat{x},1\right)\wp_{1}\rho_{1}+\wp\left(1|\hat{x},2\right)\wp_{2}\rho_{2}+\wp\left(1|\hat{x},3\right)\wp_{3}\rho_{3}+\wp\left(1|\hat{x},4\right)\wp_{4}\rho_{4}.\end{array}$$ (10)

And owing to the fact that a pure state cannot be obtained by convex combination of other pure states, one has

$$\begin{array}[c]{l}\tilde{\rho}{{}_{0}^{\hat{z}}=}\wp_{1}\rho_{1},\\ \tilde{\rho}{{}_{1}^{\hat{z}}=}\wp_{2}\rho_{2},\\ \tilde{\rho}{{}_{0}^{\hat{x}}=}\wp_{3}\rho_{3},\\ \tilde{\rho}{{}_{1}^{\hat{x}}=}\wp_{4}\rho_{4}.\end{array}$$ (11)

Here

$$\wp\left(0|\hat{z},1\right)=\wp\left(1|\hat{z},2\right)=\wp\left(0|\hat{x},3\right)=\wp\left(1|\hat{x},4\right)=1,$$

and other $\wp\left(a|\hat{n},\xi\right)=0$. By summing four terms up in Eq. (11) and taking trace, the left side gives $\mathrm{tr}(\tilde{\rho}{{}^{\hat{z}}}+\tilde{\rho}{{}^{\hat{x}}})=2\mathrm{tr}(\rho_{B})=2$. But the right side gives $\mathrm{tr}(\wp_{1}\rho_{1}+\wp_{2}\rho_{2}+\wp_{3}\rho_{3}+\wp_{4}\rho_{4})=\mathrm{tr}(\rho_{B})=1$, then one can obtain the contradiction “2=1”, i.e., the EPR steering paradox, for any 2-qubit pure entangled state.

After that, in 2021, Liu et al. found a 4-qubit mixed entangled state

$$\rho\left(\theta\right)=\cos^{2}\theta\left|LC_{4}\right\rangle\left\langle LC_{4}\right|+\sin^{2}\theta\left|LC_{4}^{\prime}\right\rangle\left\langle LC_{4}^{\prime}\right|,$$ (12)

where

$$\begin{array}[c]{c}\left|LC_{4}\right\rangle=\dfrac{1}{2}\left(\left|0000\right\rangle+\left|1100\right\rangle+\left|0011\right\rangle-\left|1111\right\rangle\right),\\[9.0pt] \left|LC_{4}^{\prime}\right\rangle=\dfrac{1}{2}\left(\left|0100\right\rangle+\left|1000\right\rangle+\left|0111\right\rangle-\left|1011\right\rangle\right),\end{array}$$ (13)

are the linear cluster states Liu2021 . Alice prepares the state $\rho\left(\theta\right)$ as in Eq. (12). She keeps 1,2 particles and sends 3,4 particles to Bob. In the 2-setting steering protocol $\left\{\hat{n}_{1},\hat{n}_{2}\right\}$ $\left(\hat{n}_{1}\neq\hat{n}_{2}\right)$, with

$$\begin{array}[c]{c}\hat{n}_{1}=\sigma_{z}\sigma_{z}\equiv zz,\\ \hat{n}_{2}=\sigma_{y}\sigma_{x}\equiv yx.\end{array}$$ (14)

In the protocol, Bob asks Alice to carry out either one of two possible projective measurements on her qubits, i.e.,

$$\begin{array}[c]{l}P_{00}^{\hat{n}_{1}}=\left|00\right\rangle\left\langle 00\right|,\\ P_{01}^{\hat{n}_{1}}=\left|01\right\rangle\left\langle 01\right|,\\ P_{10}^{\hat{n}_{1}}=\left|10\right\rangle\left\langle 10\right|,\\ P_{11}^{\hat{n}_{1}}=\left|11\right\rangle\left\langle 11\right|,\\ P_{00}^{\hat{n}_{2}}=\left|\uparrow+\right\rangle\left\langle\uparrow+\right|,\\ P_{01}^{\hat{n}_{2}}=\left|\uparrow-\right\rangle\left\langle\uparrow-\right|,\\ P_{10}^{\hat{n}_{2}}=\left|\downarrow+\right\rangle\left\langle\downarrow+\right|,\\ P_{11}^{\hat{n}_{2}}=\left|\downarrow-\right\rangle\left\langle\downarrow-\right|,\end{array}$$ (15)

where $|\pm\rangle=(1/\sqrt{2})(\left|0\right\rangle\pm\left|1\right\rangle)$, $|\updownarrow\rangle=(1/\sqrt{2})(\left|0\right\rangle\pm{\rm i}\left|1\right\rangle)$. After Alice’s measurement, Bob’s unnormalized conditional states are

$$\begin{array}[c]{l}\tilde{\rho}_{00}^{\hat{n}_{1}}=\dfrac{1}{4}\cos^{2}\theta\left(\left|00\right\rangle+\left|11\right\rangle\right)\left(\left\langle 00\right|+\left\langle 11\right|\right),\\[9.0pt] \tilde{\rho}_{01}^{\hat{n}_{1}}=\dfrac{1}{4}\sin^{2}\theta\left(\left|00\right\rangle+\left|11\right\rangle\right)\left(\left\langle 00\right|+\left\langle 11\right|\right),\\[9.0pt] \tilde{\rho}_{10}^{\hat{n}_{1}}=\dfrac{1}{4}\sin^{2}\theta\left(\left|00\right\rangle-\left|11\right\rangle\right)\left(\left\langle 00\right|-\left\langle 11\right|\right),\\[9.0pt] \tilde{\rho}_{11}^{\hat{n}_{1}}=\dfrac{1}{4}\cos^{2}\theta\left(\left|00\right\rangle-\left|11\right\rangle\right)\left(\left\langle 00\right|-\left\langle 11\right|\right),\\[9.0pt] \tilde{\rho}_{00}^{\hat{n}_{2}}=\dfrac{1}{8}\left(\left|00\right\rangle+{\rm i}\left|11\right\rangle\right)\left(\left\langle 00\right|-{\rm i}\left\langle 11\right|\right),\\[9.0pt] \tilde{\rho}_{01}^{\hat{n}_{2}}=\dfrac{1}{8}\left(\left|00\right\rangle-{\rm i}\left|11\right\rangle\right)\left(\left\langle 00\right|+{\rm i}\left\langle 11\right|\right),\\[9.0pt] \tilde{\rho}_{10}^{\hat{n}_{2}}=\dfrac{1}{8}\left(\left|00\right\rangle-{\rm i}\left|11\right\rangle\right)\left(\left\langle 00\right|+{\rm i}\left\langle 11\right|\right),\\[9.0pt] \tilde{\rho}_{11}^{\hat{n}_{2}}=\dfrac{1}{8}\left(\left|00\right\rangle+{\rm i}\left|11\right\rangle\right)\left(\left\langle 00\right|-{\rm i}\left\langle 11\right|\right).\end{array}$$ (16)

If Bob’s states have a LHS description, they must satisfy Eq. (5) and Eq. (6). Because the eight states of Eq. (16) are pure states, it is sufficient to take $\xi$ from $1$ to $8$. And a pure state cannot be obtained by convex combination of other pure states, one has

$$\begin{array}[c]{c}\tilde{\rho}_{00}^{\hat{n}_{1}}=\wp_{1}\rho_{1},\\ \tilde{\rho}_{01}^{\hat{n}_{1}}=\wp_{2}\rho_{2},\\ \tilde{\rho}_{10}^{\hat{n}_{1}}=\wp_{3}\rho_{3},\\ \tilde{\rho}_{11}^{\hat{n}_{1}}=\wp_{4}\rho_{4},\\ \tilde{\rho}_{00}^{\hat{n}_{2}}=\wp_{5}\rho_{5},\\ \tilde{\rho}_{01}^{\hat{n}_{2}}=\wp_{6}\rho_{6},\\ \tilde{\rho}_{10}^{\hat{n}_{2}}=\wp_{7}\rho_{7},\\ \tilde{\rho}_{11}^{\hat{n}_{2}}=\wp_{8}\rho_{8}.\end{array}$$ (17)

By summing eight terms up in Eq. (17) and taking trace, the left side gives $\mathrm{tr}(\tilde{\rho}^{\hat{n}_{1}}+\tilde{\rho}^{\hat{n}_{2}})=2\mathrm{tr}\left(\rho_{B}\right)=2$. But the right side gives $\mathrm{tr}\big{(}{\sum_{\xi=1}^{8}}\wp_{\xi}\rho_{\xi}\big{)}=\mathrm{tr}(\rho_{B})=1$, then one can obtained an EPR steering paradox “2=1” for the specific 4-qubit mixed entangled state.

II.2 “2=1” for $N$-qubit states

In this subsection, we study the EPR steering paradox “2=1” for $N$-qubit states. Our result is the following theorem.

Theorem 1.

In the 2-setting steering protocol $\left\{\hat{n}_{1},\hat{n}_{2}\right\}$, Alice and Bob share an $N$-qubit state $\rho_{AB}$. Assume that Alice measures along $\hat{n}_{1}$ and $\hat{n}_{2}$, and then Bob obtains $\tilde{\rho}_{a}^{\hat{n}_{1}}$ and $\tilde{\rho}_{a^{\prime}}^{\hat{n}_{2}}$, respectively, where $a,a^{\prime}$ is the measurement result of Alice. There will be a contradiction of “2=1” if $\rho_{AB}$ satisfies simultaneously “the pure state requirement” and “the measurement requirement”. The two requirements are:

1. The pure state requirement: Bob’s unnormalized conditional states $\{\tilde{\rho}_{a}^{\hat{n}_{1}}\}$ and $\{\tilde{\rho}_{a^{\prime}}^{\hat{n}_{2}}\}$ are all pure states.

2. The measurement requirement: any one of $\{\tilde{\rho}_{a}^{\hat{n}_{1}}\}$ is different from any one of $\{\tilde{\rho}_{a^{\prime}}^{\hat{n}_{2}}\}$.

Let Alice and Bob share an $N$-qubit entangled state

$$\rho_{AB}=\sum_{\alpha}p_{\alpha}\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle\left\langle\psi_{AB}^{\left(\alpha\right)}\right|.$$ (18)

Alice has $M\left(M<N\right)$ particles and Bob has $(N-M)$ particles. In the 2-setting steering protocol $\left\{\hat{n}_{1},\hat{n}_{2}\right\}$ $\left(\text{with }\hat{n}_{1}\neq\hat{n}_{2}\right)$, Alice performs $2^{M+1}$ projective measurements, each of them measuring $M$ particles of Alice. For each projective measurement $\hat{P}_{a}^{\hat{n}_{k}}$, Bob obtains the corresponding unnormalized state $\tilde{\rho}_{a}^{\hat{n}_{k}}=\mathrm{tr}_{A}\left[\left({{P}_{a}^{\hat{n}_{k}}\otimes\mathds{1}}\right)\rho_{AB}\right]$, with $\hat{n}_{k}$ the measurement direction, $a$ the Alice’s measurement result, ${\mathds{1}}$ the $2^{N-M}\times 2^{N-M}$ identity matrix. The wave-function $|\psi_{AB}^{\left(\alpha\right)}\rangle$ may be written as

$$\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=\sum_{i}\left(s_{i+}^{\left(\alpha\right)}\left|+\phi_{i}\right\rangle\left|\eta_{i+}^{\left(\alpha\right)}\right\rangle+s_{i-}^{\left(\alpha\right)}\left|-\phi_{i}\right\rangle\left|\eta_{i-}^{\left(\alpha\right)}\right\rangle\right),$$ (19)

or

$$\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=\sum_{j}\left(t_{j+}^{\left(\alpha\right)}\left|+\varphi_{j}\right\rangle\left|\varepsilon_{j+}^{\left(\alpha\right)}\right\rangle+t_{j-}^{\left(\alpha\right)}\left|-\varphi_{j}\right\rangle\left|\varepsilon_{j-}^{\left(\alpha\right)}\right\rangle\right),$$ (20)

where $i,j=1,2,\cdots,2^{M-1}$. Eq. (19) and Eq. (20) are two representations of $\left|\psi_{AB}^{\alpha}\right\rangle$ in different bases. Here $\left|\pm\phi_{i}\right\rangle$ and $\left|\pm\varphi_{j}\right\rangle$ are the eigenstates of the operator $\hat{P}_{a}^{\hat{n}_{k}}$ $\left(\text{with }k=1,2\right)$, respectively. And $\left\{\left|\pm\phi_{i}\right\rangle\right\}$ and $\left\{\left|\pm\varphi_{j}\right\rangle\right\}$ are two sets of complete basis of $2^{M}$-dimensional Hilbert space, respectively. At the same time, $\left|\pm\phi_{i}\right\rangle$ and $\left|\pm\varphi_{j}\right\rangle$ also represent Alice’s particles. $|\eta_{i\pm}^{(\alpha)}\rangle$ and $|\varepsilon_{j\pm}^{(\alpha)}\rangle$ are Bob’s collapsed states (unnormalized), where

$$\left|\eta_{i\pm}^{\left(\alpha\right)}\right\rangle\left\langle\eta_{i\pm}^{\left(\alpha\right)}\right|=\text{tr}_{A}\bigg{[}\left(\left|\pm\phi_{i}\right\rangle\left\langle\pm\phi_{i}\right|\otimes{\mathds{1}}\right)\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle\left\langle\psi_{AB}^{\left(\alpha\right)}\right|\bigg{]},$$ (21)

and

$$\left|\varepsilon_{j\pm}^{\left(\alpha\right)}\right\rangle\left\langle\varepsilon_{j\pm}^{\left(\alpha\right)}\right|=\text{tr}_{A}\bigg{[}\left(\left|\pm\varphi_{j}\right\rangle\left\langle\pm\varphi_{j}\right|\otimes{\mathds{1}}\right)\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle\left\langle\psi_{AB}^{\left(\alpha\right)}\right|\bigg{]},$$ (22)

where ${\mathds{1}}$ is a $2^{(N-M)}\times 2^{(N-M)}$ identity matrix. $s_{i\pm}^{(\alpha)}$ and $t_{j\pm}^{(\alpha)}$ are complex numbers satisfying

$$\sum_{i}\left(\left|s_{i+}^{\left(\alpha\right)}\right|^{2}+\left|s_{i-}^{\left(\alpha\right)}\right|^{2}\right)=1,$$ (23)

and

$$\sum_{j}\left(\left|t_{j+}^{\left(\alpha\right)}\right|^{2}+\left|t_{j-}^{\left(\alpha\right)}\right|^{2}\right)=1.$$ (24)

The two requirements can be rewritten as:

1. The pure state requirement: $|\eta_{i\pm}^{(\alpha)}\rangle$ and $|\varepsilon_{j\pm}^{(\alpha)}\rangle$ are independent of $\alpha$.

2. The measurement requirement: for the result obtained by Bob, any one of $\{|\eta_{i\pm}\rangle\}$ is different from any one of $\{|\varepsilon_{j\pm}\rangle\}$

The pure state requirement guarantees that Bob’s unnormalized conditional states are all pure states. After Alice’s measurement, Bob obtains the states

$$\tilde{\rho}_{a_{i\pm}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{i\pm}^{\left(\alpha\right)}\right|^{2}\left|\eta_{i\pm}^{\left(\alpha\right)}\right\rangle\left\langle\eta_{i\pm}^{\left(\alpha\right)}\right|,$$ (25)

and

$$\tilde{\rho}_{a_{j\pm}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{j\pm}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{j\pm}^{\left(\alpha\right)}\right\rangle\left\langle\varepsilon_{j\pm}^{\left(\alpha\right)}\right|.$$ (26)

Since $p_{\alpha}$, $|s_{i\pm}^{(\alpha)}|^{2}$ and $|t_{j\pm}^{(\alpha)}|^{2}$ are all non-negative, any terms cannot cancell. Bob’s un-normalized conditional states are pure if and only if

$$\left\{\begin{array}[c]{l}s_{i\pm}^{\left(\alpha\right)}\left|\eta_{i\pm}^{\left(\alpha\right)}\right\rangle=c_{\alpha^{\prime}\left(i\pm\right)}^{\alpha}s_{i\pm}^{(\alpha^{\prime})}\left|\eta_{i\pm}^{(\alpha^{\prime})}\right\rangle,\\[9.0pt] t_{j\pm}^{\left(\alpha\right)}\left|\varepsilon_{j\pm}^{\left(\alpha\right)}\right\rangle=d_{\alpha^{\prime}\left(j\pm\right)}^{\alpha}t_{j\pm}^{(\alpha^{\prime})}\left|\varepsilon_{j\pm}^{(\alpha^{\prime})}\right\rangle,\end{array}\right.$$ (27)

where $c_{\alpha^{\prime}\left(i\pm\right)}^{\alpha}$ and $d_{\alpha^{\prime}\left(j\pm\right)}^{\alpha}$ are arbitrary complex numbers related to $i\pm$ and $j\pm$, respectively. For convenience, we analyze in the representation with $\left\{\left|\pm\phi_{i}\right\rangle\otimes\left|\eta_{i\pm}\right\rangle\right\}$ as basis. The projection of $|\psi_{AB}^{(\alpha)}\rangle$ in Eq. (19) onto $|\pm\varphi_{j}\rangle$ is

$$\left\langle\pm\varphi_{j^{\prime}}|\psi_{AB}^{\left(\alpha\right)}\right\rangle =\sum_{i}s_{i\pm}^{\left(\alpha\right)}\left\langle\pm\varphi_{j^{\prime}}\left|\pm\phi_{i}\right.\right\rangle\left|\eta_{i\pm}^{\left(\alpha\right)}\right\rangle,$$

$$\equiv\sum_{i}V_{\left(j^{\prime}\pm\right)\left(i\pm\right)}s_{i\pm}^{\left(\alpha\right)}\left|\eta_{i\pm}^{\left(\alpha\right)}\right\rangle,$$ (28)

where $V_{(j^{\prime}\pm)(i\pm)}\equiv\langle\pm\varphi_{j^{\prime}}|\pm\phi_{i}\rangle$, $\{V_{(j^{\prime}\pm)(i\pm)}\}$ can be written as

$$V=\left(\begin{array}[c]{cccccc}V_{\left(1+\right)\left(1+\right)}&\cdots&V_{\left(1+\right)\left(2^{M-1}+\right)}&V_{\left(1+\right)\left(1-\right)}&\cdots&V_{\left(1+\right)\left(2^{M-1}-\right)}\\ \vdots&&\vdots&\vdots&&\vdots\\ V_{\left(2^{M-1}+\right)\left(1+\right)}&\cdots&V_{\left(2^{M-1}+\right)\left(2^{M-1}+\right)}&V_{\left(2^{M-1}+\right)\left(1-\right)}&\cdots&V_{\left(2^{M-1}+\right)\left(2^{M-1}-\right)}\\ V_{\left(1-\right)\left(1+\right)}&\cdots&V_{\left(1-\right)\left(2^{M-1}+\right)}&V_{\left(1-\right)\left(1-\right)}&\cdots&V_{\left(1-\right)\left(2^{M-1}-\right)}\\ \vdots&&\vdots&\vdots&&\vdots\\ V_{\left(2^{M-1}-\right)\left(1+\right)}&\cdots&V_{\left(2^{M-1}-\right)\left(2^{M-1}+\right)}&V_{\left(2^{M-1}-\right)\left(1-\right)}&\cdots&V_{\left(2^{M-1}-\right)\left(2^{M-1}-\right)}\end{array}\right).$$

Then we have

$$t_{j^{\prime}\pm}^{\left(\alpha\right)}\left|\varepsilon_{j^{\prime}\pm}^{\left(\alpha\right)}\right\rangle=\sum_{i}V_{\left(j^{\prime}\pm\right)\left(i\pm\right)}s_{i\pm}^{\left(\alpha\right)}\left|\eta_{i\pm}^{\left(\alpha\right)}\right\rangle.$$ (29)

The pure state requirement Eq. (27) can be expressed as

$$\left\{\begin{array}[c]{l}s_{i\pm}^{\left(\alpha\right)}\left|\eta_{i\pm}^{\left(\alpha\right)}\right\rangle=c_{\alpha^{\prime}\left(i\pm\right)}^{\alpha}s_{i\pm}^{(\alpha^{\prime})}\left|\eta_{i\pm}^{(\alpha^{\prime})}\right\rangle,\\ \left|\chi_{j\pm}^{\left(\alpha\right)}\right\rangle=d_{\alpha^{\prime}\left(j\pm\right)}^{\alpha}\left|\chi_{j\pm}^{\left(\alpha^{\prime}\right)}\right\rangle.\end{array}\right.$$ (30)

Where $|\chi_{j\pm}^{(\alpha)}\rangle\equiv\sum_{i}V_{(j\pm)(i\pm)}s_{i\pm}^{(\alpha)}|\eta_{i\pm}^{(\alpha)}\rangle$.

The measurement requirement suggests that if Alice chooses different measurements $\hat{P}_{a}^{\hat{n}_{1}}$ or $\hat{P}_{a^{\prime}}^{\hat{n}_{2}}$, Bob cannot get the same result. We prove that in Appendix A and Appendix B the results obtained by Bob cannot be the same in different measurements. And “the different measurements” refers to the measurements in different directions ($\hat{n}_{1}$ and $\hat{n}_{2}$). After Alice’s measurements, Bob obtains $s_{i\pm}^{(\alpha)}|\eta_{i\pm}^{(\alpha)}\rangle$ and $|\chi_{j\pm}^{(\alpha)}\rangle=\sum_{i}V_{(j\pm)(i\pm)}s_{i\pm}^{(\alpha)}|\eta_{i\pm}^{(\alpha)}\rangle$. It can be seen that $|\chi_{j\pm}^{(\alpha)}\rangle$ is obtained by superposition of $|\eta_{i\pm}^{(\alpha)}\rangle$. If Bob’s two results are required to be different, $|\chi_{j\pm}^{(\alpha)}\rangle$ contains at least two summation terms. This also requires: (1) At least two terms in the summation in $|\psi_{AB}^{(\alpha)}\rangle$ are nonzero. That is, at least two $s_{i\pm}^{(\alpha)}$ in $|\psi_{AB}^{(\alpha)}\rangle$ are nonzero. (2) The matrix $\{V_{(j\pm)(i\pm)}\}$ has at least two nonzero matrix elements in each row. That is, the two measurements ${{P}_{a}^{\hat{n}_{1}}}$ and ${{P}_{a^{\prime}}^{\hat{n}_{2}}}$ are different. (3) $|\psi_{AB}^{(\alpha)}\rangle\neq\big{[}\sum_{i}(s_{i+}^{(\alpha)}|+\phi_{i}\rangle+s_{i-}^{(\alpha)}|-\phi_{i}\rangle)\big{]}\otimes|\eta_{\ell}^{(\alpha)}\rangle$, where $|\eta_{\ell}^{(\alpha)}\rangle$ is one of $\big{\{}|\eta_{i\pm}^{(\alpha)}\rangle\big{\}}$. That is, each $|\psi_{AB}^{(\alpha)}\rangle$ is an entangled state.

Proof.

Here, we prove that for $N$-qubit state $\rho_{AB}$, the difference between quantum theory and classical theory can be expressed as “2=1”, as long as the pure state requirement and the measurement requirement are satisfied. It is well known that if $\rho_{AB}$ satisfies two requirements at the same time, Bob’s unnormalized conditional states are all pure states. And for different measurements $\hat{P}_{a}^{\hat{n}_{1}}$ and $\hat{P}_{a^{\prime}}^{\hat{n}_{2}}$, Bob ’s results are different. Without loss of generality, we assume that Bob’s $2^{M+1}$ unnormalized conditional states are different. Then for the quantum results we have

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{1+}^{\left(\alpha\right)}\right|^{2}\left|\eta_{1+}\right\rangle\left\langle\eta_{1+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{2^{M-1}+}^{\left(\alpha\right)}\right|^{2}\left|\eta_{2^{M-1}+}\right\rangle\left\langle\eta_{2^{M-1}+}\right|,\\ \tilde{\rho}_{a_{2^{M-1}-}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{2^{M-1}-}^{\left(\alpha\right)}\right|^{2}\left|\eta_{2^{M-1}-}\right\rangle\left\langle\eta_{2^{M-1}-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{1-}^{\left(\alpha\right)}\right|^{2}\left|\eta_{1-}\right\rangle\left\langle\eta_{1-}\right|,\\ \tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{1+}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{1+}\right\rangle\left\langle\varepsilon_{1+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{2^{M-1}+}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{2^{M-1}+}\right\rangle\left\langle\varepsilon_{2^{M-1}+}\right|,\\ \tilde{\rho}_{a_{2^{M-1}-}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{2^{M-1}-}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{2^{M-1}-}\right\rangle\left\langle\varepsilon_{2^{M-1}-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{1-}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{1-}\right\rangle\left\langle\varepsilon_{1-}\right|.\end{array}\right.$$ (31)

Suppose Bob’s states have a LHS description, they must satisfy Eq. (5) and Eq. (6). Then, Bob will check the following set of $2^{M+1}$ equations:

$$\tilde{\rho}_{a}^{\hat{n}_{k}}={\displaystyle\sum\limits_{\xi}}\wp\left(a|\hat{n}_{k},\xi\right)\wp_{\xi}\rho_{\xi}.$$ (32)

If these $2^{M+1}$ equations have a contradiction, that is they cannot have a common solution for the sets $\{\wp_{\xi}\rho_{\xi}\}$ and $\wp(a|\hat{n}_{k},\xi)$, then Bob is convinced that the LHS models are non-existent and that Alice can steer the state of his qubits.

In the quantum result, there are $2^{M+1}$ pure states in Eq. (31). Then in the LHS description, it is sufficient to take $\xi$ from 1 to $2^{M+1}$. It is a fact that a density matrix of pure state can only be expanded by itself. Therefore, any $\tilde{\rho}_{a}^{\hat{n}_{k}}$ in the equation (32) contains only one term. So for the LHS models results we have

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{1}}=\wp_{1}\rho_{1},\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}}^{\hat{n}_{1}}=\wp_{2^{M-1}}\rho_{2^{M-1}},\\ \tilde{\rho}_{a_{2^{M-1}-}}^{\hat{n}_{1}}=\wp_{2^{M-1}+1}\rho_{2^{M-1}+1},\\ \cdots\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}=\wp_{2^{M}}\rho_{2^{M}},\\ \tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M}+1}\rho_{2^{M}+1},\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M}+2^{M-1}}\rho_{2^{M}+2^{M-1}},\\ \tilde{\rho}_{a_{2^{M-1}-}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M}+2^{M-1}+1}\rho_{2^{M}+2^{M-1}+1},\\ \cdots\\ \tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M+1}}\rho_{2^{M+1}}.\end{array}\right.$$ (33)

Finally, we sum up terms in Eq. (33) and take the trace. The left side gives $\mathrm{tr}(\tilde{\rho}^{\hat{n}_{1}}+\tilde{\rho}^{\hat{n}_{2}})=2\mathrm{tr}(\rho_{B})=2$, the result of quantum. While the right side gives $\mathrm{tr}(\wp_{1}\rho_{1}+\cdots+\wp_{2^{M+1}}\rho_{2^{M+1}})=\mathrm{tr}(\rho_{B})=1$, the result of the classical LHS models. This leads to the contradiction “2=1”, which represents the EPR paradox in the 2-setting steering protocol. ∎

Remark 1.

For 2-qubit pure state Eq. (2), Bob’s unnormalized conditional states are always pure. So only need to verify whether it satisfies the measurement requirement. It can be seen that in the 2-setting protocol $\{\hat{z},\hat{x}\}$, Bob ’s results are $\{|0\rangle\langle 0|,|1\rangle\langle 1|\}$ and $\{|+\rangle\langle+|,|-\rangle\langle-|\}$, respectively chenjl2016 . Obviously, this satisfies the measurement requirement. According to our analysis, such state Eq. (2) can get the contradiction “2=1”.

Remark 2.

For the 4-qubit mixed state Eq. (12), it is necessary to analyze whether it satisfies both the pure state requirement and the measurement requirement. In the 2-setting protocol $\{\hat{n}_{1},\hat{n}_{2}\}=\{\hat{z}\hat{z},\hat{y}\hat{x}\}$, Bob’s eight conditional states are all pure states as shown in Eq. (16). And for two different measurements $P_{a}^{\hat{n}_{1}}$ and $P_{a^{\prime}}^{\hat{n}_{2}}$, Bob’s results are different. It is obvious that the pure state requirement and the measurement requirement are satisfied at the same time, so the specific 4-qubit mixed state Eq. (12) can also get the contradiction “2=1”.

In the EPR steering paradox, we propose a theorem for $N$-qubit quantum states which contains two requirements: the pure state requirement and the measurement requirement. If Alice and Bob share an $N$-qubit mixed state, there will be a contradiction of “2=1” only when the pure state requirement and the measurement requirement are satisfied at the same time. If they share an $N$-qubit pure state, the pure state requirement is automatically satisfied. In this situation, after Alice’s measurement, Bob must get a pure state, which only needs to meet the measurement requirement. And our results are completely consistent with the previous conclusions. This confirms the correctness of our conclusion.

III Bell-like basis measurement

Here we show a more specific example of the Bell-like basis measurement for the $N$-qubit mixed states. Let us consider Alice and Bob share an $N$-qubit entangled state $\rho_{AB}=\sum_{\alpha}p_{\alpha}|\psi_{AB}^{(\alpha)}\rangle\langle\psi_{AB}^{(\alpha)}|$, in which$|\psi_{AB}^{(\alpha)}\rangle$ may be written as

$$\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=\sum_{i}\left(s_{i+}^{\left(\alpha\right)}\left|+\phi_{i}\right\rangle\left|\eta_{i+}^{\left(\alpha\right)}\right\rangle+s_{i-}^{\left(\alpha\right)}\left|-\phi_{i}\right\rangle\left|\eta_{i-}^{\left(\alpha\right)}\right\rangle\right).$$ (34)

and $\sum_{i}(|s_{i+}^{(\alpha)}|^{2}+|s_{i-}^{(\alpha)}|^{2})=1$. Alice has $M(M<N)$ particles and Bob has $(N-M)$ particles. In particular, in the 2-setting steering protocol $\left\{\hat{n}_{1},\hat{n}_{2}\right\}$ $\left(\text{with }\hat{n}_{1}\neq\hat{n}_{2}\right)$, Alice performs the Bell-like basis measurement on her qubits. Then according to the theorem, we analyze whether this example can obtain the contradiction “2=1”, and if yes, what conditions $\rho_{AB}$ need to meet with the Bell-like basis measurement.

The Bell-like basis measurement can be written as

$$\left\{\begin{array}[c]{l}P_{a_{1+}}^{\hat{n}_{k}}=\left(\cos\beta_{k}\left|+\phi_{1}\right\rangle+\sin\beta_{k}\left|-\phi_{1}\right\rangle\right)\left(\cos\beta_{k}\left\langle+\phi_{1}\right|+\sin\beta_{k}\left\langle-\phi_{1}\right|\right),\\ \vdots\\ P_{a_{2^{M-1}+}}^{\hat{n}_{k}}=\left(\cos\beta_{k}\left|+\phi_{2^{M-1}}\right\rangle+\sin\beta_{k}\left|-\phi_{2^{M-1}}\right\rangle\right)\left(\cos\beta_{k}\left\langle+\phi_{2^{M-1}}\right|+\sin\beta_{k}\left\langle-\phi_{2^{M-1}}\right|\right),\\ P_{a_{2^{M-1}-}}^{\hat{n}_{k}}=\left(\sin\beta_{k}\left|+\phi_{2^{M-1}}\right\rangle-\cos\beta_{k}\left|-\phi_{2^{M-1}}\right\rangle\right)\left(\sin\beta_{k}\left\langle+\phi_{2^{M-1}}\right|-\cos\beta_{k}\left\langle-\phi_{2^{M-1}}\right|\right),\\ \vdots\\ P_{a_{1-}}^{\hat{n}_{k}}=\left(\sin\beta_{k}\left|+\phi_{1}\right\rangle-\cos\beta_{k}\left|-\phi_{1}\right\rangle\right)\left(\sin\beta_{k}\left\langle+\phi_{1}\right|-\cos\beta_{k}\left\langle-\phi_{1}\right|\right),\end{array}\right.$$ (35)

in which $\beta_{k}\in\left[0,2\pi\right)$. Alice performs the measurement along $\hat{n}_{k}$ directions (with $k=1,2$). And $|\pm\phi_{i}\rangle$ ( $i=1,2,\cdots,2^{M-1}$) is a set of complete basis of $2^{M}$-dimensional Hilbert space. We prove that in Appendix C. After Alice’s measurement, Bob obtains

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{i+}}^{\hat{n}_{k}}=\sum_{\alpha}p_{\alpha}\left|\omega_{i+}^{k\left(\alpha\right)}\right\rangle\left\langle\omega_{i+}^{k\left(\alpha\right)}\right|,\\[6.0pt] \tilde{\rho}_{a_{i-}}^{\hat{n}_{k}}=\sum_{\alpha}p_{\alpha}\left|\omega_{i-}^{k\left(\alpha\right)}\right\rangle\left\langle\omega_{i-}^{k\left(\alpha\right)}\right|,\end{array}\right.$$ (36)

with

$$\begin{array}[c]{c}\left|\omega_{i+}^{k\left(\alpha\right)}\right\rangle=s_{i+}^{\left(\alpha\right)}\cos\beta_{k}\left|\eta_{i+}^{\left(\alpha\right)}\right\rangle+s_{i-}^{\left(\alpha\right)}\sin\beta_{k}\left|\eta_{i-}^{\left(\alpha\right)}\right\rangle,\\[6.0pt] \left|\omega_{i-}^{k\left(\alpha\right)}\right\rangle=s_{i+}^{\left(\alpha\right)}\sin\beta_{k}\left|\eta_{i+}^{\left(\alpha\right)}\right\rangle-s_{i-}^{\left(\alpha\right)}\cos\beta_{k}\left|\eta_{i-}^{\left(\alpha\right)}\right\rangle.\end{array}$$ (37)

Firstly, in Bell-like basis measurement, the transformation matrix $\{V_{(j\pm)(i\pm)}\}$ is

$$V=\left(\begin{array}[c]{cccccc}\cos\left(\beta_{1}-\beta_{2}\right)&0&\cdots&\sin\left(\beta_{1}-\beta_{2}\right)&0&\cdots\\ 0&\cos\left(\beta_{1}-\beta_{2}\right)&\cdots&0&\sin\left(\beta_{1}-\beta_{2}\right)&\cdots\\ \vdots&\vdots&\ddots&\vdots&\vdots&\vdots\\ -\sin\left(\beta_{1}-\beta_{2}\right)&0&\cdots&\cos\left(\beta_{1}-\beta_{2}\right)&0&\cdots\\ 0&-\sin\left(\beta_{1}-\beta_{2}\right)&\cdots&0&\cos\left(\beta_{1}-\beta_{2}\right)&\cdots\\ \vdots&\vdots&\vdots&\vdots&\vdots&\vdots\end{array}\right).$$ (38)

It is obvious that there are only two nonzero matrix elements $V_{jq+}$ as well as $V_{jq-}$ (with $q=1,2,\cdots,2^{M-1}$) in each row of matrix $\{V_{(j\pm)(i\pm)}\}$, and $|\chi_{j\pm}^{(\alpha)}\rangle$ contains only two terms. Similarly, only $|\eta_{q+}^{(\alpha)}\rangle$ and $|\eta_{q-}^{(\alpha)}\rangle$ contribute to $|\psi_{AB}^{(\alpha)}\rangle$, and we can consider $|\psi_{AB}^{(\alpha)}\rangle$ as

$$\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=s_{q+}^{\left(\alpha\right)}\left|+\phi_{q}\right\rangle\left|\eta_{q+}^{\left(\alpha\right)}\right\rangle+s_{q-}^{\left(\alpha\right)}\left|-\phi_{q}\right\rangle\left|\eta_{q-}^{\left(\alpha\right)}\right\rangle.$$ (39)

The measurement requirement also requires $|\psi_{AB}^{(\alpha)}\rangle\neq\big{[}\sum_{i}(s_{i+}^{(\alpha)}|+\phi_{i}\rangle+s_{i-}^{(\alpha)}|-\phi_{i}\rangle)\big{]}\otimes|\eta_{\ell}^{(\alpha)}\rangle$, so $|\eta_{q+}^{(\alpha)}\rangle\neq|\eta_{q-}^{(\alpha)}\rangle$. In this way, the measurement requirement is satisfied. Secondly, the pure state requirement requires that $|\eta_{q\pm}^{(\alpha)}\rangle$ is independent of $\alpha$, i.e.

$$\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=s_{q+}^{\left(\alpha\right)}\left|+\phi_{q}\right\rangle\left|\eta_{q+}\right\rangle+s_{q-}^{\left(\alpha\right)}\left|-\phi_{q}\right\rangle\left|\eta_{q-}\right\rangle.$$ (40)

It is apparent that after a series of analysis, the form of $|\psi_{AB}^{(\alpha)}\rangle$ is simple and only contains two terms. There is an interesting question worthy of our further analysis, that is, whether $|\psi_{AB}^{(\alpha)}\rangle$ and $|\psi_{AB}^{(\alpha^{\prime})}\rangle$ can contain the same states? Suppose that $|\psi_{AB}^{(\alpha)}\rangle$ and $|\psi_{AB}^{(\alpha^{\prime})}\rangle$ are

$$\left\{\begin{array}[c]{l}\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=s_{p+}^{\left(\alpha\right)}\left|+\phi_{p}\right\rangle\left|\eta_{p+}\right\rangle+s_{p-}^{\left(\alpha\right)}\left|-\phi_{p}\right\rangle\left|\eta_{p-}\right\rangle,\\[6.0pt] \left|\psi_{AB}^{(\alpha^{\prime})}\right\rangle=s_{p+}^{(\alpha^{\prime})}\left|+\phi_{p}\right\rangle\left|\eta_{p+}\right\rangle+s_{p-}^{(\alpha^{\prime})}\left|-\phi_{p}\right\rangle\left|\eta_{p-}\right\rangle,\end{array}\right.$$ (41)

with $p\in\{1,2,\cdots,2^{m}\}$. The pure state requirement requires

$$\left\{\begin{array}[c]{c}s_{p+}^{(\alpha^{\prime})}=c_{\alpha^{\prime}\left(p+\right)}^{\alpha}s_{p+}^{\left(\alpha\right)},\\ s_{p-}^{(\alpha^{\prime})}=c_{\alpha^{\prime}\left(p-\right)}^{\alpha}s_{p-}^{\left(\alpha\right)},\end{array}\right.$$ (42)

and

$$\begin{array}[c]{l}V_{jp+}s_{p+}^{\left(\alpha\right)}\left|\eta_{p+}\right\rangle+V_{jp-}s_{ip-}^{\left(\alpha\right)}\left|\eta_{p-}\right\rangle\\ =d_{\alpha^{\prime}j}^{\alpha}\left(V_{jp+}s_{p+}^{(\alpha^{\prime})}\left|\eta_{p+}\right\rangle+V_{jp-}s_{p-}^{(\alpha^{\prime})}\left|\eta_{p-}\right\rangle\right).\end{array}$$ (43)

The measurement requirement requires $|\eta_{p+}^{(\alpha)}\rangle\neq|\eta_{p-}^{(\alpha)}\rangle$. According to Eq. (42) and Eq. (43), we have

$$c_{\alpha^{\prime}\left(p+\right)}^{\alpha}=c_{\alpha^{\prime}\left(p-\right)}^{\alpha}=\frac{1}{d_{\alpha^{\prime}j}^{\alpha}}\equiv c_{\alpha^{\prime}}^{\alpha}.$$ (44)

So

$$\left|\psi_{AB}^{(\alpha^{\prime})}\right\rangle=c_{\alpha^{\prime}}^{\alpha}\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle,$$ (45)

that means $|\psi_{AB}^{(\alpha)}\rangle$ and $|\psi_{AB}^{(\alpha^{\prime})}\rangle$ are the same state. In summary, $|\psi_{AB}^{(\alpha)}\rangle$ and $|\psi_{AB}^{(\alpha^{\prime})}\rangle$ cannot contain the same state.

Therefore, for arbitrary $\alpha$ and $\alpha^{\prime}$, we have

$$\left\{\begin{array}[c]{l}\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=s_{q+}^{\left(\alpha\right)}\left|+\phi_{q}\right\rangle\left|\eta_{q+}\right\rangle+s_{q-}^{\left(\alpha\right)}\left|-\phi_{q}\right\rangle\left|\eta_{q-}\right\rangle,\\ \left|\psi_{AB}^{(\alpha^{\prime})}\right\rangle=s_{q^{\prime}+}^{(\alpha^{\prime})}\left|+\phi_{q^{\prime}}\right\rangle\left|\eta_{q^{\prime}+}\right\rangle+s_{q^{\prime}-}^{(\alpha^{\prime})}\left|-\phi_{q^{\prime}}\right\rangle\left|\eta_{q^{\prime}-}\right\rangle,\end{array}\right.$$ (46)

with $q\neq q^{\prime}$ and $q,q^{\prime}=1,2,\cdots,2^{M-1}$. And after Alice’s measurement, Bob’s results are

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{q+}}^{\hat{n}_{k}}=p_{\alpha}\left|\omega_{q+}^{k\left(\alpha\right)}\right\rangle\left\langle\omega_{q+}^{k\left(\alpha\right)}\right|,\\[6.0pt] \tilde{\rho}_{a_{q-}}^{\hat{n}_{k}}=p_{\alpha}\left|\omega_{q-}^{k\left(\alpha\right)}\right\rangle\left\langle\omega_{q-}^{k\left(\alpha\right)}\right|.\end{array}\right.$$ (47)

Similarly, suppose every Bob’s state has a LHS description. Bob can check the following set of $2^{M+1}$ equations:

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{k}}={\displaystyle\sum\limits_{\xi}}\wp\left(a_{1+}|\hat{n}_{k},\xi\right)\wp_{\xi}\rho_{\xi},\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}}^{\hat{n}_{k}}={\displaystyle\sum\limits_{\xi}}\wp\left(a_{2^{M-1}+}|\hat{n}_{k},\xi\right)\wp_{\xi}\rho_{\xi},\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{k}}={\displaystyle\sum\limits_{\xi}}\wp\left(a_{1-}|\hat{n}_{k},\xi\right)\wp_{\xi}\rho_{\xi},\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}-}}^{\hat{n}_{k}}={\displaystyle\sum\limits_{\xi}}\wp\left(a_{2^{M-1}-}|\hat{n}_{k},\xi\right)\wp_{\xi}\rho_{\xi}.\end{array}\right.$$ (48)

According to Eq. (47) , Bob’s unnormalized conditional states are pure. A density matrix of pure state can only be expanded by itself, therefore, from Eq. (47) we have

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{k}}=\wp_{1}\rho_{1},\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}}^{\hat{n}_{k}}=\wp_{2^{M-1}}\rho_{2^{M-1}},\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{k}}=\wp_{2^{M-1}+1}\rho_{2^{M-1}+1},\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}-}}^{\hat{n}_{k}}=\wp_{2^{M}}\rho_{2^{M}}.\end{array}\right.k=1,2$$ (49)

By summing them up and taking the trace, the left side gives $\mathrm{tr}(\tilde{\rho}^{\hat{n}_{1}}+\tilde{\rho}^{\hat{n}_{2}})=2\mathrm{tr}(\rho_{B})=2$. But the right side gives $\mathrm{tr}(\wp_{1}\rho_{1}+\cdots+\wp_{2^{M+1}}\rho_{2^{M+1}})=\mathrm{tr}(\rho_{B})=1$. This leads to the contradiction “2=1”

In summary, we discuss the steering paradox with the Bell-like basis measurement. And it shows that for Bell-like basis measurement, when $\rho_{AB}$ satisfied both the pure state requirement and the measurement requirement, we can obtain the contradiction “2=1”. In this case, $|\psi_{AB}^{(\alpha)}\rangle$ and $|\psi_{AB}^{(\alpha^{\prime})}\rangle$ cannot contain the same state, and only contain two items. It is evident that the maximum value range of $\alpha$ is from $1$ to $2^{M-1}$ ($M$ is the number of particles of Alice), that is, the maximum rank of $\rho_{AB}$ is $2^{M-1}$ for the Bell-like basis measurement.

IV Nonexistence of contradiction “2 = 1” for 2-qubit mixed states

In this section, we apply the results to the 2-qubit mixed states. We get the following corollary.

Corollary 1.

If Alice and Bob share a 2-qubit mixed state, there is no contradiction “2 = 1”.

From the above analysis, we know that for the Bell-like basis measurement, the maximum rank of $\rho_{AB}$ is $2^{M-1}$. Suppose Alice and Bob share a 2-qubit entangled state $\rho_{AB}$ and each of them has a qubit, then $M=1$. If Alice performs the Bell-like basis measurement, the maximum rank of $\rho_{AB}$ is $1$. That is, for 2-qubit entangled states with the Bell-like basis measurement, only the $2$-qubit prue entangled states have the contradiction “2=1”. So is there a contradiction “2=1” for 2-qubit mixed states? Next, we prove that there is no contradiction “2 = 1” for 2-qubit mixed states

Proof.

Suppose Alice and Bob share a 2-qubit mixed state $\rho_{AB}=\sum_{\alpha}p_{\alpha}|\psi_{AB}^{(\alpha)}\rangle\langle\psi_{AB}^{(\alpha)}|$, in which

$$\left|\psi_{AB}^{\left(\alpha\right)}\right\rangle=s_{1}^{\left(\alpha\right)}\left|\phi_{1}\right\rangle\left|\eta_{1}\right\rangle+s_{2}^{\left(\alpha\right)}\left|\phi_{2}\right\rangle\left|\eta_{2}\right\rangle.$$ (50)

Here we take $|\eta_{1}^{(\alpha)}\rangle$ $=$ $|\eta_{1}\rangle$ and $|\eta_{2}^{(\alpha)}\rangle$ $=$ $|\eta_{2}\rangle$ to satisfy the pure state requirement. And the measurement requirement requests $s_{1}^{(\alpha)},s_{2}^{(\alpha)}\neq 0$, and $|\eta_{1}\rangle\neq|\eta_{2}\rangle$. Alice takes one particle and the other belongs to Bob. Similaily, in the 2-setting steering protocol $\{\hat{n}_{1},\hat{n}_{2}\}$, suppose that Alice’ s projective measurements are $\{|\phi_{i}\rangle\langle\phi_{i}|\}$ and $\{|\varphi_{j}\rangle\langle\varphi_{j}|\}$, where $|\langle\phi_{i}|\varphi_{j}\rangle|<1$ and $i,j=1,2$. Then Bob’s unnormalized conditional states are

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{i}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{i}^{\left(\alpha\right)}\right|^{2}\left|\eta_{i}\right\rangle\left\langle\eta_{i}\right|,\\[6.0pt] \tilde{\rho}_{a_{j}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|\chi_{j}^{\left(\alpha\right)}\right\rangle\left\langle\chi_{j}^{\left(\alpha\right)}\right|,\end{array}\right.$$ (51)

in which $|\chi_{j}^{\left(\alpha\right)}\rangle=\sum_{i}V_{ji}s_{i}^{\left(\alpha\right)}|\eta_{i}^{\left(\alpha\right)}\rangle$, $V_{ji}=\left\langle\varphi_{j}|\phi_{i}\right\rangle$. In order to get the contradiction, for any $\alpha$ and $\alpha^{\prime}$, the conditional state of Bob needs to satisfy

$$\left\{\begin{array}[c]{l}\left|\chi_{j}^{\left(\alpha\right)}\right\rangle=d_{\alpha^{\prime}j}^{\alpha}\left|\chi_{j}^{(\alpha^{\prime})}\right\rangle,\\[6.0pt] s_{i}^{\left(\alpha\right)}=c_{\alpha^{\prime}i}^{\alpha}s_{i}^{\left(\alpha^{\prime}\right)}.\end{array}\right.$$ (52)

According to Eq. (52) , we have $c_{\alpha^{\prime}(i)}^{\alpha}=c_{\alpha^{\prime}(i)}^{\alpha}=1/{d_{\alpha^{\prime}j}^{\alpha}}$. Then we can obtain $|\psi_{AB}^{(\alpha)}\rangle$ and $|\psi_{AB}^{(\alpha^{\prime})}\rangle$ are the same state, similarly. Therefore, there is no contradiction “2=1” in 2-qubit mixed states. ∎

V Conclusion

We have presented a simple EPR steering paradox that shows the incompatibility of the local-hidden-state models with quantum theory for any $N$-qubit entangled state based on a 2-setting steering protocol. The argument is valid for any $N$-qubit entangled state, not only $N$-qubit pure entangled states, but more importantly, $N$-qubit mixed entangled states. We propose a simple theorem and prove that for any $N$-qubit state satisfying simultaneously “the pure state requirement” and “the measurement requirement”, then the contradiction “2=1” can be obtained. In the example of Bell-like basis measurement, we obtain that the maximum rank of the $N$-qubit mixed state is $2^{M-1}$ ($M$ is the number of particles of Alice), and prove that there is no contradiction “2=1” in the 2-qubit mixed state. In general, we obtain the contradiction “2=1” in a more general case. Furthermore, if one considers the EPR steering scenario in $k$-setting for arbitrary $N$-qudit entangled mixed state, then following the similar approach one can arrive at a full contradiction, i.e., “$k=1$”.

Appendix A Can Bob have the same state in the same measurement?

In order to obtain the contradiction “2=1”, we analyze whether Bob can have the same pure state in the same measurement, and whether Bob can have the same pure state in different measurements.

After Alice’s measurement, Bob obtains

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{1+}^{\left(\alpha\right)}\right|^{2}\left|\eta_{1+}\right\rangle\left\langle\eta_{1+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{m+}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{m+}^{\left(\alpha\right)}\right|^{2}\left|\eta_{m+}\right\rangle\left\langle\eta_{m+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{n+}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{n+}^{\left(\alpha\right)}\right|^{2}\left|\eta_{n+}\right\rangle\left\langle\eta_{n+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{2^{M-1}+}^{\left(\alpha\right)}\right|^{2}\left|\eta_{2^{M-1}+}\right\rangle\left\langle\eta_{2^{M-1}+}\right|,\\ \tilde{\rho}_{a_{2^{M-1}-}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{2^{M-1}-}^{\left(\alpha\right)}\right|^{2}\left|\eta_{2^{M-1}-}\right\rangle\left\langle\eta_{2^{M-1}-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{n-}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{n-}^{\left(\alpha\right)}\right|^{2}\left|\eta_{n-}\right\rangle\left\langle\eta_{n-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{m-}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{m-}^{\left(\alpha\right)}\right|^{2}\left|\eta_{m-}\right\rangle\left\langle\eta_{m-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}=\sum_{\alpha}p_{\alpha}\left|s_{1-}^{\left(\alpha\right)}\right|^{2}\left|\eta_{1-}\right\rangle\left\langle\eta_{1-}\right|,\\ \tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{1+}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{1+}\right\rangle\left\langle\varepsilon_{1+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{m^{\prime}+}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|s_{m^{\prime}+}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{m^{\prime}+}\right\rangle\left\langle\varepsilon_{m^{\prime}+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{n^{\prime}+}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|s_{n^{\prime}+}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{n^{\prime}+}\right\rangle\left\langle\varepsilon_{n^{\prime}+}\right|,\\ \cdots\\ \tilde{\rho}_{a_{2^{M-1}+}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{2^{M-1}+}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{2^{M-1}+}\right\rangle\left\langle\varepsilon_{2^{M-1}+}\right|,\\ \tilde{\rho}_{a_{2^{M-1}-}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{2^{M-1}-}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{2^{M-1}-}\right\rangle\left\langle\varepsilon_{2^{M-1}-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{n^{\prime}-}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|s_{n^{\prime}-}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{n^{\prime}-}\right\rangle\left\langle\varepsilon_{n^{\prime}-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{m^{\prime}-}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|s_{m^{\prime}-}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{m^{\prime}-}\right\rangle\left\langle\varepsilon_{m^{\prime}-}\right|,\\ \cdots\\ \tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}=\sum_{\alpha}p_{\alpha}\left|t_{1-}^{\left(\alpha\right)}\right|^{2}\left|\varepsilon_{1-}\right\rangle\left\langle\varepsilon_{1-}\right|.\end{array}\right.$$ (53)

In Eq. (53), assuming that $\left|\eta_{m+}\right\rangle=\left|\eta_{n+}\right\rangle$ , i.e., only $2^{M+1}-1$ different pure states appear in the quantum result of Eq. (53), so that it is sufficient to take $\xi$ from $1$ to $2^{M+1}-1$, namely, one can take the ensemble as

$$\left\{\wp_{\xi}\rho_{\xi}\right\}=\left\{\wp_{1}\rho_{1},\wp_{2}\rho_{2},\wp_{3}\rho_{3},\cdots,\wp_{2^{M+1}-1}\rho_{2^{M+1}-1}\right\}.$$ (54)

Then Eq. (53) can be written as

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{1}}={\displaystyle\sum\limits_{\xi=1}^{2^{M+1}-1}}\wp\left(a_{1+}|\hat{n}_{1},\xi\right)\wp_{\xi}\rho_{\xi},\\ \cdots\\ \tilde{\rho}_{a_{m+}}^{\hat{n}_{1}}={\displaystyle\sum\limits_{\xi=1}^{2^{M+1}-1}}\wp\left(a_{m+}|\hat{n}_{1},\xi\right)\wp_{\xi}\rho_{\xi},\\ \cdots\\ \tilde{\rho}_{a_{n+}}^{\hat{n}_{1}}={\displaystyle\sum\limits_{\xi=1}^{2^{M+1}-1}}\wp\left(a_{n+}|\hat{n}_{1},\xi\right)\wp_{\xi}\rho_{\xi},\\ \cdots\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}={\displaystyle\sum\limits_{\xi=1}^{2^{M+1}-1}}\wp\left(a_{1-}|\hat{n}_{1},\xi\right)\wp_{\xi}\rho_{\xi},\\ \tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}={\displaystyle\sum\limits_{\xi=1}^{2^{M+1}-1}}\wp\left(a_{1+}^{\prime}|\hat{n}_{2},\xi\right)\wp_{\xi}\rho_{\xi},\\ \cdots\\ \tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}={\displaystyle\sum\limits_{\xi=1}^{2^{M+1}-1}}\wp\left(a_{1-}^{\prime}|\hat{n}_{2},\xi\right)\wp_{\xi}\rho_{\xi}.\end{array}\right.$$ (55)

Since the $2^{M+1}$ states on the left-hand side of Eq. (53) are all pure states, a pure state cannot be obtained by a convex sum of other different states. Therefore,

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{1}}=\wp\left(a_{1+}|\hat{n}_{1},1\right)\wp_{1}\rho_{1},\\ \wp\left(a_{1+}|\hat{n}_{1},2\right)=\wp\left(a_{1+}|\hat{n}_{1},3\right)=\cdots=\wp\left(a_{1+}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (56)

Similarly, one has

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{2+}}^{\hat{n}_{1}}=\wp\left(a_{2+}|\hat{n}_{1},2\right)\wp_{2}\rho_{2},\\ \wp\left(a_{2+}|\hat{n}_{1},1\right)=\wp\left(a_{2+}|\hat{n}_{1},3\right)=\cdots=\wp\left(a_{2+}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (57)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{m+}}^{\hat{n}_{1}}=\wp\left(a_{m+}|\hat{n}_{1},m\right)\wp_{m}\rho_{m},\\ \wp\left(a_{m+}|\hat{n}_{1},1\right)=\wp\left(a_{m+}|\hat{n}_{1},2\right)=\cdots=\wp\left(a_{m+}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (58)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{n+}}^{\hat{n}_{1}}=\wp\left(a_{n+}|\hat{n}_{1},m\right)\wp_{m}\rho_{m},\\ \wp\left(a_{n+}|\hat{n}_{1},1\right)=\wp\left(a_{n+}|\hat{n}_{1},2\right)=\cdots=\wp\left(a_{n+}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (59)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}=\wp\left(a_{1-}|\hat{n}_{1},2^{M}-1\right)\wp_{2^{M}-1}\rho_{2^{M}-1},\\ \wp\left(a_{2^{M}}|\hat{n}_{1},1\right)=\wp\left(a_{2^{M}}|\hat{n}_{1},2\right)=\cdots=\wp\left(a_{2^{M}}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (60)

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}=\wp\left(a_{1+}^{\prime}|\hat{n}_{2},1\right)\wp_{2^{M}}\rho_{2^{M}},\\ \wp\left(a_{1+}^{\prime}|\hat{n}_{2},2\right)=\wp\left(a_{1+}^{\prime}|\hat{n}_{2},3\right)=\cdots=\wp\left(a_{1+}^{\prime}|\hat{n}_{2},2^{M+1}-1\right)=0.\end{array}\right.$$ (61)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}=\wp\left(a_{1-}^{\prime}|\hat{n}_{2},2^{M+1}-1\right)\wp_{2^{M+1}-1}\rho_{2^{M+1}-1},\\ \wp\left(a_{1-}^{\prime}|\hat{n}_{2},1\right)=\wp\left(a_{1-}^{\prime}|\hat{n}_{2},2\right)=\cdots=\wp\left(a_{1-}^{\prime}|\hat{n}_{2},2^{M+1}-2\right)=0.\end{array}\right.$$ (62)

Because ${\sum\limits_{a}}\wp(a|\hat{n},\xi)=1$, one has

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{1}}=\wp_{1}\rho_{1},\\ \cdots\\ \tilde{\rho}_{a_{m+}}^{\hat{n}_{1}}=\wp\left(a_{m+}|\hat{n}_{1},m\right)\wp_{m}\rho_{m},\\ \cdots\\ \tilde{\rho}_{a_{n+}}^{\hat{n}_{1}}=\wp\left(a_{n+}|\hat{n}_{1},n\right)\wp_{n}\rho_{n},\\ \cdots\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}=\wp_{2^{M}-1}\rho_{2^{M}-1},\\ \tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M}}\rho_{2^{M}},\\ \cdots\\ \tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M+1}-1}\rho_{2^{M+1}-1}.\end{array}\right.$$ (63)

where $\wp\left(a_{m+}|\hat{n}_{1},m\right)+\wp\left(a_{n+}|\hat{n}_{1},n\right)=1$, The sum on the left-hand side of Eq. (63) is $2\rho_{B}$, and the sum on the right-hand side is $\rho_{B}$. By summing terms in Eq. (63) and taking trace, we arrive at the contradiction “2=1”. So in the same measurement, if Bob gets the same pure state, we can get the contradiction “2=1”.

Appendix B Can Bob have the same state in different measurements?

In Eq. (53), assuming that $\left|\eta_{m+}\right\rangle=\left|\varepsilon_{m^{\prime}+}\right\rangle$ , i.e., only $2^{M+1}-1$ different pure states appear in the quantum result of Eq. (53). Similarly, it is sufficient to take $\xi$ from $1$ to $2^{M+1}-1$, one can take the ensemble as Eq. (54). Then one has Eq. (55). The $2^{M+1}$ states on the left-hand side of Eq. (53) are all pure states. The same reasoning can be used to obtain

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1+}}^{\hat{n}_{1}}=\wp\left(a_{1+}|\hat{n}_{1},1\right)\wp_{1}\rho_{1},\\ \wp\left(a_{1+}|\hat{n}_{1},2\right)=\wp\left(a_{1+}|\hat{n}_{1},3\right)=\cdots=\wp\left(a_{1+}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (64)

Similarly, one has

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{2+}}^{\hat{n}_{1}}=\wp\left(a_{2+}|\hat{n}_{1},2\right)\wp_{2}\rho_{2},\\ \wp\left(a_{2+}|\hat{n}_{1},1\right)=\wp\left(a_{2+}|\hat{n}_{1},3\right)=\cdots=\wp\left(a_{2+}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (65)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{m+}}^{\hat{n}_{1}}=\wp\left(a_{m+}|\hat{n}_{1},m\right)\wp_{m}\rho_{m},\\ \wp\left(a_{m+}|\hat{n}_{1},1\right)=\wp\left(a_{m+}|\hat{n}_{1},2\right)=\cdots=\wp\left(a_{m+}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (66)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}=\wp\left(a_{1-}|\hat{n}_{1},2^{M}\right)\wp_{2^{M}}\rho_{2^{M}},\\ \wp\left(a_{2^{M}}|\hat{n}_{1},1\right)=\wp\left(a_{2^{M}}|\hat{n}_{1},2\right)=\cdots=\wp\left(a_{2^{M}}|\hat{n}_{1},2^{M+1}-1\right)=0.\end{array}\right.$$ (67)

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}=\wp\left(a_{1+}^{\prime}|\hat{n}_{2},1\right)\wp_{2^{M}+1}\rho_{2^{M}+1},\\ \wp\left(a_{1+}^{\prime}|\hat{n}_{2},2\right)=\wp\left(a_{1+}^{\prime}|\hat{n}_{2},3\right)=\cdots=\wp\left(a_{1+}^{\prime}|\hat{n}_{2},2^{M+1}-1\right)=0.\end{array}\right.$$ (68)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{{}_{m^{\prime}+}}^{\prime}}^{\hat{n}_{2}}=\wp\left(a_{{}_{m^{\prime}+}}^{\prime}|\hat{n}_{2},m^{\prime}\right)\wp_{2^{M}+m^{\prime}}\rho_{2^{M}+m^{\prime}},\\ \wp\left(a_{1}^{\prime}|\hat{n}_{2},1\right)=\wp\left(a_{1}^{\prime}|\hat{n}_{2},2\right)=\cdots=\wp\left(a_{1}^{\prime}|\hat{n}_{2},2^{M+1}-1\right)=0.\end{array}\right.$$ (69)

$$\cdots$$

$$\left.\begin{array}[c]{l}\tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}=\wp\left(a_{1-}^{\prime}|\hat{n}_{2},2^{M+1}-1\right)\wp_{2^{M+1}-1}\rho_{2^{M+1}-1},\\ \wp\left(a_{1-}^{\prime}|\hat{n}_{2},1\right)=\wp\left(a_{1-}^{\prime}|\hat{n}_{2},2\right)=\cdots=\wp\left(a_{1-}^{\prime}|\hat{n}_{2},2^{M+1}-2\right)=0.\end{array}\right.$$ (70)

Because ${\sum\limits_{a}}\wp(a|\hat{n},\xi)=1$, one has

$$\left\{\begin{array}[c]{l}\tilde{\rho}_{a_{1}}^{\hat{n}_{1}}=\wp_{1}\rho_{1},\\ \cdots\\ \tilde{\rho}_{a_{m+}}^{\hat{n}_{1}}=\wp_{m}\rho_{m},\\ \cdots\\ \tilde{\rho}_{a_{1-}}^{\hat{n}_{1}}=\wp_{2^{M}}\rho_{2^{M}},\\ \tilde{\rho}_{a_{1+}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M}+1}\rho_{2^{M}+1},\\ \cdots\\ \tilde{\rho}_{a_{{}_{m^{\prime}+}}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M}+m^{\prime}}\rho_{2^{M}+m^{\prime}},\\ \cdots\\ \tilde{\rho}_{a_{1-}^{\prime}}^{\hat{n}_{2}}=\wp_{2^{M+1}-1}\rho_{2^{M+1}-1}.\end{array}\right.$$ (71)

Here $\wp(a_{m+}|\hat{n}_{1},m)=1$, $\wp(a_{{}_{m^{\prime}+}}^{\prime}|\hat{n}_{2},m^{\prime})=1$, no more $\wp(a_{m+}|\hat{n}_{1},m)+\wp(a_{{}_{m^{\prime}+}}^{\prime}|\hat{n}_{2},m^{\prime})=1$. The sum on the left-hand side of Eq. (71) is $2\rho_{B}$, and the sum on the right-hand side is $\rho_{B}+\wp_{i}\rho_{i}$. By summing terms in Eq. (71) and taking trace, we no longer get the contradiction “2=1”. So in the different measurements, if Bob gets the same pure state, we can’t get the contradiction “2=1”.

Appendix C $\left|\pm\phi_{i}\right\rangle$ is a set of complete basis of Hilbert space

Here we show that $|\pm\phi_{i}\rangle$ ( $i=1,2,\cdots,2^{M-1}$) which in the Bell-like basis measurement is a set of complete basis of $2^{M}$-dimensional Hilbert space. For any $k$( $k=1,2$), we have $\sum_{i=1}^{2^{M-1}}(P_{a_{i+}}^{\hat{n}_{k}}+P_{a_{i-}}^{\hat{n}_{k}})=1$. That can be expanded as

$${\displaystyle\sum\limits_{i=1}^{2^{M-1}}}\left(P_{a_{i+}}^{\hat{n}_{k}}+P_{a_{i-}}^{\hat{n}_{k}}\right)$$

$$= {\displaystyle\sum\limits_{i=1}^{2^{M-1}}}\left(\cos^{2}\beta_{k}\left|+\phi_{i}\right\rangle\left\langle+\phi_{i}\right|+\cos\beta_{k}\sin\beta_{k}\left|+\phi_{i}\right\rangle\left\langle-\phi_{i}\right|+\sin\beta_{k}\cos\beta_{k}\left|-\phi_{i}\right\rangle\left\langle+\phi_{i}\right|+\sin^{2}\beta_{k}\left|-\phi_{i}\right\rangle\left\langle-\phi_{i}\right|\right.$$

$$\left.+\sin^{2}\beta_{k}\left|+\phi_{i}\right\rangle\left\langle+\phi_{i}\right|-\sin\beta_{k}\cos\beta_{k}\left|+\phi_{i}\right\rangle\left\langle-\phi_{i}\right|-\cos\beta_{k}\sin\beta_{k}\left|-\phi_{i}\right\rangle\left\langle+\phi_{i}\right|+\cos^{2}\beta_{k}\left|-\phi_{i}\right\rangle\left\langle-\phi_{i}\right|\right)$$

$$= {\displaystyle\sum\limits_{i=1}^{2^{M-1}}}\left(\left|+\phi_{i}\right\rangle\left\langle+\phi_{i}\right|+\left|-\phi_{i}\right\rangle\left\langle-\phi_{i}\right|\right)$$

$$= {\mathds{1}}$$ (72)

It is obvious that $|\pm\phi_{i}\rangle$ is a set of complete basis of $2^{M}$-dimensional Hilbert space.