Causality in the Quantum World
Pienaar, Jacques
2017-07-31 00:00:00
VIEWPOINT A new model extends the denition of causality to quantum-mechanical systems. by Jacques Pienaar athematical models for deducing cause-effect relationships from statistical data have been successful in diverse areas of science (see Ref. M [1] and references therein). Such models can be applied, for instance, to establish causal relationships between smoking and cancer or to analyze risks in con- struction projects. Can similar models be extended to the microscopic world governed by the laws of quantum me- chanics? Answering this question could lead to advances in quantum information and to a better understanding of the foundations of quantum mechanics. Developing quantum Figure 1: In statistics, causal models can be used to extract extensions of causal models, however, has proven challeng- cause-effect relationships from empirical data on a complex ing because of the peculiar features of quantum mechanics. system. Existing models, however, do not apply if at least one component of the system (Y) is quantum. Allen et al. have now For instance, if two or more quantum systems are entangled, proposed a quantum extension of causal models. (APS/Alan it is hard to deduce whether statistical correlations between Stonebraker) them imply a cause-effect relationship. John-Mark Allen at the University of Oxford, UK, and colleagues have now pro- posed a quantum causal model based on a generalization of an old principle known as Reichenbach’s common cause is a third variable that is a common cause of both. In the principle [2]. latter case, the correlation will disappear if probabilities are conditioned to the common cause. For example, the inci- Historically, statisticians thought that all information dence of tsunamis in Chile is statistically correlated with about a system could be represented in terms of statistical that of tsunamis in Japan. In statistical terms, the combined correlations among its variables. Nowadays, however, it is probability for two tsunamis is greater than the product of recognized that the concept of causal information goes be- the separate probabilities for tsunamis in Chile and Japan. yond that of correlation. For example, compare the statistical But neither event is a cause of the other. If we condition statement “the number of cars is correlated with the amount the tsunamis’ probabilities on the knowledge that an earth- of air pollution” and the causal statement “cars cause air pol- quake has occurred in the Paciﬁc basin, then we should ﬁnd lution.” The statistical statement goes both ways: Knowing that the events are independent: the combined (conditional) there are more cars, I can infer that the air is more pol- probability of the two is equal to the product of the separate luted. Similarly, knowing the air is more polluted, I can (conditional) probabilities. In other words, the correlation infer that there are more cars. The causal statement tells us disappears. Given our knowledge of an earthquake, the more; namely, if we change the number of cars, we can affect news that a tsunami occurred in Chile no longer gives us air pollution, but not vice versa—polluting the air by other any extra information about the probability that a tsunami means (say by building factories) will not affect the number occurred in Japan. Reichenbach’s conditional independence of cars. Causal information is different from correlations be- suggests that the earthquake might be the common cause of cause it tells us how the system changes under interventions. the tsunamis in the two regions. In classical causal models, statistical and causal infor- Besides providing clues about causal relationships, con- mation are related by the Reichenbach’s principle. This ditional independence relations tell us how to update the principle states that two correlated variables must have a probability of an event based on new information related common cause: either one is a cause of the other or there to the event—a procedure called Bayesian inference. The two types of inference (Bayesian and causal), connected by International Institute of Physics, Lagoa Nova, Natal - RN, 59078- Reichenbach’s principle, are the heart of causal models. A 970, Brazil quantum extension of such models should provide a frame- physics.aps.org 2017 American Physical Society 31 July 2017 Physics 10, 86 work for both. Allen et al. realized that by replacing “deterministic” In the case of two entangled particles, Reichenbach’s with “unitary” in Reichenbach’s principle they could ob- principle would suggest that the correlations between the tain a new version of quantum causal models. In particular, particles could be explained by a common cause. However, their quantum version of the Reichenbach principle allowed we also know that quantum statistics can violate Bell’s in- them to relate conditional independence to quantum causal equalities, which means that variables serving as common relationships like those described in Costa and Shrapnel’s causes that could make the correlation disappear cannot ex- model. What’s more, these conditional independence re- ist. A quantum causal model should redeﬁne the connection lations could then be used to perform Bayesian inference. between causal statements and statistical observations by ac- Allen et al.’s result combines both causal interventions and counting for this phenomenon (see Fig. 1). It should also tell Bayesian inference into a single model, succeeding where us how to derive conditional independence relations, which others had failed. in turn allow us to perform Bayesian updating of probabili- Several research groups, including mine, are still explor- ties. Finding a model that meets both of these requirements ing a range of alternative quantum causal theories. But has been challenging. the new model by Allen and colleagues is the ﬁrst to meet Most early attempts at quantum causal models proceeded all requirements of a quantum causal model, providing a by deﬁning causal structures for quantum systems and then uniquely quantum deﬁnition of causality. Thanks to results ﬁnding which conditional independence relationships re- like this, we may ﬁnd that quantum mechanics has a causal mained intact [3]. However, these models could not perform interpretation, just like classical mechanics. We might also Bayesian inference because conditional independence was reveal the mechanisms that are behind observed correlations no longer a prerequisite for identifying a common cause. and pinpoint the interventions that manipulate such mecha- Matthew Leifer and Robert Spekkens [4] attempted to in- nisms. In a few words, this would amount to bringing back corporate Bayesian inference in a quantum framework using some cause-effect “intuition” into the spooky and bizarre “conditional quantum states” in place of conditional proba- world of quantum mechanics. bilities, but this creative approach was found to be applica- ble only in restricted cases. Fabio Costa and Sally Shrapnel This research is published in Physical Review X. [5] set aside the problem of conditional independences to focus on causal interventions. For example, instead of con- sidering the conditional independence of tsunamis in Chile REFERENCES and Japan, their approach would consider whether creating or preventing earthquakes (an intervention) would trigger [1] J. Pearl, Causality: Models, Reasoning and Inference (Cam- or suppress the tsunami events through physical processes. bridge University Press, Cambridge, 2009). [2] J.-M. A. Allen, J. Barrett, D. C. Horsman, C. M. Lee, and R. W. This model allowed causal relationships to be deﬁned, but Spekkens, ``Quantum Common Causes and Quantum Causal it lacked the conditional independences with which to per- Models,'' Phys. Rev. X 7, 031021 (2017). form Bayesian inference. [3] K. Laskey, ``Quantum Causal Networks,'' arXiv:0710.1200; R. Building on the work of Costa and Shrapnel, Allen and his R. Tucci, ``Quantum Bayesian Nets,'' Int. J. Mod. Phys. B 9, 295 colleagues set out to restore conditional independence as a (1995); T. Fritz, ``Beyond Bell's Theorem II: Scenarios with Arbi- prerequisite for common causes. To do so, they took advan- trary Causal Structure,'' Commun. Math. Phys. 341, 391 (2015); tage of an old physical argument that derives Reichenbach’s J. Henson, R. Lal, and M. F. Pusey, ``Theory-Independent Lim- principle by assuming that statistical data are the result of its on Correlations from Generalized Bayesian Networks,'' New a deterministic model. For instance, rolling dice in a casino J. Phys. 16, 113043 (2014); J. Pienaar and . Brukner, ``A might appear random, but it could be explained, in princi- Graph Separation Theorem for Quantum Causal Models,'' New ple, by a croupier whose skills allow him to determine the J. Phys. 17, 073020 (2015). [4] M. S. Leifer and R. W. Spekkens, ``Towards a Formulation of outcome of each throw. While it is debatable whether quan- Quantum Theory as a Causally Neutral Theory of Bayesian In- tum systems are compatible with this type of determinism, ference,'' Phys. Rev. A 88, 052130 (2013). they are compatible with another type of determinism called [5] F. Costa and S. Shrapnel, ``Quantum Causal Modelling,'' New J. unitary evolution. A process is called unitary if it conserves Phys. 18, 063032 (2016). quantum information. Compatibility with unitarity is a cen- tral tenet of quantum mechanics. 10.1103/Physics.10.86 physics.aps.org 2017 American Physical Society 31 July 2017 Physics 10, 86
http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.pngPhysicsAmerican Physical Society (APS)http://www.deepdyve.com/lp/american-physical-society-aps/causality-in-the-quantum-world-xODSekgcqJ

VIEWPOINT A new model extends the denition of causality to quantum-mechanical systems. by Jacques Pienaar athematical models for deducing cause-effect relationships from statistical data have been successful in diverse areas of science (see Ref. M [1] and references therein). Such models can be applied, for instance, to establish causal relationships between smoking and cancer or to analyze risks in con- struction projects. Can similar models be extended to the microscopic world governed by the laws of quantum me- chanics? Answering this question could lead to advances in quantum information and to a better understanding of the foundations of quantum mechanics. Developing quantum Figure 1: In statistics, causal models can be used to extract extensions of causal models, however, has proven challeng- cause-effect relationships from empirical data on a complex ing because of the peculiar features of quantum mechanics. system. Existing models, however, do not apply if at least one component of the system (Y) is quantum. Allen et al. have now For instance, if two or more quantum systems are entangled, proposed a quantum extension of causal models. (APS/Alan it is hard to deduce whether statistical correlations between Stonebraker) them imply a cause-effect relationship. John-Mark Allen at the University of Oxford, UK, and colleagues have now pro- posed a quantum causal model based on a generalization of an old principle known as Reichenbach’s common cause is a third variable that is a common cause of both. In the principle [2]. latter case, the correlation will disappear if probabilities are conditioned to the common cause. For example, the inci- Historically, statisticians thought that all information dence of tsunamis in Chile is statistically correlated with about a system could be represented in terms of statistical that of tsunamis in Japan. In statistical terms, the combined correlations among its variables. Nowadays, however, it is probability for two tsunamis is greater than the product of recognized that the concept of causal information goes be- the separate probabilities for tsunamis in Chile and Japan. yond that of correlation. For example, compare the statistical But neither event is a cause of the other. If we condition statement “the number of cars is correlated with the amount the tsunamis’ probabilities on the knowledge that an earth- of air pollution” and the causal statement “cars cause air pol- quake has occurred in the Paciﬁc basin, then we should ﬁnd lution.” The statistical statement goes both ways: Knowing that the events are independent: the combined (conditional) there are more cars, I can infer that the air is more pol- probability of the two is equal to the product of the separate luted. Similarly, knowing the air is more polluted, I can (conditional) probabilities. In other words, the correlation infer that there are more cars. The causal statement tells us disappears. Given our knowledge of an earthquake, the more; namely, if we change the number of cars, we can affect news that a tsunami occurred in Chile no longer gives us air pollution, but not vice versa—polluting the air by other any extra information about the probability that a tsunami means (say by building factories) will not affect the number occurred in Japan. Reichenbach’s conditional independence of cars. Causal information is different from correlations be- suggests that the earthquake might be the common cause of cause it tells us how the system changes under interventions. the tsunamis in the two regions. In classical causal models, statistical and causal infor- Besides providing clues about causal relationships, con- mation are related by the Reichenbach’s principle. This ditional independence relations tell us how to update the principle states that two correlated variables must have a probability of an event based on new information related common cause: either one is a cause of the other or there to the event—a procedure called Bayesian inference. The two types of inference (Bayesian and causal), connected by International Institute of Physics, Lagoa Nova, Natal - RN, 59078- Reichenbach’s principle, are the heart of causal models. A 970, Brazil quantum extension of such models should provide a frame- physics.aps.org 2017 American Physical Society 31 July 2017 Physics 10, 86 work for both. Allen et al. realized that by replacing “deterministic” In the case of two entangled particles, Reichenbach’s with “unitary” in Reichenbach’s principle they could ob- principle would suggest that the correlations between the tain a new version of quantum causal models. In particular, particles could be explained by a common cause. However, their quantum version of the Reichenbach principle allowed we also know that quantum statistics can violate Bell’s in- them to relate conditional independence to quantum causal equalities, which means that variables serving as common relationships like those described in Costa and Shrapnel’s causes that could make the correlation disappear cannot ex- model. What’s more, these conditional independence re- ist. A quantum causal model should redeﬁne the connection lations could then be used to perform Bayesian inference. between causal statements and statistical observations by ac- Allen et al.’s result combines both causal interventions and counting for this phenomenon (see Fig. 1). It should also tell Bayesian inference into a single model, succeeding where us how to derive conditional independence relations, which others had failed. in turn allow us to perform Bayesian updating of probabili- Several research groups, including mine, are still explor- ties. Finding a model that meets both of these requirements ing a range of alternative quantum causal theories. But has been challenging. the new model by Allen and colleagues is the ﬁrst to meet Most early attempts at quantum causal models proceeded all requirements of a quantum causal model, providing a by deﬁning causal structures for quantum systems and then uniquely quantum deﬁnition of causality. Thanks to results ﬁnding which conditional independence relationships re- like this, we may ﬁnd that quantum mechanics has a causal mained intact [3]. However, these models could not perform interpretation, just like classical mechanics. We might also Bayesian inference because conditional independence was reveal the mechanisms that are behind observed correlations no longer a prerequisite for identifying a common cause. and pinpoint the interventions that manipulate such mecha- Matthew Leifer and Robert Spekkens [4] attempted to in- nisms. In a few words, this would amount to bringing back corporate Bayesian inference in a quantum framework using some cause-effect “intuition” into the spooky and bizarre “conditional quantum states” in place of conditional proba- world of quantum mechanics. bilities, but this creative approach was found to be applica- ble only in restricted cases. Fabio Costa and Sally Shrapnel This research is published in Physical Review X. [5] set aside the problem of conditional independences to focus on causal interventions. For example, instead of con- sidering the conditional independence of tsunamis in Chile REFERENCES and Japan, their approach would consider whether creating or preventing earthquakes (an intervention) would trigger [1] J. Pearl, Causality: Models, Reasoning and Inference (Cam- or suppress the tsunami events through physical processes. bridge University Press, Cambridge, 2009). [2] J.-M. A. Allen, J. Barrett, D. C. Horsman, C. M. Lee, and R. W. This model allowed causal relationships to be deﬁned, but Spekkens, ``Quantum Common Causes and Quantum Causal it lacked the conditional independences with which to per- Models,'' Phys. Rev. X 7, 031021 (2017). form Bayesian inference. [3] K. Laskey, ``Quantum Causal Networks,'' arXiv:0710.1200; R. Building on the work of Costa and Shrapnel, Allen and his R. Tucci, ``Quantum Bayesian Nets,'' Int. J. Mod. Phys. B 9, 295 colleagues set out to restore conditional independence as a (1995); T. Fritz, ``Beyond Bell's Theorem II: Scenarios with Arbi- prerequisite for common causes. To do so, they took advan- trary Causal Structure,'' Commun. Math. Phys. 341, 391 (2015); tage of an old physical argument that derives Reichenbach’s J. Henson, R. Lal, and M. F. Pusey, ``Theory-Independent Lim- principle by assuming that statistical data are the result of its on Correlations from Generalized Bayesian Networks,'' New a deterministic model. For instance, rolling dice in a casino J. Phys. 16, 113043 (2014); J. Pienaar and . Brukner, ``A might appear random, but it could be explained, in princi- Graph Separation Theorem for Quantum Causal Models,'' New ple, by a croupier whose skills allow him to determine the J. Phys. 17, 073020 (2015). [4] M. S. Leifer and R. W. Spekkens, ``Towards a Formulation of outcome of each throw. While it is debatable whether quan- Quantum Theory as a Causally Neutral Theory of Bayesian In- tum systems are compatible with this type of determinism, ference,'' Phys. Rev. A 88, 052130 (2013). they are compatible with another type of determinism called [5] F. Costa and S. Shrapnel, ``Quantum Causal Modelling,'' New J. unitary evolution. A process is called unitary if it conserves Phys. 18, 063032 (2016). quantum information. Compatibility with unitarity is a cen- tral tenet of quantum mechanics. 10.1103/Physics.10.86 physics.aps.org 2017 American Physical Society 31 July 2017 Physics 10, 86

Journal

Physics
– American Physical Society (APS)

Published: Jul 31, 2017

Recommended Articles

Loading...

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.

“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”

Daniel C.

“Whoa! It’s like Spotify but for academic articles.”

@Phil_Robichaud

“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”

@deepthiw

“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”

Bookmark this article. You can see your Bookmarks on your DeepDyve Library.

To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.

Subscribe to Journal Email Alerts

To subscribe to email alerts, please log in first, or sign up for a DeepDyve account if you don’t already have one.

Follow a Journal

To get new article updates from a journal on your personalized homepage, please log in first, or sign up for a DeepDyve account if you don’t already have one.