arXiv:1803.04843v4 [quant-ph] 18 Apr 2021 ytmi nymaigu ntecneto measurement of system. context another the by in quantum meaningful a in- only of is physical reality system The through measurement. established In- during is systems, teraction quantum correlation system. among how relation quantum and the a quan- describes describe of it a not stead, does properties for theory state independent Quantum absolute the to 11]. no [10, relative system is described tum there be should system, system another quantum suggests a further themselves. that (RQM) systems Mechanics quantum basic Quantum the more Relational of is properties systems the in- quantum than correlation Sim- two induced that between concluded system. decoherence is formation the of it of [7–9], theory rest environment the by the developing meaningful of only in state is ilarly, given subsystem a a to of relative state rela- quantum asserts The a [4–6] mechanics that recognized. quantum of further formulation was state state tive system quan- quantum reference of a a development of on dependency recent the more interpretations, In tum measuring the on [1–3]. depends apparatus system quantum de- a the days of that early scription emphasized the had In Bohr mechanics, measurement. quantum a of in system observations quantum experimental a the of explain to theory physical ∗ [email protected] h dao Q stogtpooig tessentially It provoking. thought is RQM of idea The a as developed originally was mechanics Quantum Keywords: nec numbers: a is PACS is theory. It observers relativity different the system. from with the mechanics results on sync measurement are performed of they measurement tion if any system mechanic from quantum quantum outcomes a in re observer of reso local the descriptions the a consistent and in to abandoned confirmed relative is is be exchange This must necessary. observer becomes super measurement c of system composite assumption a For measu who the observer evo location. quantum any super time from a results a its surement of assumes description describe mechanics However, to quantum Traditional possible system. is reference concep it a significant system, in quantum results traditio given the formulation duri to exchanged the relati equivalent is importantly, the mathematically information how is mutual that studying how mulation By and independe elements. transformed the fundamental is of most instead the systems, as quantum among properties unu esrmn n unu prto hoyi develo is theory operation quantum and measurement Quantum .INTRODUCTION I. eainlQatmMcais unu esrmn,Entan Measurement, Quantum Mechanics, Quantum Relational unu ehnc rmRltoa Properties Relational from Mechanics Quantum 36.a 03.65.-w 03.65.Ta, atI:MaueetadEPR and Measurement II: Part a ig,C 22,USA 92121, CA Diego, San Dtd pi 0 2021) 20, April (Dated: ina .Yang M. Jianhao r bevrdpnet hsi nedtemi thesis main themselves the indeed properties is relational this This in the is observer-dependent. say It are we system. that observing differ- sense the system between and properties observed system relational of the observed set In different describe a using can system. ently reference she the case and such results system measurement) observed (or, complete between have interaction not the It does of observers. observer information another multiple that be possible can is prop- there relational quantified, the and are selected erties is Even system [13–30]. reference the QRFs a if switching con- called ensure when to descriptions is how sistent particularly which re- QRF, extensive on system, are activities There quantum search (QRF). a frame reference also quantum sys- is reference itself The tem systems. envi- differently reference the described different in be to can relative system system another quantum A or ronment. apparatus setup, an measurement be refer- can a It The in selected. system. arbitrarily reference is system properties a ence to relational from relative the theory are Second, themselves mechanics quantum properties? system? a relational reference reconstruct the to and properties How system relational observed the rel- the quantify of between to quantum aspect start- how of this include a with formulation ativity as associated the considered Questions constructing be mechanics. for rela- should point properties The system independent ing quantum the a system. of of one instead than of properties basic tional properties more are independent proper- systems relational the quantum the quan- two that sys- between a claims another ties theory to since RQM relative First, the described tem, be relativity. should system of tum aspects two implies a unu esrmn hoy More theory. measurement quantum nal ulcneune.W hwta o a for that show We consequences. tual ∗ nit pc-iesprtdsubsystems, separated space-like onsists .Dffrn oa bevr a achieve can observers local Different s. rnzdo h nomto regarding information the on hronized sayse hncmiigquantum combining when step essary uinwtotepiil aln out calling explicitly without lution uino P aao.Information paradox. EPR of lution tpoete faqatmsystem, quantum a of properties nt gmaueet edrv h for- the derive we measurement, ng a ntnaeul nwtemea- the know instantaneously can eetms eepiil relative. explicitly be must rement nlpoaiiyapiuematrix amplitude probability onal e eeb aigterelational the taking by here ped ainlfruaino quantum of formulation lational ugse httesynchroniza- the that suggested lmn,EPR glement, 2 of Ref. [10]. There are several fundamental questions for a given quantum system, description of its time evo- associated with the second aspect of relativity: Is such lution can be implicitly relative, while description of a description compatible with traditional quantum theory quantum operation must be explicitly relative. Informa- which appears as observer independent (at least the non- tion exchange is relative to a local observer in quantum relativistic quantum mechanics)? If a quantum system mechanics. The assumption of Super Observer should should be described as observer-dependent, how is the be abandoned, so as the notion of observer independent objectivity of a physical reality preserved? In what con- description of physical reality. Different local observers dition RQM can provide results that are different from can achieve consistent descriptions of a quantum system traditional quantum mechanics? if they are synchronized on the outcomes from any mea- Recently we proposed a formulation to address the surement performed on the system. Traditional quantum first aspect of RQM [32]. In this formulation, the re- mechanics was originally developed to explain observa- lational properties between the two quantum systems tion results from microscopic system that is much smaller are the most fundamental elements to formulate quan- than the measuring apparatus. In those situations, RQM tum mechanics. In addition, a novel framework of cal- and traditional quantum mechanics are practically equiv- culating the probability of an outcome when measuring alent. However, for a composite system that is spatially a quantum system is proposed by modeling the bidirec- much larger than a typical apparatus, the necessity of tional probe-response interaction process in a measure- RQM formulation become obvious. This is clearly illus- ment. The fundamental relational property is defined trated in the analysis of EPR paradox [34]. The paradox as relational probability amplitude. The probability of is seemingly inevitable in traditional quantum mechan- a measurement outcome is proportional to the summa- ics but can be resolved by removing the assumption of tion of probability amplitude product from all alternative the Super Observer who knows measurement results in- measurement configurations. The properties of quan- stantaneously from local observer from any location. It tum systems, such as superposition and entanglement, is suggested that the synchronization of measurement re- are manifested through the rules of counting the alterna- sults from different observers is a necessary step when tives. As a result, the framework gives mathematically combining quantum mechanics with Relativity Theory. equivalent formulation to Born’s rule. Wave function is The works presented here is inspired by the main idea found to be summation of relational probability ampli- of the original RQM theory [10]. However, there are sig- tudes, and Schr¨odinger Equation is derived when there is nificant advancements compared with the original RQM no entanglement in the relational probability amplitude theory. They are summarized in Section VII. matrix. Although the relational probability amplitude is The paper is organized as following. Firstly we briefly the most basic properties, there are mathematical tools review the relational formulation of quantum mechanics such as wave function and reduced density matrix that in Section II. In Section III we present the measurement describe the observed system without explicitly called out theory based on the relational formulation and in Sec- the measuring system. Thus, the formulation in Ref. [32] tion IV the formulation is extended to general quantum is mathematically compatible to the traditional quantum operation. It turns out that Schr¨odinger Equation, for- mechanics. mulations for selective and non-selective measurement, can all be derived from the general quantum operation. This paper has two goals. The first goal is to extend Section V analyzes the criteria on whether a quantum the formulation presented in Ref. [32] to the quantum process must be described by calling out the observer ex- measurement theory. One of the key concept in this for- plicitly. The result is applied to resolve the EPR paradox mulation is the entanglement measure of the relational in Section VI. Lastly, the conceptual consequences, the probability amplitude matrix. The entanglement mea- potential applications of this formulation, and the con- sure quantifies the difference between time evolution and clusion remarks are presented in Section VII. measurement. When there is change in the entanglement entropy, we expect to derive the quantum measurement theory, which is missing in Ref. [32]. This paper intends to complete the formulation for quantum measurement II. RELATIONAL FORMULATION OF QUANTUM MECHANICS and quantum operation in the context of RQM. The re- formulation is mathematically equivalent to the tradi- tional quantum measurement theory and open quantum A. Terminologies system theory. The second goal of this paper is to in- vestigate how the measurement theory is perceived from A Quantum System, denoted by symbol S, is an object different observers. Here we assume these different ob- under study and follows the quantum mechanics postu- servers use the same reference frame. How the measure- lates [32]. An Apparatus, denoted as A, can refer to the ment theory is transformed when switching QRF is not measuring devices, the environment that S is interact- in the scope of this paper. Therefore, this investigation ing with, or the system from which S is created. All only partially addresses the second aspect of RQM. But systems are quantum systems, including any apparatus. as shown later, it already results in several important Depending on the selection of observer, the boundary conceptual implications. For instance, We assert that between a system and an apparatus can change. For ex- 3 ample, in a measurement setup, the measuring system is an apparatus A, the measured system is S. However, the composite system S + A as a whole can be consid- ered a single system, relative to another apparatus A′. In an ideal measurement to measure an observable of S, the apparatus is designed in such a way that at the end of the measurement, the pointer state of A has a distin- guishable, one to one correlation with the eigenvalue of the observable of S. The definition of Observer is associated with an appa- ratus. An observer, denoted as , is an intelligence who can operate and read the pointerO variable of the appara- tus. Whether or not this observer is a quantum system FIG. 1. Schematic illustration of the entities for the termi- is irrelevant in our formulation. However, there is a re- nologies. The overlapping of the measured system S and ap- striction that is imposed by the principle of locality. An paratus A represents there is interaction in a measurement. observer is defined to be physically local to the appara- The relational properties between S and A must be described tus he associates with. This prevents the situation that relative a QRF. There can be multiple QRFs. Selecting a O can instantaneously read the pointer variable of the different QRF, can have a different description of the rela- tional properties in a quantum measurement event. apparatusO that is space-like separated from . Receiv- ing the information from A at a speed fasterO than the speed of light is prohibited. An observer cannot be asso- ciated with two or more apparatuses in the same time if the apparatus. Both systems encode information each these apparatuses are space-like separated. other, allowing an observer to infer measurement results In the traditional quantum measurement theory pro- of S by reading pointer variable of A. The key insight posed by von Neumann [35], both the quantum system learned here is that quantum measurement is a question- and the measuring apparatus follow the same quantum and-answer bidirectional process. The measuring system mechanics laws. Von Neumann further distinguished two interacts (or, disturbs) the measured system. The inter- separated measurement stages, Process 1 and Process action in turn alters the state of the measuring system. 2. Mathematically, an ideal measurement process is ex- As a result, a correlation is established, allowing the mea- pressed as surement result for S to be inferred from the pointer vari- able of A. Ψ = ψ a0 = c s a0 | iSA | Si| i X i| ii| i A Quantum Reference Frame (QRF) is a quantum sys- i (1) tem where all the descriptions of the relational properties ci si ai sn an between S and A is referred to. There can be multi- −→ X | i| i −→ | i| i i ple QRFs. How the descriptions are transformed when Initially, both S and A are in a product state described by switching QRFs is not in the scope of this study. But we expect the theories developed in Ref. [29–31] can be Ψ SA. In Process 2, referring to the first arrow in Eq.(1), the| i quantum system S and the apparatus A interact. applicable here. In this paper, we only consider the de- However, as a combined system they are isolated from in- scription relative to one QRF, denoted as F . It is also teraction with any other system. Therefore, the dynam- possible to choose A as the reference frame. In that case, ics of the total system is determined by the Schr¨odinger F and A are the same quantum system in a measure- Equation. Process 2 establishes a one to one correla- ment [29, 31]. Fig. 1 shows a schematic illustration of tion between the eigenstate of observable of S and the the entities in the relational formulation. pointer state of A. After Process 2, there are many pos- A Quantum State of S describes the complete informa- tion an observer can know about S. A quantum state sible outcomes to choose from. In the next step which is O called Process 1, referring to the second arrow in Eq.(1), encodes the relational information of S relative to A or one of these possible outcomes (labeled with eigenvalue other systems that S previously interacted with [11]. The n) emerges out from the rest1. An observer knows the information encoded in the quantum state is the complete outcome of the measurement via the pointer variable of knowledge an observer can say about S, as it determines

1 Traditional quantum mechanics does not provide a theoretical the quantum system. Therefore, it is an epistemic, rather than description of Process 1. In the Copenhagen Interpretation, this ontological, variable. In this view, the collapse of wave function is considered as the “collapse’ of the wave function into an eigen- is just an update of the observer’s description on the condition state of the measured observable. The nature of this wave func- of knowing the measurement outcome. For example, Quantum tion collapse has been debated over many decades. Recent inter- Bayesian theory [36, 37] formulates how Bayesian theorem can pretations of quantum theory advocate that the wave function be utilized to describe such process. The relational argument of simply encodes the information that an observe can describe on the wave function “collapse” is presented in Section III. 4

~ the possible outcomes of next measurement. As we will But eiSp/ itself is a probabilistic quantity. Although SA explain later, the state for a quantum system is not a Rij is a probability amplitude, not a probability real fundamental concept. Instead, it is a derivative concept number, we assume it follows certain rules in the clas- from the relational properties. sical probability theory, such as multiplication rule, and sum of alternatives in the intermediate steps. The set of kets si , representing distinct measure- B. Basic Formulation ment events for S{|, cani} be considered as eigenbasis of Hilbert space with dimension N, and s is an eigen- HS | ii This section briefly describes the framework of rela- vector. Since each measurement outcome is distinguish- tional formulation of quantum mechanics [32]. the frame- able, si sj = δij . Similarly, the set of kets aj is eigenbasish | i of Hilbert space with dimension{| Ni} work is based on a detailed analysis of the interaction HA process during measurement of a quantum system. First, for the apparatus system A. The bidirectional process a ⇋ s is called a potential measurement configura- from experimental observations, a measurement of a vari- | j i | ii able on a quantum system yields multiple possible out- tion. A potential measurement configuration comprises comes randomly. Each potential outcome is obtained possible eigen-vectors of S and A that involve in the mea- with a certain probability. We call each measurement surement event, and the relational weight quantities. It can be represented by Γ : s , a , RSA,QAS . with a distinct outcome a quantum event. Denote these ij {| ii | j i ij ji } alternatives events with a set of kets s for S, where To derive the properties of S based on the relational {| ii} (i =0,...,N 1), and a set of kets aj for A, where R, we examine how the probability of measuring S with (j =0,...,M− 1). A potential measurement{| i} outcome is a particular outcome of variable q is calculated. It turns − represented by a pair of kets ( si , aj ). Second, a physi- out such probability is proportional to the sum of weights cal measurement is a bidirectional| i process,| i the measuring from all applicable measurement configurations, where system and the measured system interact and modify the the weight is defined as the product of two relational state of each other. The probability of finding a poten- probability amplitudes corresponding to the applicable tial measurement outcome represented by a pair of kets measurement configuration. Identifying the applicable ( s , a ), p , should be calculated by modeling such measurement configuration manifests the properties of a | ii | j i ij bidirectional process. This implies pij can be expressed quantum system. For instance, before measurement is as product of two numbers, actually performed, we do not know that which event will occur to the quantum system since it is completely p QASRSA. (2) probabilistic. It is legitimate to generalize the potential ij ∝ ji ij measurement configuration as aj si ak . In AS SA other words, the measurement configuration| i → | i → in the| i joint Qji and Rij are not necessarily real non-negative num- Hilbert space starts from a , but can end at a , or ber since each number alone only models a unidirectional | j i | j i process which is not a complete measurement process. any other event, ak . Indeed, the most general form | i On the other hand, pij is a real non-negative number of measurement configuration in a bipartite system can since it models an actual measurement process. To sat- be aj sm sn ak . Correspondingly, we | i → | i → | i → | i isfy such requirement, we further assume generalize Eq.(2) by introducing a quantity for such con- figuration, AS SA Qji = (Rij )∗. (3) ASSA AS SA SA SA Wjmnk = Qjm Rnk = (Rmj )∗Rnk . (5) AS SA Written in a different format, Qji = (R )ji† . This The second step utilizes Eq.(3). This quantity is inter- AS SA means Q = (R )†. Eq.(2) then becomes preted as a weight associated with the potential measure- ment configuration aj sm sn ak . Suppose SA 2 | i → | i → | i → | i pij = R /Ω (4) we do not perform actual measurement and inference in- | ij | formation is not available, the probability of finding S AS where Ω is a real number normalization factor. Qji and in a future measurement outcome can be calculated by SA ASSA Rij are called relational probability amplitudes. Given summing Wjmnk from all applicable alternatives of mea- the relation in Eq.(3), we will not distinguish the nota- surement configurations. tion R versus Q, and only use R. The relational matrix With this framework, the remaining task to calculate RSA gives the complete description of S. It provides a the probability is to correctly count the applicable alter- framework to derive the probability of future measure- natives of measurement configuration. This task depends ment outcome. on the expected measurement outcome. For instance, SA Rij can be explicitly calculated using the path Inte- suppose the expected outcome of an ideal measurement gral formulation [33]. In the context of path integral, is event si , i.e., measuring variable q gives eigenvalue SA iSp/~ | i Rij is defined as the sum of quantity e , where Sp is qi. The probability of event si occurs, pi, is propor- ASSA| i the action of the composite system S + A along a path. tional to the summation of Wjmnk from all the possible Physical interaction between S and A may cause change configurations related to si . Mathematically, we select ASSA | i of Sp, which is the phase of the probability amplitude. all Wjmnk with m = n = i, sum over index j and k, and 5 obtain the probability pi. C. Entanglement Measure

M SA SA SA 2 The description of S using the reduced density ma- pi (Rij )∗Rik = Rij . (6) X X trix ρS is valid regardless there is entanglement between ∝ =0 | | j,k j S and A. To determine whether there is entanglement between S and A, a parameter to characterize the entan- This leads to the definition of wave function ϕi = Rij , Pj glement measure should be introduced. There are many so that p = ϕ 2. The quantum state can be de- i | i| forms of entanglement measure [38, 39], the simplest one scribed either by the relational matrix R, or by a set is the von Neumann entropy. Denote the eigenvalues of of variables ϕi . The vector state of S relative to A, A { } T the reduced density matrix ρS as λi ,i =0,...,N, the is ψ S = (ϕ0, ϕ1,...,ϕN ) where superscript T is the von Neumann entropy is defined as{ } transposition| i symbol. More specifically,

H(ρS)= λilnλi. (13) ψ A = ϕ s where ϕ = R . (7) − X | iS X i| ii i X ij i i j A change in H(ρS) implies there is change of entangle- The justification for the above definition is that the prob- ment between S and A. Unless explicitly pointed out, ability of finding S in eigenvector si in future measure- we only consider the situation that S is described by a | i ment can be calculated from it by defining a projection single relational matrix R. In this case, the entanglement ˆ operator Pi = si si . Noted that si are orthogonal measure E = H(ρ ). Since ρ = RR , the entanglement | ih | {| i} S S † eigenbasis, the probability is rewritten as: measure is sometimes expressed as H(R). Theorem 1 in Appendix A provides a simple criteria to determine p = ψ Pˆ ψ = ϕ 2 (8) i h | i| i | i| whether H(R) = 0 based on the decomposition of Rij . H(ρS) enables us to distinguish different quantum dy- Eqs.(6) and (7) are introduced on the condition that 2 namics. Given a quantum system S and its referencing there is no entanglement between quantum system S apparatus A, there are two types of the dynamics be- and A. If there is entanglement between them, the sum- tween them. In the first type of dynamics, there is no mation in Eq.(6) over-counts the applicable alternatives physical interaction and no change in the entanglement of measurement configurations and should be modified measure between S and A. S is not necessarily isolated accordingly. A more generic approach to describe the in the sense that it can still be entangled with A, but the quantum state of S is the reduced density matrix formu- entanglement measure remains unchanged. This type of lation, which is defined as dynamics is defined as time evolution. In the second type of dynamics, there is a physical interaction and correla- ρS = RR† (9) tion information exchange between S and A, i.e., the von Neumann entropy H(ρS) changes. This type of dynamics The probability pi is calculated using the projection op- is defined as quantum operation. Quantum measurement erator Pˆ = s s i | iih i| is a special type of quantum operation with a particular outcome. Whether the entanglement measure changes 2 pi = T rS(PˆiρˆS)= Rij . (10) distinguishes a dynamic as either a time evolution or a X | | j quantum operation. Particularly, when S is in an isolated state, its dynam- The effect of a quantum operation on the relational ics is governed by the Schr¨odinger Equation [32]. One of probability amplitude matrix can be expressed through the purposes of this paper is to provide the formulation an operator. Defined an operator Mˆ in Hilbert space of quantum operations when the entanglement measure ˆ S as Mij = si M sk , The new relational probability between S and A changes. amplitudeH matrixh | is| obtainedi by

(RSA ) = M (RSA ) , or III. QUANTUM MEASUREMENT new ij X ik init kj k (11)

Rnew = MRinit. A. Expectation of A Measurement Theory

Consequently, the reduced density becomes, The entanglement measure defined in Section II C characterizes the quantum correlation between the mea- ρnew = Rnew(Rnew)† = MρinitM †. (12) sured system S and the apparatus system A. The cor- relation enables the inference of measurement outcome. A change in entanglement measure implies change in the quantum correlation, consequently, change in the capa- 2 See Section II C for the definition of entanglement. bility of inference. The capability of inference can be 6 described by the mutual information variable, which is a fixed QRF F . How the measurement formulation is defined as [40] transformed when switching QRFs is not in the scope of this paper but studies elsewhere [31]. We start the I(S, A)= H(ρS )+ H(ρA) H(ρSA). (14) formulation with a simpler case that the S +A composite − system is initially in a product state. where H(ρ) is the von Neumann entropy of a density ma- trix3. Mutual information is a quantity that measures the amount of information about S through knowing infor- B. Product Initial State mation about A. For a pure bipartite state, H(ρSA)=0 and H(ρS)= H(ρA), thus I(S, A)=2H(ρS), only differs In Ref. [32], it is shown that when the composite sys- from the Von Neumann entropy of the reduced density tem S + A is described by a relational probability am- matrix of S by a factor of 2. Thus, in this condition, it plitude matrix R and assuming S + A is in an isolated is equivalent to state that quantum operation is a pro- environment, it is mathematically equivalent to describe cess that alters the mutual information between S and the composite system with a wave function, A. The term information exchange used in the following text strictly refers the changes of mutual information. Ψ = R s a . (15) Although the cause of information exchange is the | i X ij | ii| j i ij physical interaction, the measurement theory in this pa- per does not aim to explain the detailed physical process Suppose S and A initially are unentangled, they can be on how A records a particular outcome. Instead, the described as a product state, Ψ0 SA = Rij si aj | i Pij | i| i measurement theory just describes how the mutual in- where Rij = cidj . This implies that Ψ0 SA can be writ- formation is transferred from one system to another. In | i ten as ψ0 S φ0 A, where ψ0 S = i ci si and φ0 A = the context of RQM, the goal is to describe how the re- | i | i | i P | i | i j dj aj . S + A as a whole follows the Schr¨odinger lational probability amplitude matrix R is transformed Equation.P | i Since there is interaction between S and A, during measurement, and how mutual information is ex- the overall unitary operator cannot be decomposed to changed in the process. UˆSA = UˆS UˆA. Instead, UˆSA can be decomposed such Suppose the measurement is performed using appa- that6 it gives⊗ the following map ratus A and the initial correlation matrix is R0. Al- though measurement dynamics involves information ex- Ψ1 SA = UˆSA Ψ0 SA change between S and A, the composite system S + A is | i | i = Uˆ ψ0 φ0 isolated, and can be described as a unitary process. This SA| iS| iA (16) is Process 2 in the von Neumann measurement theory. = Mˆ ψ0 a X m| iS | mi The result is that the relational matrix R0 is mapped to m R′, denoted as R0 R′. R′ is then used by the intrinsic → observer to calculate the probability of a particular where the set of operators Mˆ m satisfies the completeness OI ˆ ˆ measurement outcome. Here the intrinsic observer is the condition m MmMm† = I. Appendix A shows that such one who reads the pointer variables of apparatus A, while a decompositionP always exists as long as the initial state any other observer who does not access the apparatus is is a product state. Substitute ψ0 = i ci si to Eq.(16) called external observer. As pointed out in Ref. [10], if an | i P | i external observer only knows there is a measurement Ψ1 SA = ( (Mˆ m)ikck) si am (17) OE | i X X | i| i process occurred, but does not know the measurement im k outcome, his description of the measurement process is This gives the new relational matrix R’ with element limited to R0 R′. On the other hand, the intrin- sic observer, → who reads the pointer variable of A, OI R′ = (Mˆ ) c knows the measurement outcome after the measurement im X m ik k k finishes. This additional information on the exact mea- (18) surement outcome allows to infer the final quantum = (Mˆm)ik(R0)kj = (MmR0)ij . OI X X state of S. It results in another map R′ R′′. This is jk j Process 1 in the von Neumann measurement→ theory. In summary, a measurement theory should describe Rim′ cannot be decomposed into a format of cidj , which how the relational matrix R and the mutual information implies H(R′) = 0 according to Theorem 1, and we can- 6 are changed during the measurement process. In this not define a wave function for S. R′ has now encoded the paper we also assume the description is with respect to correlation between S and A and can be used to predict the probability of a possible measurement outcome. At the end of the measurement, who operates and reads the outcome of his apparatusOAI knows the measurement 3 However, there is speculation that quantum mutual information outcome as A ends up in a distinguishable pointer state should be defined as I(S,A) = H(ρS ) − H(ρSA), see remark in am . This allows to infer exactly the resulting state Ref. [41] |of Si. Since there isOI no additional interaction between S 7 and A, the process can be modeled as a local operator The last statement of the above paragraph needs IS P A where P = a a . According to Theo- more qualification. At the beginning of a measurement ⊗ m m | mih m| rem 2 in Appendix C, the relational matrix is updated to H(R) = 0. At the end of the measurement H(Rm′′ )=0as S A T A T Rm′′ = I R′(Pm) = R′(Pm ) . Substituting R′ obtained well. The entanglement measure appears to be the same earlier, one has at the beginning and at the end of the measurement. However, during the measurement process, H(R) does (R ) = R (P T ) m′′ ij X in′ m nj not stay as a constant. This can be seen from Eq.(16). n ˆ The correlation matrix Rim′ = (Mm)ikck. It is not 4 Pk = (Mˆ ) c δ δ difficult to calculate H(R′) = pmln(pm). From X X n ik k nm jm (19) − Pm n k Eq.(14) we can analyze the change of mutual information between S and A. Initially S and A share no mutual in- = ( (Mˆ ) c )(δ ) X m ik k jm formation. In the initial phase of measurement, S and k A interact and become entangled. Information from S

The last step shows that (Rm′′ )ij can be written as ci′ dj′ is encoded in A. The mutual information increases to 2 pmln(pm). This allows an observer to infer prob- with c′ = (Mˆ m)ikck and d′ = δjm. From Theorem 1, m i Pk j ability− P of measurement outcome through A, but without H(R′′ ) = 0. Therefore, we can use Eq.(7) to calculate m knowing the exact measurement outcome. At the later the wave function of S corresponding to am , | i phase of an ideal projection measurement, A becomes m disentangled with S again and converges into a particu- ϕ = (R′′ ) i X m ij j lar pointer state am with a probability pm, this allows to infer exactly| whichi state S is in. When the mea- = (Mˆ ) c δ OI X m ik k X jm (20) surement ends, S and A share no mutual information k j again. During the measurement, the mutual information = (Mˆ ) c is changed as 0 2 m pmln(pm) 0. X m ik k →− P → k The increase of mutual information in the first arrow is described as a unitary process of the composite system, Recall the initial state of S is ψ0 S = i ci si , the re- and the decrease of mutual information in the second ar- | i mP | i sulting state vector for S, ψm = ϕi si , can be writ- row is described by a projection operator. The update | i Pi | i ten as ψm = Mˆ m ψ0 S without normalization. Apply- of the relational matrix from R′ to R′′ was perceived as ing the| normalizationi | i factor, and omitting the subscript “wave function collapse” in the Copenhagen Interpreta- referring to S, one finally gets tion. However, in this paper this update is not associated with a physical reality change. Instead, it is interpreted ˆ Mm ψ0 as change of description of the relational matrix due to ψm = | i (21) | i ˆ ˆ the fact that knows the exact measurement outcome. ψ0 Mm† Mm ψ0 OI qh | | i An external observer does not know the measure- OE The normalization factor is the probability of finding A in ment outcome and therefore still describes S with R′. We see that even though both and describe S the pointer state am after Process 2, i.e., the probability OI OI of measurement with| i outcome m. This can be verified by through a relational matrix, the relational matrix itself combining Eq.(D2) in Appendix C and Eq.(16), is relative, as pointed out in the introduction section. can obtain the measurement outcome through com- S A OE pm = Ψ1 I Pm Ψ1 munication with . But this means there is a physical h | ⊗ | i (22) OI ˆ ˆ interaction between A′ and A. An interaction between = ψ0 Mm† Mm ψ0 h | | i A′ and A disqualifies to describe the composite sys- OE Due to the correlation in Eq.(16), the probability of find- tem S + A as a unitary time evolution. Thus Process 1 cannot be described as a unitary process by either ing A in am is exactly the probability of inferring S | i or . In other words, Process 1 cannot be describedOI in the resulting state ψm . If repeats the same ex- OE periment many times,| he shalli O findI that the outcome m by the Schr¨odinger Equation. One of the preconditions for applying Schr¨odinger Equation is that there should occurs with a frequency of pm, even though the outcome of a particular measurement is random. have no information exchange between the observed sys- Eqs.(21) and (22) typically appear in textbooks as a tem and the reference apparatus. But for an observer to postulate for quantum measurement [38] [40]. In deriving know the exact measurement outcome, such information these results, a mysterious ancillary system is introduced. exchange is unavoidable. The property of the ancillary system is traced out at the end to obtain Eq.(21) and (22). As shown in this section, the ancillary system is nothing but the apparatus A. Its property can be traced out because the initial state is a 4 From Eq.(16), the reduced density operator of S isρ ˆ = ˆ ˆ product state, and at the end of the measurement, S and TrA(|Ψ1ihΨ1|) = Pm Mm|ψ0ihψ0|Mm† = Pm pm|ψmihψm|. A are still in a product state. Thus H(R′)= −Tr(ˆρln(ˆρ)) = − Pm pmln(pm). 8

As mentioned earlier, measurement theory is to de- Then A is projected to a particular state am . Similar to velop a physical model that describes how mutual infor- the approach in deriving Eq.(19), the relational| i matrix S A T A mation is exchanged during measurement. The detailed is updated to Rm′′ = I R′(Pm) where Pm = φm φm , physical process of interaction is not explained here. For we get | ih | example, after the S and A become entangled, one must assume there is no further interaction between S and A (R ) = R (P T ) m′′ ij X in′ m nj in order to model the process with the local projection n S A operator I Pm. It may be just an approximation. ˆ ⊗ = si an USA Ψ0 aj φm φm an Ref. [42] provides tremendous amount of physical de- Xh |h | | ih | ih | i n (25) tails to describe this process. The measurement process = φ a a s Uˆ Ψ0 a φ goes through several sub-processes such as registration, Xh m| nih n|h i| SA| ih j | mi truncation, decoherence, and the emergence of a unique n outcome that is interpreted using quantum statistics me- = si φm UˆSA Ψ0 aj φm chanics [42]. It is of great interest to find out the physical h |h | | ih | i details on the measurement process, but the primary in- where the property an an = I is applied in the Pn | ih | terest of the measurement theory developed here is how third step. Since (Rm′′ )ij can be written as product of the relational matrix R and the mutual information are two terms with index i and j separated, according to changed during the measurement process. Theorem 1, H(R) = 0. We can use Eq.(7) to calculate the wave function of S associated with outcome a | mi C. Entangled Initial State m ϕ = (R′′ ) i X m ij j When S and A are initially entangled, A already has = s φ Uˆ Ψ0 a φ (26) some level of correlation with S. In a sense that A has al- h i|h m| SA| i Xh j | mi ready measured S since the information of A can be used j to infer information of S. One may ask what the goal of = dm si φm UˆSA Ψ0 subsequent measurement is in this case. In the situation h |h | | i that S and A are initially in product state, an opera- where dm = j aj φm is a normalization constant. The tion involving interaction between S and A increases the probability ofP findingh | measurementi outcome associated mutual information, thus allowing A to infer information with am is given by Eq.(D2) in Appendix C, of S. Similarly, in the case when S and A are initially | i S A entangled, the goal of the measurement can be further p′ = Ψ1 I P Ψ1 m h | ⊗ m| i increasing the mutual information. After more mutual 2 = s φ Uˆ Ψ0 information is encoded in A, a subsequent projection op- X |h i|h m| | i| (27) eration can be applied so that A evolves to a unique dis- i ˆ ˆ tinguishable pointer state. Since the mutual information = Ψ0 U † φm φm U Ψ0 . h | | ih | | i is defined as I(S, A) = H(ρS )+ H(ρA) H(ρSA), the maximum mutual information for a pure− bipartite state The resulting state vector of S before normalization is is Imax =2lnN where N is the rank of matrix R. We can define the amount of unmeasured mutual information as ψ = ϕm s | mi X i | ii i Iu(S, A)=2lnN I(S, A). (23) − = d s s φ Uˆ Ψ0 (28) m X | iih i|h m| SA| i Thus, the goal of measurement is to minimize Iu(S, A). i Alternatively, the goal of measurement can be set as = dm φm UˆSA Ψ0 to alter the probability distribution pm such that the h | | i probability to find S in a particular{ state} is adjusted as Normalization requires that dm = 1/ pm′ . In order desired, or such that the expectation value of an observ- to simplify Eq.(28), we rewrite the initialp entangled bi- able of S matches a desired value. We will discuss both partite state using the Schmidt decomposition Ψ0 = cases in this section. | i (US VA) i λi s˜i a˜i where US VA is a local uni- Denote the initial entangled state for S+A as Ψ0 = ⊗ P | i| i ⊗ | iSA tary transformation. λi is the Schmidt coefficient, which Rij si aj . The interaction between S and A is still Pij | i| i essentially is the eigenvalue of the relational matrix R. described as a unitary operation over the whole S + A This gives composite system, Ψ1 = Uˆ Ψ0 . The relational | iSA SA| iSA matrix R′ is 1 ψm = φm UˆSA(US VA) s˜i a˜i | i p Xh | ⊗ | i| i R′ = s a Uˆ Ψ0 m′ i ij h i|h j | SA| i p (29) (24) 1 = Rkl si aj UˆSA sk al . = Mˆ mi s˜i X h |h | | i| i p X | i kl p m′ i 9 where Mˆ mi = λi φm UˆSA(US VA) a˜i . Note that Mˆ mi ΛSA = UˆS UˆA and there is no further projection opera- depends on the initialh | state itself,⊗ Eq.(29)| i is not a simple tion. 6 S and⊗A are entangled at the end of the operation. form. If Ψ0 is a product state, λ0 = 1 and λi = 0for i> A more general global linear map on a bipartite system | i ˆ can be decomposed to Λ = α Bˆ Cˆ where Bˆ 0, Eq.(29) is reduced to Eq.(21). Given that m Pm = I, SA k k k k k P ˆ P ⊗ it is easy to verify the completeness property of Mˆ mi is local operator to S and Ck is local operator to A [38]. ΛSA is a general operation in the sense that Bˆk or Cˆk 2 Mˆ † Mˆ = δ λ I are not necessarily project operators, and the resulting X mi mj ij i S m S + A can be in a product state or an entangled state. It (30) ˆ ˆ is convenient to re-express the relational matrix by intro- Mmi† Mmj = IS . X X duce a linear operator Rˆ = Rij si aj . According to m ij Pij | ih | Theorem 2, the operation of ΛSA on S + A transforms the initial relational operator Rˆ0 to Since ψm ψm = 1, from Eq.(29) one gets pm′ = h ˆ | ˆ i ij s˜i Mmi† Mmj s˜j . It follows from Eq.(30) that ˆ ˆ ˆ ˆ ˆT P h | |2 i R =ΛSA(R0)= αkBkR0Ck (33) p′ = λ = 1. From the expression for p′ , X m m i i m k Pthe mutualP information after the unitary operation can be calculated as I′(S, A) = 2 m pm′ ln(pm′ ). On the The reduced density operator for S after the general other hand, the initial mutual− P information I(S, A) = quantum operation is 2 2 2 λi ln( λi ). If the goal of measurement is to i T − P | | | | ρˆ = RˆRˆ = α α Bˆ Rˆ0(Cˆ†Cˆ ) Rˆ0Bˆ† (34) increase the mutual information, one wishes to find a S † X k l∗ k l k l unitary operator Uˆ such that I′(S, A) > I(S, A), that is, kl

2 2 Suppose the initial composite state of S + A is Ψ0 = pm′ ln(pm′ ) < λi ln( λi ). (31) | i X X | | | | λi s˜i a˜i , the relational operator Rˆ0 can be expressed m i i P ˆ | i| i as R0 = i λi s˜i a˜i . Substitute this into Eq.(34), On the other hand, if the goal of measurement is not P | ih | necessarily to increase the mutual information, but to ρˆS = RˆRˆ† increase the probability that S is in a state inferred by ˆ ˆ ˆ T ˆ = λiλj αkαl∗Bk s˜i a˜i (C†Ck) a˜j s˜j B† A being in the pointer state am . The initial probability X | ih | l | ih | l | i 2 ijkl (35) before measurement operation is pm = Rim . After Pi | | measurement operation, we want p′ >pm. This means = (λ λ α α∗ a˜ Cˆ†Cˆ a˜ )Bˆ s˜ s˜ Bˆ†. m X i j k l h j | l k| ii k| iih j | l the goal of measurement is to find a unitary operator ijkl UˆSA such that If the local operator Cˆ is a unit operator, ΛSA = 2 s˜i Mˆ † Mˆ mj s˜j > Rim . (32) αkBˆk IA = ΛS IA. This means ΛSA only op- Xh | mi | i X | | Pk ⊗ ⊗ ij i erates on S and has no impact on A. Eq.(35) becomes

Neither Eq.(31) nor Eq.(32) is simple to solve. It is not ρˆS =ΛSρˆ0ΛS† (36) clear that for a given initial correlation matrix R, a uni- 2 tary operator Uˆ that satisfies either Eq.(31) or Eq.(32) whereρ ˆ0 = λ s˜i s˜i is the initial density operator SA Pi i | ih | always exists. This is an open topic for future research. for S. However, if ΛSA =ΛS IA, Eq.(36) does not hold 6 ⊗ The quantum measurement theory developed in the in general. RQM context is equivalent to the Open Quantum System Eq.(35) can be derived through the partial trace theory, if we replace the environment system in OQS with approach as well. The initial density operator the reference apparatus system in this formulation. The of the composite system is ρSA = Ψ0 Ψ0 = | ih | details of the equivalency is left in Appendix E. λiλj s˜i a˜i s˜j a˜j . After applying the general op- Pij | i| ih |h | eration ΛSA, the reduced density operator of S is

IV. GENERAL QUANTUM OPERATION ρˆS = T rA(ΛSAρSAΛSA† )

= T r ( λ λ α α∗Bˆ s˜ s˜ Bˆ† Cˆ a˜ a˜ Cˆ†) A X i j k l k| iih j | l ⊗ k| iih j | l In Section III we only consider the selective measure- ijkl ment. At the end of the selective measurement opera- = λ λ α α∗Bˆ s˜ s˜ Bˆ† T r (Cˆ a˜ a˜ Cˆ†) tions, the apparatus A is in a definite state, and S and A X i j k l k| iih j | l { A k| iih j | l } are in a product composite state. There is other type of ijkl quantum operation where at the end of the operation, S = λ λ α α∗( a˜ Cˆ†Cˆ a˜ )Bˆ s˜ s˜ Bˆ† X i j k l h j | l k| ii k| iih j | l and A are in an entangled state and there are still multi- ijkl ple possible outcomes. This is the non-selective measure- (37) ment [43]. For instance, the composite system S + A can go through the interaction characterized by an operator which is the same as Eq.(35). 10

An application of Eq.(35) is briefly described as follow- Thus, ϕi and ϕi′ are different only by an unimportant ing. In Section III, we have been assuming the measuring constant. The description of S relative to A′ is equivalent apparatus is A. The initial interaction between S and A to the description relative to A. In this case, there is no is described as a unitary operation on S + A. This is negative consequence5 to describe S without calling out equivalent to the case that the general map ΛSA is a uni- A. tary operator and can be decomposed according to (16). If S is entangled with another system A but the S + A However, the measurement of S can be performed using composite system is unentangled with any other system, another apparatus A′. In this case, A′ shall interact with based on the same reasoning in the previous paragraph, either S or the S+A composite system. The general map the S + A composite system can be described without ΛSA is not a unitary operator anymore. Instead, it can calling out the reference system. The state vector of the be considered as a quantum operation decomposed from composite system is [32] a unitary operator for the S + A + A′ composite system. If A′ interacts with both S and A, the resulting reduced Ψ = ϕ m = ϕ s a | i X m| i X ij | i j i density operator of S is given by Eq.(35). If the appa- m ij ratus A′ only interacts with S and has no impact on A, (38) ˆ = Rij si aj . C is a unit operator, and the resulting reduced density X | i| i operator of S is given by Eq.(36). ij Eq.(33) is the most general form of equation describing S itself is described by the reduced density matrix, different types of dynamics between S and A, depending on how the map Λ is decomposed. If Λ = Uˆ Uˆ , SA SA S A ′ ⊗ ρˆS = T rA Ψ Ψ = ( RikRi∗′k) si si it results in the Schr¨odinger Equation [32]. If ΛSA is a | ih | X′ X | ih | unitary operator but decomposed according to Eq.(16), ii k (39) it describes process 2 of the von-Neumann measurement = (RR†)ii′ si si′ . S A X′ | ih | process. If ΛSA = I Pm where Pm = am am , it de- ii scribes the process 1 of⊗ the measurement| process.ih | Lastly, Given Ψ can be described without calling out the refer- the most general decomposition of ΛSA gives Eq.(33). | i One logical conclusion is that Schr¨odinger Equation can- ence system and the reduced density matrix ρS is derived from Ψ , it is logical to deduce that ρ can be described not describe all these quantum dynamics, particularly, | i S cannot described the process 1 in the measurement pro- without calling out the reference system either. cess, as discussed in the previous section. Since time evolution is defined as a quantum process that there is no change of entanglement measure be- tween S and any other system, the argument presented V. EXPLICIT V.S. IMPLICIT RELATIVITY above holds true for any given moment during time evolu- tion. Supposed the time evolution Hamiltonian operator is known to any observer, there is no need to call out In the relational formulation of quantum mechanics, explicitly the reference system in the description. even though a quantum system S should be described Quantum measurement process, on the other hand, is relative to a reference system A, there are mathematical different. The measurement process comprises two sub- tools that provide equivalent descriptions without explic- processes. In process 2, the composite system S + A can itly calling out the reference system A. When S and A be described as a unitary process. There is information are unentangled, S can be described by a wave function exchange between S and A, but S+A as a whole does not defined in Eq.(7). When S and A are entangled, S is exchange information with other system. The intrinsic described by a reduced density matrix that traces out observer, , who is associated with A, describes Process the information of A. On the other hand, the measure- 2 accordingOI to Eq.(18); An external observer, , who is ment theory developed here indicates the observer must OE associated with A′, describes the same process according be called out explicitly when describing a measurement to Eq.(16). On the condition that both and have process. We wish to obtain a cohesive conclusion on when the same information of the total HamiltonianOI operator,OE the reference system must be explicitly called out. they can have the equivalent descriptions on S. As for If S is unentangled with any other quantum system, it Process 1, it is modeled as a projection operator and is can be described with a wave function defined as Eq.(7). not a unitary process. , who is associated with A and Supposed there are two different observers with their own reads the outcome fromOIA, gains additional information pieces of apparatus A and A′, and the relational matrices compared with other observers , who do not know the are R and R′, respectively. The wave function for S is OE ϕi = Rij relative to A, and ϕ′ = R′ relative Pj i Pj ij to A′. However, since there is no entanglement between S and A, or between S and A′, according to Theorem 5 Note that if Rij = ci′ dj′ , ϕ′ = d′ci′ 6= ϕ. We do not prove 1, Rij can be decomposed as Rij = cidj . Therefore, Rij = cidj′ here but only show it is a possible decomposition. The ϕi = ci j dj = dci, where d = j dj is just a constant. proof is not necessary because our goal is to show it is possible P P to describe S without calling out A. If R′ is decomposed as Rij = cid′ , ϕ′ = ci d′ = d′ci. ij j i Pj j 11 exact the measurement outcome. This is not a unitary interacts with each other and the reduced entropy of S1 process6 relative to either or . We conclude that a or S2 changes, S1+ S2 as a whole undergoes time evo- measurement process, comprisingOI OE both Processes 1 and lution and the process is implicitly relative. However, if 2, must be described by calling out the reference system either S1 or S2 interacts with another system outside the explicitly. A quantum process that must be described by composite system, the process is explicitly relative. This calling out the reference system is defined to be explicitly is at the heart of the EPR paradox. relative. On the other hand, a quantum process that can be described without calling out the reference system is implicitly relative. We can summary our conclusions with VI. EPR the following statement: A. Hidden Assumptions For a given system and the initial condition, the time evolution process is implicitly rela- tive, while the measurement process is explic- In traditional quantum mechanics, since the reference itly relative. system is not typically called out in traditional quan- tum mechanics, there are assumptions on the reference By definition, in a measurement process the entropy of apparatus that are not obvious. Two of such hidden as- the observed system S, H(R), is changed. A change of sumptions are: entropy of S implies there is information exchange be- tween S and another system. Thus, the above statement 1. An unentangled reference apparatus always exists can be restated as regardless the composition of the observed system S. For instance, S can be as large as the Universe. Given a system and the initial condition, if a quantum process induces information ex- 2. Suppose S comprises multiple subsystems and change from S, the process is explicitly rel- these subsystems are space-like separated. When ative. A measures a subsystem of S, the observer knows the measurement result instantaneously regardless The statement of “given a system and the initial con- where the observer locates. dition” is important. In the Process 2 of a quantum measurement process, S interacts with A during a mea- Let us call such an observer who knows the measurement surement. There is entropy change for S, thus the process result instantaneously as a Super Observer . Because is explicitly relative, i.e., relative to . If we change the always exists and knows the changes of theOS relational boundary of the system to S + A,O weI need another ap- matrixOS R instantaneously, one can choose the apparatus paratus A′. This means the observer is changed to . associated with as an absolute reference. A quantum can derive equivalent description of S comparedO toE state can then beOS described as an absolute state. The OthatE relative to , as shown in Eqs.(16) and (18). This assumption that there exists a Super Observer enables is on the conditionOI that knows the total Hamiltonian the notion of absolute state for a quantum system 7. In operator of S + A. However,OE and may not always most of physical processes where S is an isolated system share the same information, asOE seen inOI Process 1. and the locations of its subsystems are sufficiently close, Suppose system S comprises two subsystems S1 and an observer-independent state will not lead to paradox. S2, and the two subsystems are space-like separated. If Mathematically it is more convenient and elegant to de- S1+ S2 do not interact with other system, the entropy scribe a quantum state as observer-independent. How- of S1+ S2 is unchanged. Even in the case S1 and S2 ever, when a quantum system comprises two entangled subsystems and the two subsystems are remotely sepa- rated, the view of can lead to the paradox described in the EPR paperO [34].S Ref. [11] had already provided a 6 One may argue that if we include the observer herself into the thorough analysis of the EPR paradox in the RQM con- composite system S + A + O, the entire composite system can text. This analysis presented in this paper is in general be treated as an isolated system and described as going through consistent with the argument in Ref. [11], however, we a unitary process. However the inclusion of O into the described system means there is yet another apparatus is involved that bring more insights on the role of a hidden Super Ob- can measure the S + A + O composite system, thus it means a server in the EPR argument, and explore the implication change of observer by definition. Furthermore, such approach of resolution from information exchange perspective. still cannot explain why a single outcome is selected at the end of measurement. In fact, the decoherence theory follows such reasoning by considering O as environment of the apparatus. But the decoherence theory does not explain why at the end of a measurement a single outcome is singled out from all possible 7 The role of a privileged observer is also proposed in Ref [10] in outcomes after decoherence takes place. The Quantum Bayesian the effort to resolve the classical and quantum world separation Theory models Process 1 as a probability update after additional issue in the Copenhagen Interpretation. This privileged system data is collected. Obviously, the Bayesian probability update is is similar to the super observer (or, super-apparatus) contains not a unitary process. collection of “all the macroscopic objects around us” [10]. 12

B. EPR Argument are non-commutative. According to Heisenberg Uncer- tainty Principle, β cannot simultaneously have definite The EPR argument is briefly reviewed as following. eigenvalues for σz and σx. Therefore, there are elements Assuming two systems α and β are initially in the same of physical reality of β that the quantum mechanics can- physical location and interact for a period of time. They not describe. This leads to the conclusion that quantum become entangled and then move away from each other mechanics is an incomplete theory. with a space-like separation. We will adopt Bohm’s ver- The issue here is that the definition of realism assumes sion of the EPR argument by assuming α and β are two the element of physical reality is observer-independent. It spin half particles. The quantum state of the composite assumes the measurement of Alice on α reveals a phys- system can be decomposed based on the up and down ical reality that is observer-independent, and Bob at a eigenstates along the z direction, or left and right eigen- remote location knows the same physical reality instan- states along the x direction8: taneously. But both Alice and Bob are local observers, such definition is not operational to them, unless faster 1 that light interaction is permitted. If, however, there is Ψ = ( αu βu αd βd ) | i √2 | i| i − | i| i another observer, Charles, who always know the state of (40) 1 α and β at the same time, any measurement on either α = ( αl βl αr βr ). or β is known to Charles instantaneously. With the help √ | i| i − | i| i 2 of Charles, the definition of absolute physical reality is In the context of RQM, Eq.(40) assumes there is an ob- operational. However, Charles is a super observer ac- server, Alice, with apparatus A that can measure α and cording to our definition. Such an observer is imaginary, β, and A is unentangled with the composite system. The although we unintentionally assume he always exists, and issue here is that after α and β are remotely separated, we build physical concepts with such assumption. It is there is no apparatus that can perform measurement on the assumption that there exists a Super Observer that α and β at the same time. Alice needs to perform mea- allows the definition of absolute element of physical real- surement on α or β once at a time. Alternatively, there ism. According to the criteria presented in Section V, for can be two local observers, Alice with apparatus A and the given composite system α + β, the process of Alice’s located with particle α, and Bob with apparatus B and measurement on α extracts information the composite located with particle β, to perform the measurements at system. Therefore it must be described by calling out the same time. the observer explicitly. The EPR paper then proposed a definition of realism as following, C. Resolution If, without in any way disturbing a system, we can predict with certainty the value of a phys- The original definition of an element of physical real- ical quantity, then there exists an element of ism depends on a Super Observer and is not operational. physical reality corresponding to this physical Instead, the definition should be modified as following, quantity. [34] If, without in any way disturbing a system, a Now Alice performs a measurement on α and supposed local observer can predict with certainty the the outcome is spin up. Traditional quantum mechanics value of a physical quantity, then there exists states that wave function vector Ψ collapses to αu βu , an element of physical reality corresponding thus β is deterministically in the| spini up state| afteri| Al-i to this physical quantity, relative this local ob- ice’s measurement. If instead Alice performs a measure- server. ment along x axis and finds α is in the spin left eigen- With this modified definition, let’s proceed the EPR rea- state, β is deterministically in the left eigenstate after soning to see whether it leads to the conclusion of incom- the measurement. A measurement in the location where pleteness of quantum mechanics. If Alice performs a mea- α is does not cause a state change for β that is space-like surement on α along the z direction and the outcome is separated, otherwise it violates the principle of locality spin up, the wave function after measurement is updated demanded by special relativity. Since one can predict the to α β . This just means that β is in the spin up state spin of β in both z and x directions without disturbing u u according| i| i to Alice. The element of physical reality is true it, by the above definition of realism, β can simultane- only relative to Alice. From Bob’s perspective, before he ously have elements of reality for the spin properties in knows the Alice’s measurement result, he still views the both z and x directions. Denoting these properties as composite system in the original state, no quantum event eigenvalues of operators σ and σ . However, σ and σ z x z x happened yet. In other words, Bob still predicts that fu- ture measurement on β will find it is in spin up state with fifty percentage of chance. At this point, both observers

1 1 are out of synchronization on the relational information 8 Note that |α i = (|α i + |α i) and |α i = (|α i−|α i) l √2 u d r √2 u d of the two particles, thus give different descriptions of 13 particle β. Alice’s description is “Given the condition the correlation is preserved. Such entangled quantum that α is in the spin up state, β is in spin up state with state encode not only the classical correlation, but also unit probability”, while Bob’s description is still “β is in the additional information of the composite system (see spin up state with fifty percent of chance”. Note that explanation in Appendix F). When Alice measures par- Alice’s description contains a new condition “α is in the ticle α, she effectively measures the composite system, spin up state” while Bob’s description doesn’t contain because she obtains information not only about α, but such condition. Both descriptions are valid. To verify also about the correlation between α and β. In addition, the physical description Alice obtained on particle β after the measurement induces decoherence of the α + β com- measuring particle α, Alice can travel to Bob’s location posite system. Before Alice performs the measurement, to perform a measurement, or can send the measurement it is meaningless to speculate what spin state particle β result to Bob and ask Bob to perform a measurement. is in since it is in a maximum mixed state. When Alice Suppose Alice sends the measurement outcome to Bob. measures α along z direction and obtains result of spin Bob updates the wave function accordingly to αu βu , up, she knows that in this condition, β is also in spin up same as the wave function relative to Alice. He| nowi| cani and later this is confirmed by Bob. If instead, she mea- confirm the physical reality that β is in spin up state sures α along x direction and obtains result of spin left, βu with unit probability. However, in this state, he she knows that in this new condition, β is in spin left |cannoti predict deterministically that β is in spin left or and later confirmed by Bob. But such correlation is not right, since β = 1 ( β + β ). Similarly, if Alice a causal relation. To better understand this non-causal | ui √2 | li | ri performs a measurement on α along the x direction and relation, supposed there are N identical but distinguish- the outcome is spin left, β is deterministically in the spin able copies of the entangled pairs described by Eq. (40) left state relative to Alice, but nothing happened from and each pair has a label n = 1, 2, 3,...,N. Alice mea- Bob’s perspective. If Alice sends the measurement result sures the α particles sequentially along the z direction to Bob, Bob updates the wave function accordingly to and she does not send measurement results to Bob. Bob αl βl . He now can confirm the physical reality that β independently measures the β particles along z direction is| spini| i left state β . However, in this state, he cannot with same sequence. Both of them observe their own | li predict deterministically that β is in spin up or down. measurement results for σz as randomly spin up or spin Since Alice cannot perform measurement on α along z down with fifty percent of chance for each. When later and x directions in the same time, Bob cannot confirm they meet and compare measurement results, they find β has spin values in both z and x directions simulta- the sequence of σz values are exactly the matched. They neously. The reality that β simultaneously have definite can even choose a random sequence of z or x direction values for σz and σx cannot be verified. This is consistent but both follow the exact sequence in their independent with the Heisenberg Uncertainty Principle. There is no measurements. When later they meet and compare mea- incompleteness issue for quantum mechanics. Hence the surement results, they still find their measured values are original EPR argument no longer holds with the modifi- the same sequentially. cation on the definition of physical realism.

D. Non-causal Correlation E. Implications However, there is still a puzzle here. It appears Bob’s measurement outcome on β “depends” on which direc- The EPR experiment shows that a quantum measure- tion Alice chooses to measure α. Since Alice’s measure- ment should be explicitly described as observer depen- ment does not impact the physical property of particle dent. The idea of observer-independent quantum state β, exactly what spin state β is in before Alice’s measure- should be abandoned since it assumes there exists a Su- ment? To answer this subtle question, we first note that per Observer. Assumption of a Super Observer is non- it is Alice’s new gained information on β, not the physi- operational since practically a physical observer is always cal reality of β, that depends on the axis along which the local. When we measure a microscopic object which is measurement is performed. To confirm the new-found re- much smaller than the apparatus, the apparatus can de- ality of β relative to Alice, Alice sends the measurement tect change in any part of the system under observation. result to Bob who performs a subsequent measurement. This is what quantum mechanics was originally devel- There is no faster-than-light action here. One cannot oped for. However, when one wishes to extend the quan- assume there exists an absolute reality for β. Secondly, tum description to a system that is spatially much larger it is true that Bob’s measurement outcome correlates to than a typical apparatus system in a measurement, the the Alice’s measurement result. But this is an informa- locality of the apparatus becomes important. The impli- tional correlation, not a causal relation. This correlation cation here is that information exchange between quan- is encoded in the entangled state of the composite system tum systems is local. Description of a quantum process α + β described in (40). Since the entanglement is pre- involving information exchange between quantum sys- served even when both particles are space-like separated, tems must factor in such locality principle. 14

VII. DISCUSSION AND CONCLUSION B. Predictability

A. Conceptual Consequences One critical question to ask is in what situation the RQM formulation can predict results different from tra- ditional quantum mechanics. Is it yet another quantum The relational formulation of quantum measurement mechanics interpretation only, or is there new physics results in many conceptual implications. We will high- underlined the formulation? To answer the question, we light the key conceptual consequences in this section. need to know the limitation of traditional quantum me- chanics that RQM helps to remove. Quantum mechan- Explicit relativity of measurement process. A key con- ics was initially developed as a physical theory to ex- clusion of this work is that for a given quantum system, plain results of observation of microscopic systems, such it time evolution process is implicitly relative, while a as spectrum of light emitted from hydrogen atoms. In measurement process is explicitly relative. The crite- such condition, the observed system as a whole is much rion to distinguish time evolution and measurement is smaller than the apparatus. An observer can read the whether the entanglement entropy of the observed system results at once even when the observed system consists changes. Therefore, if the entanglement entropy changes, multiple subsystems. The assumption of Super Observer the process is explicitly relative. becomes operational, even though it is conceptually in- Relativity of Information. The notion of information correct. The assumption of Super Observer practically here refers to the correlation between the observed sys- makes explicit relativity unnecessary. Thus, there is no tem and the measuring system, and is measured by the need to rely on a relational formulation of quantum me- entropy of reduced density matrix of the observed sys- chanics. Traditional quantum mechanics is adequate to tem. Change of the entropy means change of informa- describe the microscopic physical world in this condition. tion. Thus, a quantum process to extract information However, when one wishes to construct a quantum the- from a system must be described explicitly relative to ory for composite system that is spatially larger than an observer. Consequently, there is no absolute informa- the typical measuring apparatus by orders of magnitude, tion to all observers in quantum mechanics, just as there REQ formulation becomes necessary, as manifested in the is no absolute spacetime in Relativity. Even when the EPR analysis. A typical procedure to construct a quan- time evolution is described by the Schr¨odinger Equation tum description is to define the boundary of the system 9 without calling out the observer explicitly, the underlined such that it can be approximated as an isolated system . notion that the quantum theory is relative should not be Then, for a given set of information, such as an initial forgotten. quantum state and the Hamilton operator, its time evo- lution is described as a unitary process. If, however, an No Super Observer. The assumption that there ex- event occurs such that one of the subsystem starts to in- ists a Super Observer can revert explicit relativity back teract with another system outside the composite system to implicit relativity, because a Super Observer instanta- and causes information exchange, it must be described neously knows the outcome of a measurement performed as explicitly relative. A different observer who does not at any remote location. One can always describe a pro- know the event, must exchange information with observer cess referring to the Super Observer. Thus, there is no who knows the event through classical communication or need to call out observer explicitly which in turns al- additional measurement if we expect these two observers lows absolute quantum state. However, the assumption to have equivalent descriptions of the same quantum pro- of Super Observer is non-operational, because practically cess. The need of information synchronization, which is a physical apparatus is practically finite and local. Con- a result of our relational formulation, becomes a nec- sequently, the measurement event is also local. essary component for an accurate quantum description. Objectivity of a quantum state. The relational nature Global quantum operation on such a composite system is of a quantum state does not imply a quantum state is not practical. If indeed there is a need to describe results subjective. Space-like separated local observers can rec- from a global operation, extra caution must be taken care oncile the different descriptions of the same quantum to synchronize the measurement outcomes from different system through classical communication of information remote subsystems. obtained from local measurements, as shown in the anal- ysis of EPR paradox. This is significant since it gives the meaning of objectivity of a quantum state. Objectivity can be defined as the ability of different observers coming 9 Such an approximation may not be always possible if the inter- to a consensus independently [8]. This synchronization action from the environment cannot be isolated. Suppose the of latest information is operational, and it is necessary observed system S is interacting with the environmental system E. Changing the boundary to include E, we have S + E. But for consistent descriptions of the same quantum system S +E is interacting with a larger surrounding environmental sys- from different observers. Exactly what the mechanism is tem E′, and so on. One cannot describe the composite system to achieve synchronization is not specified here, but such as unitary process unless extending the boundary to the whole a mechanism must follow the quantum theory. Universe. 15

Having equivalent description of a physical law from sity matrix. Therefore, RQM and traditional QM are different observers is a basic requirement in the Relativ- compatible mathematically. We further show that for a ity Theory. How quantum measurement is described in given system, its time evolution process is implicitly rel- the context of Relativity? This is an interesting ques- ative, while a measurement process is explicitly relative. tion to investigate, given that a quantum measurement The measurement process, when applied to microscopic must be described as observer dependent. We speculate physical system that are much smaller than the appa- that the need for information synchronization in a quan- ratus, can be practically described without calling out tum measurement is a necessary step when one wishes to the reference system. In such cases, RQM and tradi- combine quantum mechanics with the Relativity Theory. tional QM are practically equivalent. This is important because it confirms that although the main idea of RQM seems radical, it does not change the practical applica- C. Comparison with the Original RQM Theory tion of quantum mechanics. These points were not clear in Ref [10]. The works presented here is inspired by the main idea of the original RQM theory [10]. However, there are sev- D. Relation to QRF Theory eral significant improvements that should be pointed out. The works of Refs. [7, 10, 12] establish the idea that relational properties are more basic, and a quantum sys- As discussed in the introduction section, the RQM tem must be described relative to another quantum sys- principle consists two aspects. First, we need to refor- tem. However, they do not provide a clear formulation mulate quantum mechanics relative to a QRF which can on how a quantum system should be described relative to be in a superposition quantum state, and show how quan- another system and what the basic relational properties tum theory is transformed when switching QRFs. In this are. On the other hand, our formulation gives a clear aspect, we accept the basic quantum theory as it is, in- quantification of the relational property, which is the re- cluding Schr¨odinger equation, Born’s rule, von Neunman lational probability amplitude. The introduction of the measurement theory, but add the QRF into the formula- relational probability amplitude is based on a detailed tions and derive the transformation theory when switch- analysis of measurement process. It enables us to de- ing QRFs [29–31]. Second, we go deeper to reformulate velop a framework to calculate probability during quan- the basic theory itself from relational properties, but rel- tum measurement. We further show that the relational ative to a fixed QRF. Here the fixed QRF is assume to be probability amplitude can be calculated using Feynman in a simple eigen state. This is what we do in Ref. [32] and path integral [33]. the present work. A complete RQM theory should com- The second improvement in this works comes from the bine these two aspects together. This means one will need introduction of the concept of entanglement to the RQM to reformulate the basic quantum theory from relational theory. We recognize not only that a quantum system properties and relative to a quantum reference frame that must be described relative to another quantum system, exhibits superposition behavior. Therefore, a future step but also that the entanglement between these two sys- is to investigate how the relational probability amplitude tems impacts the formulation the observed system is de- matrix should be formulated when the QRF possesses su- scribed. If there is no entanglement, the observed system perposition properties, and how the relational probability can be described by a wave function. If there is entan- amplitude matrix and the measurement formulations are glement, a reduced density matrix is more appropriate transformed when switching QRFs. mathematical tool. In addition, entanglement measure plays a pivot role in determining a system is undergoing a time evolution or measurement process. This allows E. Conclusions us to reconstruct both the Schr¨odinger equation [32] and the measuring theory. When one states that a quan- Quantum measurement and quantum operation the- tum system must be described relative to another quan- ory is developed here based on the relational formula- tum system, one can further quantify this relativity via tions of quantum mechanics [32]. The relational prop- the entanglement measure between these two systems. erties are the starting point to construct the quantum However, the concept of entanglement is not presented measurement and quantum operation theory. We show in Ref [10]. The reconstruction attempts in Ref [10, 12] that how the relational probability amplitude matrix is to derive the laws of quantum mechanics based on quan- transformed and how mutual information is exchanged tum logic does not yet include quantum measurement during measurement. The resulting formulation is math- theory. ematically compatible with the traditional quantum me- Thirdly, although a quantum system must be described chanics. relative to another quantum system, our work shows that The significance of our formulation comes from the there are mathematical tools that can describe the ob- conceptual consequences. We assert that for a given served system without explicitly calling out the reference quantum system, description of its time evolution can system, such as the wave function or the reduced den- be implicitly relative, while description of a quantum op- 16 eration must be explicitly relative. Information exchange Super Observer. The completeness of quantum mechan- is relative to a local observer in quantum mechanics. The ics and locality can coexist by redefining the element of assumption of Super Observer should be abandoned, so physical reality to be observer-dependent. There might as the notion of observer independent description of phys- be more results from this direction. We further spec- ical reality. Different local observers can achieve consis- ulate that the synchronization of measurement results tent descriptions of a quantum system if they are syn- from different observers is a necessary step when com- chronized on the outcomes from any measurement per- bining quantum mechanics with relativity theory. formed on the system, thus achieve an objective descrip- As stated philosophically in Ref. [45], the physical tion. The conceptual subtlety of the relativity and ob- world is made of processes instead of objects, and the jectivity of a quantum description is not obvious to rec- properties are described in terms of relationships be- ognize in traditional quantum mechanics, because tra- tween events. Based on the initial RQM reformulation ditional quantum mechanics was originally developed to effort [10], Ref. [32] and this paper together further show explain observation results from microscopic system that that quantum mechanics can be constructed by shift- is much smaller than the measuring apparatus. For those ing the starting point from the independent properties situations, RQM and traditional quantum mechanics are of a quantum system to the relational properties among practically equivalent. However, for a composite system quantum systems. The reformulation results in more that is spatially much larger than a typical apparatus, clarity of many subtle physical concepts. We envision the necessity of RQM formulation becomes clear, as man- next step is to investigate how the relational probabil- ifested in the analysis of EPR paradox. The paradox is ity amplitude matrix and the measurement formulations seemingly inevitable in traditional quantum mechanics are transformed when switching QRFs. It is our belief but can be resolved by removing the assumption of the that these efforts together is one step towards a better understanding of quantum mechanics.

[1] Bohr, N.: Quantum Mechanics and Physical Reality, Na- [15] Aharonov, Y. and Kaufherr, T. Quantum frames of ref- ture 136, 65 (1935) erence. Phys. Rev. D. 30.2,368 (1984). [2] Bohr, N.: Can Quantum Mechanical Description of Phys- [16] Palmer, M. C., Girelli, F. and Bartlett, S. D. Changing ical Reality Be Considered Completed? Phys. Rev., 48, quantum reference frames. Phys. Rev. A. 89.5, 052121 696-702 (1935) (2014). [3] Jammer, M.: The Philosophy of Quantum Mechanics: [17] Bartlett, S. D., Rudolph, T., Spekkens, R. W. and The Interpretations of Quantum Mechanics in Historical Turner, P. S. Degradation of a quantum reference frame. Perspective, Chapter 6. New York: Wiley-Interscience, N. J. Phys. 8.4, 58 (2006). (1974) [18] Poulin, D. and Yard, J. Dynamics of a quantum reference [4] Everett, H.:“Relative State” Formulation of Quantum frame. N. J. Phys. 9.5, 156 (2007). Mechanics, Rev of Mod Phys 29, 454 (1957) [19] Rovelli, C. Quantum reference systems. Class. Quantum [5] Wheeler, J. A.: Assessment of Everett’s ”Relative State” Gravity 8.2, 317 (1991). Formulation of Quantum Theory, Rev of Mod Phys 29, [20] Poulin, D. Toy model for a relational formulation of quan- 463 (1957) tum theory. Int. J. Theor. Phys. 45.7, 1189–1215 (2006). [6] DeWitt, B. S.: Quantum mechanics and reality, Physics [21] Girelli, F. and Poulin, D. Quantum reference frames and Today 23, 30 (1970) deformed symmetries. Phys. Rev. D 77.10, 104012 (2008). [7] Zurek, W. H.: Environment-induced Superselection [22] Loveridge, L., Miyadera, T. and Busch, P. Symmetry, Rules, Phys. Rev. D 26, 1862 (1982) reference frames, and relational quantities in quantum [8] Zurek, W. H.: Decoherence, Einselection, and the Quan- mechanics. Found. Phys. 48, 135–198 (2018). tum Origins of the Classical, Rev. of Mod. Phys. 75, 715 [23] J. Pienaar, A relational approach to quantum refer- (2003) ence frames for spins. arXiv preprint at arXiv:1601.07320 [9] Schlosshauer, M.: Decoherence, The Measurement Prob- (2016). lem, and Interpretation of Quantum Mechanics, Rev. [24] Angelo, R. M., Brunner, N., Popescu, S., Short, A. Mod. Phys. 76, 1267-1305 (2004) and Skrzypczyk, P. Physics within a quantum reference [10] Rovelli, C.: Relational Quantum Mechanics, Int. J. frame. J. Phys. A 44.14, 145304 (2011). Theor. Phys., 35, 1637-1678 (1996) [25] Angelo, R. M. and Ribeiro, A. D. Kinematics and dynam- [11] Smerlak M., and Rovelli, C.: Relational EPR, Found. ics in noninertial quantum frames of reference. J. Phys. Phys., 37, 427-445 (2007) A 45.46, 465306 (2012). [12] Transsinelli, M.: Relational Quantum Mechanics and [26] Bartlett, S. D., Rudolph, T., and Spekkens, R. W. Ref- Probability, Found. Phys., 48, 1092-1111 (2018) erence frames, superselection rules, and quantum infor- [13] Aharonov, Y. and Susskind, L. Charge Superselection mation. Rev. Mod. Phys. 79, 555 (2007). Rule. Phys. Rev. 155, 1428 (1967). [27] Gour, G., and Spekkens, R. W. The resource theory [14] Aharonov, Y. and Susskind, L. Observability of the Sign of quantum reference frames: manipulations and mono- Change of Spinors under 2π Rotations. Phys. Rev. 158, tones. N. J. Phys. 10.3, 033023 (2008). 1237 (1967). [28] Bartlett, S. D., Rudolph, T., Spekkens, R. W., and 17

Turner, P. S. Quantum communication using a bounded- Proof: According to the singular value decomposition, size quantum reference frame. N. J. Phys. 11, 063013 the relational matrix R can be decomposed to R = UDV , (2009). where D is rectangular diagonal and both U and V are [29] F. Giacomini, E. Castro-Ruiz, C. Brukner, Quantum Me- N N and M M unitary matrix, respectively. This chanics and the Covariance of Physical Laws in Quantum × × gives ρ = RR† = U(DD†)U †. If H(R) = 0, matrix ρ is a Reference Frame, Nat. Comm. 10:494 (2019). rank one matrix, therefore DD is diag 1, 0, 0... . This [30] A. Vanrietvelde, P. H¨ohn, F. Giacomini, and E. Castro- † means D is a rectangular diagonal matrix{ with only} one Ruiz, A Change of Perspective: Switching Quantum iφ Reference Frames via a Perspective-neutral Framework, eigenvalue e . Expanding the matrix product R = UDV Quantum 4:225 (2020), arXiv:1809.00556. gives [31] J. M. Yang, Switching Quantum Reference Frames iφ R = U D V = U 1e V1 . (A1) for Quantum Measurement, Quantum 4, 283 (2020), ij X in nm mj i j arXiv:1911.04903v4 nm [32] Yang, J. M.: A Relational Formulation of Quantum Me- iφ chanics, Sci. Rep. 8:13305 (2018), arXiv:1706.01317 We just choose ci = Ui1 and dj = e V1j to get Rij = [33] Yang, J. M.: Path Integral Implementation cidj . Conversely, if Rij = cidj , R can be written as outer of Relational Quantum Mechanics, Sci. Rep. product of two vectors, 11, www.nature.com/articles/s41598-021-88045-6, arXiv:1807.01583 T R = c1 c2 ... cn d1 d2 ... dm . (A2) [34] Einstein, A., Podolsky, B., and Rosen, N.: Can
×
Quantum-Mechanical Description of Physical Reality Be Considering vector C1 = c1,c2,...,cn as an eigenvec- Considered Complete? Phys. Rev. 47, 777-780 (1935) tor in Hilbert space , one{ can use the} Gram-Schmidt [35] Von Neumann, J.: Mathematical Foundations of Quan- HS tum Mechanics, Chap. VI. Princeton University Press, procedure [38] to find orthogonal basis set C2,...,Cn. Princeton Translated by Robert T. Beyer (1932/1955) Similarly, considering vector D1 = d1, d2,...,dm as an eigenvector in Hilbert space ,{ one can find} or- [36] Fuchs, C. A.: Quantum Mechanics as Quantum Informa- HA tion (and only a little more). arXiv:quant-ph/0205039, thogonal basis set D2,...,Dm. Under the new orthogo- (2002) nal eigenbasis, R becomes a rectangular diagonal matrix [37] Fuchs, C. A., and Schark, R.: Quantum-Bayesian Co- D = diag 1, 0, 0... . Therefore R = UDV where U and herence: The No-Nonsense Version, Rev. Mod. Phys. 85, V are two{ unitary} matrices associated with the eigen- 1693-1715 (2013) basis transformations. Then ρ = RR† = U(DD†)U †, [38] Nielsen, M. A., and Chuang, I. L.: Quantum compu- and DD† = diag 1, 0, 0... is a square diagonal matrix. tation and quantum information. Cambridge University Since the eigenvalues{ of similar} matrices are the same, Press, Cambridge (2000) the eigenvalues of ρ are (1, 0, ...), thus H(R) = 0. [39] Horodecki, R., Horodecki, P., Horodecki, M., and Horodecki, K.: Quantum Entanglement, Rev. Mod. Phys., 81, 865-942 (2009) [40] Hayashi, M., Ishizaka, S., Kawachi, A., Kimura, G., and Appendix B: Decomposition of the Unitary Ogawa, T.: Introduction to Quantum Information Sci- Operator of a Bipartite System ence, page 90, 150, 152, 197. Springer-Verlag, Berlin Hei- delberg (2015) If there is interaction between S and A, and the ini- [41] Nielsen, M. A., and Chuang, I. L. : Quantum computa- tial state of S + A is a product state, the global uni- tion and quantum information, page 366, 564. Cambridge tary operator can be decomposed into a set of measure- University Press, Cambridge (2000) ment operators that satisfies Eq. (16). The proof shown [42] Allahverdyan, A. E., Roger Balian, R., Nieuwenhuizen, here closely follows idea from Ref. [41]. Denote the ini- T. M.: Understanding quantum measurement from the tial state is product state, Ψ0 = ψ0 φ0 . First we solution of dynamical models, Phys. Rep. 525, 1-166 | i | iS | iA (2013) change the eigenbasis for A through a local unitary op- erator I Uˆ such that φ0 is the first eigenvector of [43] Breuer, H-P., Petruccione, F.: The Theory of Open S ⊗ A Quantum Systems. Oxford University Press, Oxford New the orthogonal eigenbasis, i.e., (IS UˆA) ψ0 S φ0 A = ⊗ | i ⊗ | i York (2007) ψ0 S a0 A, and am forms an orthogonal eigen- [44] Rivas, A., Huelga, S. F.: Open Quantum System, An |basisi ⊗ of |A.i The global{| i} unitary operator is changed Introduction. Springer-Verlag, Berlin Heidelberg (2012) ˆ ′ ˆ ˆ to USA = USA(IS UA† ). Define a linear operator [45] Smolin, L.: Three Roads to Quantum Gravity. Basic ˆ ˆ ′ ⊗ ˆ Books, New York (2017) Mm = am USA a0 . The set of operators Mm is what we areh looking| | for,i since we can verify it{ satisfies} Eq. (16),

ˆ ˆ ′ Appendix A: Theorem 1 USA ψ0 φ0 = USA ψ0 a0 | i| i | i| i ′ = a a Uˆ ψ0 a0 X | mih m| SA| i| i Theorem 1 H(R)=0 if and only if the matrix element m (B1) Rij can be decomposed as Rij = cidj , where ci and dj = Mˆ ψ0 a . X m| i| mi are complex numbers. m 18

The completeness condition can also be verified, Appendix E: Open Quantum System

′ ˆ ˆ ˆ ˆ ′ Mm† Mm = a0 U † am am USA a0 The measurement theory described in this paper is X Xh | SA| ih | | i m m (B2) consistent with the open quantum system (OQS) the- ′ ˆ ˆ ′ ory. OQS studies the dynamics when a quantum sys- = a0 USA† USA a0 = IS . h | | i tem interacts with its environment E [38, 43, 44]. Such interaction can result in entanglement and information Appendix C: Theorem 2 exchange between the quantum system S and the envi- ronment system E. Recall the definition of apparatus in Section II A includes the interacting environment as one Theorem 2 Applying operator Qˆ Oˆ over the composite type of apparatus, if we replace the environment system system S +A is equivalent to change⊗ the relational matrix E with the apparatus system A, the OQS theory gives R to R = QROT , where the superscript T represents a ′ the same formulations as shown in Section III. transposition. First, we give a brief review of the OQS theory. Sup- Proof: Denote the initial state vector of the composite pose the initial composite state for the quantum system and environment is described by a density matrix ρSE, system as Ψ0 = Rij si aj . Apply the composite | i Pij | i| i the interaction between S and E changes the density ma- operator Qˆ(t) Oˆ(t) to the initial state, ˆ ˆ ⊗ trix ρSE UρSEU †. The resulting density matrix of S is →ˆ ˆ ρS′ = T rE(UρSEU †). Denote the orthogonal eigen basis Ψ1 = (Qˆ Oˆ) R s a | i ⊗ X ij | ii ⊗ | j i of the environment ek . Since the orthogonal eigenba- ij sis of the environment{| isi} not necessary the same eigenba- = R Qˆ s Oˆ a sis that diagonalizes ρE, we further denote the spectral X ij | ii⊗ | j i ij decomposition of ρE as ρE = m λm e˜m e˜m . Assumed the initial state of the quantumP system| plusih environment| = R Q O s a X X ij mi nj | mi ⊗ | ni (C1) is a product state, i.e., ρSE = ρS ρE, the density op- ij mn erator of S after the interaction with⊗ the environment = ( Q R OT ) s a is X X mi ij jn | mi ⊗ | ni mn ij ρS′ = Λ(ρS) T = (QRO )mn sm an 2 ˆ ˆ X | i| i = λm ek U(ρS e˜m e˜m )U † ek mn X | | h | ⊗ | ih | | i mk (E1) where T represents the transposition of matrix. Com- = E ρ E† . X mk S mk pared the above equation to Eq.(7) for the definition of mk Ψ1 , it is clear that the relational matrix is changed to | i T R′ = QRO . where E = λ e Uˆ e˜ and satisfies the complete- mk mh k| | mi ness condition mk EmkEmk† = I. Eq.(E1) is the Kraus representationP of the linear map Λ. It is proved that Λ Appendix D: Probability in Selective Measurement can be a Kraus representation if and only if it can be induced from an extended system with initial condition Given the composite system S + A is described by ρ = ρ ρ [44]. If ρ = e0 e0 is a pure state, the SE S ⊗ E E | ih | Eq.(15), the reduced density matrix of S can be defined linear map is further simplified to Λ(ρS) = EkρSE† Pk k and Ek = ek Uˆ e0 . The operator set Ek forms a ′ ′ h | | i { } ρˆS = T rA Ψ Ψ = ( RikRi∗ k) si si POVM, and Λ(ρS) is a Complete Positive Trace Pre- | ih | X′ X | ih | ii k (D1) serving (CPTP) map [40]. To connect to the measure- = (RR†) ′ s s ′ ment theory, suppose the measurement outcome m corre- X ii i i ′ | ih | sponds to an orthogonal state φ of E, and represented ii | mi by a projection operator Pˆ = φ φ , m | mih m| and the probability of finding event si occurred to S is | i m calculated by Eq.(10). Similarly, the probability of event ρS =Λm(ρS) A 2 aj occurred to A is pj = i pij = i Rij . This | i P P | | = ek PˆmUρˆ S e˜0 e˜0 Uˆ †Pˆ† ek can be more elegantly written by introducing a partial Xh | ⊗ | ih | m| i S ˆA ˆA k projection operator I Pj where Pj = aj aj . It is (E2) ⊗ | ih | easy to verify that = φ Uˆ e˜0 ρ e˜0 Uˆ † φ k φ φ k h m| | i Sh | | mi Xh | mih m| i k pA = Ψ IS PˆA Ψ ˆ ˆ j h | ⊗ j | i = MmρSMm† 2 (D2) = Ψ aj aj Ψ = Rij . h | ih | i X | | where Mˆ = φ Uˆ e˜0 is the operator defined on i m h m| | i 19

. The probability of finding outcome m is pm = where H(ρSA) is the von Neumann entropy of the com- HS ˆ ˆ T r(Λm(ρS)) = T r(MmMm† ρS). posite system S + A. For a composite system S + A that It is evident that if we replace the environment system is described by a single relational matrix R, these two E with the apparatus system A, the OQS theory gives the variables differ only by a factor of two. However, for a same formulations as shown in Section III. In the case of composite system of S +A that is described by an ensem- initial product state, Eqs. (21) versus (E2) are effectively ble of relational matrices, the two variables can be very the same. Let’s consider the case of initial entangled different. This is illustrated by two examples described state in the OQS context. Denote the initial system plus below. environment state as pure bipartite state Ψ0 . After Case 1. S + A is in an entangled pure state de- ˆ | i scribed by Ψ SA = λi si ai in Schmidt decompo- the global unitary operation U and subsequent projection | i Pi | i| i S E S sition, where λi are the Schmidt coefficients. Subsys- I Pˆ = I φm φm , the composite state becomes ⊗ m ⊗ | ih | tem S is in a{ mixed} state. The entanglement measure S ˆE ˆ Ψ1 = (I Pm )U Ψ0 , take the partial trace over E, 2 2 | i ⊗ | i is E(ρS)= i λi ln(λi ) and the mutual information is we get I = 2 λ−2lnP(λ2). − Pi i i Case 2. S + A is in a mixed state described by ρSA = S 2 ρ = T rE( Ψ1 Ψ1 m | ih | i λi si ai si ai . In the case, H(ρS) = H(ρSA) = PH(ρ | )i| = ih |h λ|2ln(λ2). ρ is a separable bipartite = φm UˆSA Ψ0 Ψ0 Uˆ † φm (E3) diag i i i SA h | | ih | SA| i state [39, 40].− P There is no entanglement but there is = ψm ψm . 2 2 | ih | mutual information since I = i λi ln(λi ). Essentially the composite system is a mixed− P ensemble of product 2 This implies the resulting state for S is ψm = states λ , si ai . One can infer that S is in si from ˆ | i { i | i| i} | i φm USA Ψ0 , which is equivalent to Eq.(28). Taking a knowing A is in a , however such mutual information is h | | i | ii similar approach in deriving Eq.(29), we can express ψm due to classical correlation. The probability of finding S | i 2 in the eigenbasis derived from the Schmidt decomposition in an eigenvector si is just the classical probability λ . ˆ | i i of Ψ0 . The result is ψm =1/ pm′ i Mmi s˜i , where Although the reduced density operator for S, ρS = | i | i p P | i 2 the definitions of Mˆ and p are the same as those in λ si si , is the same in Case 1 and Case 2, the mi m′ Pi i | ih | Eq.(29) except replacing the apparatus system A with mutual information is different. More information is en- the environment system E. The reduced density opera- coded in the pure bipartite state in Case 1. When S + A S tor for S, ρm, is given by is described by Ψ SA = i λi si ai , besides the infer- ence information| betweeni PS and| i|A,i there is additional indeterminacy due to the superposition at the compos- m 1 ˆ ˆ ρS = Mmi s˜i s˜j Mmj† (E4) ite system level. For instance, one cannot determine the pm′ X | ih | ij composite system S + A is in s0 a0 or s1 a1 before measurement. More indeterminacy| i| beforei | measurementi| i Eq. (E4) can be considered as a generalization of Eq.(E2) means more information can be gained from measure- when the initial state is entangled. There is no simple ment. This also explains that in the EPR experiment, m form of map Λm such that ρS =Λm(ρS ) where ρS is the when Alice measures particle α, she does not only gain initial density matrix of S. A different representation of information about α, but also gain information about the m ρS can be derived by rewriting the initial state ρSE = composite system. Thus, she can predict the state of par- ρS ρE + ρcorr where ρcorr is a correlation term [44]. ticle β. On the other hand, such indeterminacy does not ⊗ exist when S + A is described by a mixture of product state as in Case 2. Since such indeterminacy is for the composite system as a whole, the reduced density oper- Appendix F: Mutual Information ator for a subsystem S, ρS, cannot reflect the difference, therefore it appears the same in Case 1 and Case 2. Entanglement between the two systems is measured by These two examples show that mutual information the parameter E(ρ) as defined by Eq.(13). In Section III, variable can substitute the entanglement measurement we also use the mutual information variable to measure only when the composite system S + A is described by a the information exchange between the measuring system single relational matrix R. To quantify change of quan- and the measured system. The mutual information be- tum correlation during a measurement, the entanglement tween S and A is defined as I = H(ρ )+H(ρ ) H(ρ ), measurement is a more appropriate parameter. S A − SA