arxiv:1509.03641v1 [quant-ph] 11 Sep 2015 Thermodynamical cost of some interpretations of quantum theory Adán Cabello, 1, a) Mile Gu, 2, 3 Otfried Gühne, 4 Jan-Åke Larsson, 5 and Karoline Wiesner 6 1) Departamento de Física Aplicada II, Universidad de Sevilla, E-41012 Sevilla, Spain. 2) Centre for Quantum Information, Institute for Interdisciplinary Information Sciences, Tsinghua University, 100084 Beijing, China. 3) Centre for Quantum Technologies, National University of Singapore, 3 Science Drive 2, Singapore 117543, Singapore. 4) Naturwissenschaftlich-Technische Fakultät, Universität Siegen, Walter-Flex-Straße 3, D-57068 Siegen, Germany. 5) Institutionen för systemteknik, Linköpings Universitet, SE-58183 Linköping, Sweden. 6) School of Mathematics, Centre for Complexity Sciences, University of Bristol, University Walk, Bristol BS8 1TW, UK. The interpretation of quantum theory is one of the longest-standing debates in physics 1. One class of interpretations sees quantum probabilities as determined by intrinsic properties of the world 2 15 ; these are called type-i interpretations. Another class sees quantum probabilities as not directly dealing with intrinsic properties of the world but with relational experiences between an observer and the world 16 26 ; these are called type-ii interpretations. It is usually believed that deciding between these two types cannot be made simply on purely physical grounds but it requires an act of metaphysical judgement 27. However, this decision is crucial for the physical meaning of phenomena such as quantum teleportation 28 or the violation of Bell s inequalities 29, and for the progress of quantum theory. Here we show that, although the problem is undecidable within the framework of quantum theory, it is decidable, under some assumptions, within the framework of thermodynamics. We prove that type-i interpretations are incompatible with the following assumptions: (i) the decision of which measurement is performed on a quantum system can be made independently of the system, (ii) a quantum system has limited memory, and (iii) Landauer s principle 30 is valid. We consider an ideal experiment in which an individual quantum system is submitted to a sequence of quantum projective measurements that leave the system in pure quantum states. We show that in any type-i interpretation satisfying (i) (iii) the system must reset its internal state, which implies that a minimum amount of heat per measurement has to be dissipated into the system s environment. We calculate a lower bound to the heat dissipated per measurement assuming that the measurements are chosen from a set of size 2 n. Then, we show that this lower bound becomes infinite in the limit of n tending to infinity. This leads to the conclusion that either type-i interpretations are untenable or at least one of the assumptions (i) (iii) has to be abandoned. Another possibility is that quantum theory is only an idealization of a deeper theory in which the actual state space is finite 31,32. This can be experimentally tested since, if this is the case, some finite heat has to be dissipated in each measurement due to Landauer s principle. The non-detection of this heat would support those interpretations in which probabilities are not determined by intrinsic properties. Consider the following ideal experiment: A single qubit is sequentially measured at time intervals by an observer who performs projective measurements randomly chosen between the Pauli observables σ z and σ x. Each of these measurements has two possible outcomes: +1 or 1. When measuring σ z, if the outcome is +1, the quantum state after the measurement is 0 ; if the outcome is 1, the quantum state is 1. Similarly, when measuring σ x, if the outcome is +1 the quantum state is + = 1 2 ( 0 + 1 ), and if the outcome is 1 the quantum state is = 1 2 ( 0 1 ). Therefore, the quantum state after each measurement is always one of the four quantum states 0, 1, +, and. The quantum state after the measurement at time t is the quantum state before the measurement at t + 1. The process is repeated infinitely many times. If the measurements are a) Electronic mail: adan@us.es randomly chosen, the quantum state has probability 1 2 to change and probability 1 2 not to change. We will consider type-i interpretations satisfying assumptions (i) (iii). In these interpretations, at any t the quantum probabilities are determined by intrinsic properties of the system and, according to assumption (i), these intrinsic properties change depending on what measurements are performed. Assumption (ii) implies that the system cannot have stored the values of the intrinsic properties for all possible sequences of measurements that the observer can perform. This implies that the system has to generate new values and store them in its memory. For that reason, the system needs to erase part of the previously existing information. Assumption (iii) is that Landauer s principle 30 holds. Landauer s principle states that the erasure of one bit of information from the information-bearing degrees of freedom of a system must be accompanied by a corresponding entropy increase in the non-information-bearing degrees of freedom of the system, therefore causing dissipation of at least kt ln 2
2 units of heat, where k is the Boltzmann constant and T is the temperature of the system. Landauer s principle has been verified in actual experiments 33,34 and is considered valid in the quantum domain 35,36. Therefore, whenever the temperature is not zero, assumption (iii) implies that the system should dissipate, at least, an amount of heat proportional to the information erased. To calculate the minimum information that the system must erase per measurement, a key observation is that our ideal experiment is an example of a stochastic inputoutput process which can be analysed in informationtheoretic terms. A stochastic process Y is a one-dimensional chain..., Y 2, Y 1, Y 0, Y 1, Y 2,... of discrete random variables {Y t } t Z that take values {y t } t Z over a finite or countably infinite alphabet Y. An input-output process Y X with input alphabet X and output alphabet Y is a collection of stochastic processes Y X { Y x } x X, where each such process Y x corresponds to all possible output sequences Y given a particular bi-infinite input sequence x. It can be represented as a finitestate automaton or, equivalently, as a hidden Markov process. For this to have physical consequences we have to consider the minimal and optimal representation. In our ideal experiment, x t is the observable measured at time t and y t the corresponding outcome. By X we denote the chain of previous measurements,..., X t 2, X t 1, by X we denote X t, X t+1,..., and by X, the chain..., Xt 1, X t, X t+1,.... Similarly, Y, Y, and Y denote the past, future and all outcomes, respectively, while Z, Z and Z denote the past, future, and all pairs of measurements and outcomes. The fact that our ideal experiment is an input-output process implies 37 that there exists a unique minimal and optimal predictor of the process, i.e., a unique finite-state machine with minimal entropy over the state probability distribution and maximal mutual information with the process future output given the process input-output past and the process future input. This machine is called the process ɛ-transducer 37 and is the extension of the so-called ɛ-machines 38,39. An ɛ-transducer of an input-output process is a tuple (X, Y, S, T ) consisting of the process input and output alphabets X and Y, the set of causal states S, and the set of corresponding conditional transition probabilities T. The causal states s t 1 S are the equivalence classes in which the set of input-output pasts Z can be partitioned in such a way that two input-output pasts z and z are equivalent if and only if the probabilities P ( Y X, Z = z ) and P ( Y X, Z = z ) are equal. The causal states are a so-called sufficient statistic of the process. They store all the information about the past needed to predict the output and as little as possible of the remaining information overhead contained in the past. The Shannon entropy over the stationary distribution of the causal states H(S) is the so-called statistical complexity, and represents the minimum internal entropy needed to be stored to optimally compute future measurement outcomes (this quantity generally depends on how our measurements X are selected, here we assume each X t is selected from a uniform probability distribution). The set of conditional transition probabilities T {P (S t+1 = s j, Y t = y S t = s i, X t = x)} governs the evolution. The fact that the ɛ-transducer is also the machine producing minimum heat can be proven as follows. The average information that must be erased per measurement is the information contained in the causal state previous to the measurement, S t 1, that is not contained in the causal state after the measurement, S t 40. This is equal to the conditional entropy (or uncertainty) of S t 1 given S t, X t, and Y t, I erased = H(S t 1 X t, Y t, S t ). (1) For the machine reproducing the outcomes of our ideal experiment, knowledge of the current state gives complete information about the last outcome, so that H(Y t X t, S t ) = 0. Then, the properties of conditional entropy give I erased = H(S t X t, S t 1 ) H(S t X t ) + H(S t 1 X t ). (2) We note that H(S t 1 X t ) = H(S t 1 ) = H(S t ) because measurements are chosen at random and ɛ-transducers are stationary, so that for our ideal experiment the amount of information erased is I erased = H(S t X t, S t 1 ) + I(S t : X t ), (3) where the mutual information I(S t : X t ) = H(S t ) H(S t X t ). States R t of another partition R of inputoutput pasts Z that are as predictive as the causal states 39 obey H(R t X t, R t 1 ) H(S t X t, S t 1 ) and the data-processing inequality (see, e.g., Ref. 41 ) guarantees that I(S t : X t ) I(R t : X t ). It follows that the quantity I erased is minimal for an ɛ-transducer. In addition, the ɛ-transducer of our ideal experiment has a particular property, namely, that there is a one-toone correspondence between causal states s t and quantum states ψ t Ψ. All input-output pasts in a given causal state s t give the same probabilities for all X, so that the quantum states given by the input-output pasts must be equal. Conversely, a given quantum state provides probabilities for all X that correspond to one causal state of the ɛ-transducer. As a consequence, the ɛ- transducer associated with our ideal experiment has four causal states that we will denote s 0, s 1, s +, and s, corresponding to the four quantum states 0, 1, +, and. Result 1. For the experiment in which a qubit is submitted to sequential measurements randomly chosen from X = {σ x, σ z } at temperature T, under assumptions (i) (iii), any interpretation of quantum theory in which probabilities are determined by intrinsic properties of the
3 system predicts that, on average, the system should dissipate, at least, 3 2kT ln 2 units of heat per measurement. Proof: Since, in our ideal experiment, the stationary distribution of the causal states is uniform and the observer randomly chooses the measurement, we can calculate I erased by using a particular causal state, e.g., S t = s 0 and a particular measurement that lead to it, e.g., X t = σ z. Using Eq. (1) and the fact that H(Y t S t ) = 0 we obtain (a) x t (b) x t I erased = H(S t 1 σ z, s 0 ) = P (S t 1 = s j σ z, s 0 ) log P (S t 1 = s j σ z, s 0 ), s j S (4) where logarithms are to base two, 0 log 0 is taken as zero, and P (S t 1 = s j σ z, s 0 ) is the conditional probability that the causal state at time-step t 1 was s j, given a measurement choice σ z and subsequent transition to causal state s 0 at time-step t. There are only three possible causal states at time t: s 0, s +, and s, whose respective conditional probabilities P (S t 1 = s j σ z, s 0 ) are 1 2, 1 4, and 1 4. Therefore, I erased = 1 2 log 1 2 21 4 log 1 4 = 3 bits. (5) 2 Applying Landauers principle we obtain the corresponding lower bound to the average heat per measurement dissipated by the system. In contrast, if probabilities are not determined by intrinsic properties of the system, then measurement outcomes are created randomly when the observables are measured, without any need to overwrite information in the system and therefore without the system dissipating heat due to Landauer s principle (Fig. 1). Result 1 assumes that the observer can only choose between two measurements. What if the observer chooses her measurements from a larger set? Result 2. For the experiment in which a qubit is submitted to sequential measurements randomly chosen from { X (n) = cos ( ) πk 2 n σz + sin ( ) } πk 2 n σx ; k = 0,..., 2 n 1, (6) under assumptions (i) (iii), any interpretation of quantum theory in which probabilities are determined by intrinsic properties of the system predicts that, on average, the system should dissipate an amount of heat per measurement that tends to infinity linearly with n. Proof: The sets X (n) and { Ψ(n) = cos ( ) ( πj 2 0 + sin πj ) } n+1 2 1, j = 0,..., 2 n+1 1 n+1 (7) have, for any n N, the same properties we exploited for the proof of Result 1. Indeed, X (1) = X, Ψ(1) = Ψ, and s 0 is associated to 0 for all n. Therefore, we can follow the same strategy as in the proof of Result 1. The number of pure quantum states attained by the system now y t FIG. 1. Thermodynamical consequences of different interpretations of quantum theory. Interpretations of quantum theory can be divided into type-i and type-ii, depending on whether or not the quantum probabilities of measurement outcomes are determined by intrinsic properties of the observed system. In some experiments in which a quantum system is sequentially measured at time intervals (X t denotes the measurement at time t and Y t its outcome), thermodynamics introduces a difference between the two types. (a) In type-i interpretations, the observed system is in a given causal state before the measurement and this state suffers an intrinsic change due to the measurement. Due to Landauer s principle, this change causes heat emission (represented in red) from the information-bearing degrees of freedom of the observed system. (b) In type-ii interpretations, the quantum state of the observed system does not correspond to any intrinsic property of the observed system. Here, the quantum state corresponds to the knowledge 16 23,26 or expectations 24,25 an external observer has. Therefore, the measurement does not cause heat emission from the observed system (represented by the grey area); Landauer s principle may still cause external observers to emit heat. grows exponentially with n and, consequently, the number of causal states in the associated ɛ-transducer also grows exponentially with n, and the needed conditional probability is cos ( ) 2 πj 2 cos ( ) 2 πj P (S t 1 = s j σ z, s 0 ) = n+1 2 n+1 2 n. y t 2 n+1 1 l=0 cos 2 ( πl 2 n+1 ) = (8) Inserting this probability in the erased information, and using that cos 2 φ < 1 when 0 < φ < π inside the logarithm, we obtain 2 n+1 1 I erased (n) = j=0 cos ( ) ( 2 πj πj ) 2 n+1 2 n log cos2 2 n+1 2 n > n. (9) Therefore, when the temperature of the system is not zero, the lower bound of the heat dissipated by the system from Landauer s principle tends to infinity linearly
4 with n. The conclusion is that interpretations in which quantum probabilities are determined by intrinsic properties (type-i interpretations) and assumptions (i) (iii) lead to an untenable prediction, namely, that the system dissipates an unbounded amount of heat in each measurement. The size of the system is irrelevant for reaching this conclusion; all we have to assume is that it is finite. Therefore, one has to abandon either type-i interpretations or one of the assumptions (i) (iii). Abandoning (i) would mean that the decision of which measurement is performed on the system cannot be made independently of the system. Abandoning (ii) would mean that, at any given time, a finite system contains infinite information. Abandoning both (i) and (ii) would allow for superdeterminism, and excluding this possibility is beyond the power of logic 18. Abandoning (iii) would mean that Landauer s principle fails at the level of type-i interpretations. Besides abandoning some of (i) (iii), another possible solution would be to assume that the set of possible measurements and states is actually discretized. Toy theories with this property have been shown to retain fundamental features of quantum theory 31,32. This solution would imply that quantum theory is only a continuous idealization of a deeper discrete theory. Nevertheless, the system should radiate a certain amount of heat that depends on the size of the system and the length of the discretization. Experiments may establish bounds on the ratio between these quantities. However, if no heat is radiated, this would support those interpretations in which quantum probabilities are not determined by intrinsic properties of the world. Ultimately, our work indicates that the long-standing question, Do the outcomes of experiments on quantum systems correspond to intrinsic properties? is not purely metaphysical. Its answer in the affirmative has considerable physical consequences, testable through experimental observation. Its falsification will be equally exciting as it will force us to embrace radically new lines of thought. Acknowledgements: This work was supported by the FQXi large grant project The Nature of Information in Sequential Quantum Measurements, projects FIS2011-29400 and FIS2014-60843-P (MINECO, Spain) with FEDER funds and the DFG. MG was supported by the John Templeton Foundation Grant 54914 Occam s Quantum Mechanical Razor: Can Quantum Theory admit the Simplest Understanding of Reality?, the National Basic Research Program of China Grants 2011CBA00300 and 2011CBA00302, and the National Natural Science Foundation of China Grants 11450110058, 61033001 and 61361136003. KW acknowledges EPSRC support via EP/E5012141. We thank D. Z. Albert, J. Anders, M. Araújo, L. E. Ballentine, H. R. Brown, Č. Brukner, J. Bub, C. Budroni, J. I. Cirac, D. Dieks, C. A. Fuchs, R. B. Griffiths, M. Kleinmann, M. Leifer, O. Lombardi, N. D. Mermin, R. Schack, J. Thompson, C. Timpson, C. Schmiegelow, A. G. White and C. Wunderlich for conversations on different aspects of the problem addressed here. 1 Wheeler, J. A. & Zurek, W. H. (eds.) Quantum Theory and Measurement (Princeton University Press, Princeton, N.J., 1983). 2 Einstein, A. Physics and reality. J. Franklin Inst. 221, 349 382 (1936). 3 Bohm, D. A suggested interpretation of the quantum theory in terms of hidden variables. I & II. Phys. Rev. 85, 166 193 (1952). 4 Goldstein, S. Bohmian mechanics. In The Stanford Encyclopedia of Philosophy, edited by E. N. Zalta (2013). 5 Everett III, H. Relative state formulation of quantum mechanics. Rev. Mod. Phys. 29, 454 462 (1957). 6 Vaidman, L. Many-worlds interpretation of quantum mechanics. In The Stanford Encyclopedia of Philosophy, edited by E. N. Zalta (2015). 7 Ballentine, L. E. The statistical interpretation of quantum mechanics. Rev. Mod. Phys. 42, 358 381 (1970). 8 Van Fraassen, B. C., A formal approach to the philosophy of science. In Paradigms and Paradoxes: The Philosophical Challenge of the Quantum Domain, edited by R. Colodny (University of Pittsburgh Press, Pittsburgh), pp. 303 366 (1972). 9 Lombardi, O. & Dieks, D. Modal interpretations of quantum mechanics. In The Stanford Encyclopedia of Philosophy, edited by E. N. Zalta (2012). 10 Bell, J. S. The theory of local beables, Epistemological Lett. No. 9 (March 1976). 11 Griffiths, R. B. Consistent histories and the interpretation of quantum mechanics. J. Stat. Phys. 36, 219 272 (1984). 12 Griffiths, R. B. The consistent histories approach to quantum mechanics. In The Stanford Encyclopedia of Philosophy, edited by E. N. Zalta (2014). 13 Ghirardi, G. C., Rimini, A. & Weber, T. Unified dynamics for microscopic and macroscopic systems. Phys. Rev. D 34, 470 491 (1986). 14 Ghirardi, G. C., Collapse theories. In The Stanford Encyclopedia of Philosophy, edited by E. N. Zalta (2011). 15 Spekkens, R. W. Evidence for the epistemic view of quantum states: A toy theory. Phys. Rev. A 75, 032110 (2007). 16 Bohr, N. The Philosophical Writings of Niels Bohr, edited by J. Faye & H. J. Folse (Ox Bow Press, Woodbridge, Connecticut, 1998). 17 Faye, J. Copenhagen interpretation of quantum mechanics. In The Stanford Encyclopedia of Philosophy, edited by E. N. Zalta (2014). 18 Wheeler, J. A. Law without law. In Quantum Theory and Measurement, edited by J. A. Wheeler & W. H. Zurek (Princeton University Press, Princeton, New Jersey, 1983), pp. 182 213. 19 Kochen, S. A new interpretation of quantum mechanics. In Symposium on the Foundations of Modern Physics: 50 Years of the Einstein-Podolsky-Rosen Experiment, edited by P. J. Lahti & P. Mittelstaedt (World Scientific, Singapore, 1985), pp. 151 169. 20 Rovelli, C. Relational quantum mechanics, Int. J. Th. Phys. 35, 1637 1678 (1996). 21 Zeilinger, A. A foundational principle for quantum mechanics. Found. Phys. 29, 631 643 (1999). 22 Zeilinger, A. The message of the quantum. Nature 438, 743 (2005). 23 Fuchs, C. A. & Peres, A. Quantum theory needs no interpretation. Phys. Today 53, No. 3, 70 71 (2000). 24 Fuchs, C. A. QBism, the Perimeter of Quantum Bayesianism. arxiv:1003.5209. 25 Fuchs, C. A., Mermin, N. D. & Schack, R. An introduction to QBism with an application to the locality of quantum mechanics. Am. J. Phys. 82, No. 8, 749 754 (2014). 26 Brukner, Č. On the quantum measurement problem. To appear in the Proceedings of the Conference Quantum UnSpeakables
5 II: 50 Years of Bell s Theorem (Vienna, 19 22 June 2014); arxiv:1507.05255. 27 Polkinghorne, J. Physics and theology. Europhysics News 45, No. 1, 28 31 (2014). 28 Bennett, C. H., Brassard, G., Crépeau, C., Jozsa, R., Peres, A. & Wootters, W. K. Teleporting an unknown quantum state via dual classical and Einstein Podolsky Rosen channels. Phys. Rev. Lett. 70, 1895 1899 (1993). 29 Bell, J. S. On the Einstein Podolsky Rosen paradox. Physics 1, 195 200 (1964). 30 Landauer, R. Irreversibility and heat generation in the computing process. IBM J. Res. Dev. 5, 183 191 (1961). 31 Schumacher, B. & Westmoreland, M. D. Modal quantum theory. Found. Phys. 42, 918 925 (2012). 32 Hanson, A. J., Ortiz, G., Sabry, A. & Tai, Y.-T. Geometry of discrete quantum computing. J. Phys. A: Math. Theor. 46, 185301 (2013). 33 Bérut, A., Arakelyan, A., Petrosyan, A., Ciliberto, S., Dillenschneider, R. & Lutz, E. Experimental verification of Landauer s principle linking information and thermodynamics. Nature 483, 187 189 (2012). 34 Jun, Y., Gavrilov, M. & Bechhoefer, J. High-precision test of Landauer s principle in a feedback trap. Phys. Rev. Lett. 113, 190601 (2014). 35 Hilt, S., Shabbir, S., Anders, J. & Lutz, E. Landauer s principle in the quantum regime. Phys. Rev. E 83, 030102(R) (2011). 36 Reeb, D. & Wolf, M. M. An improved Landauer principle with finite-size corrections. New J. Phys. 16, 103011 (2014). 37 Barnett, N. & Crutchfield, J. P. Computational mechanics of input-output processes: Structured transformations and the ɛ- transducer. arxiv:1412.2690. 38 Crutchfield, J. P. & Young, K. Inferring statistical complexity. Phys. Rev. Lett. 63, 105 108 (1989). 39 Shalizi, C. R. & Crutchfield, J. P. Computational mechanics: Pattern and prediction, structure and simplicity. J. Stat. Phys. 104, 817 879 (2001). 40 Wiesner, K., Gu, M., Rieper, E. & Vedral, V. Informationtheoretic lower bound on energy cost of stochastic computation. Proc. R. Soc. A 468, 4058 4066 (2012). 41 Cover, T. M. & Thomas, J. A. Elements of Information Theory (Wiley-Blackwell, Hoboken, New Jersey, 2006).