Unilateral determination of causal order in a cyclic process

Ilyass Mejdoub, Augustin Vanrietvelde · 2025

Read Unilateral determination of causal order in a cyclic process by Ilyass Mejdoub, Augustin Vanrietvelde (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-18540.txt.

Introduction

Ilyass Mejdoub and Augustin Vanrietvelde ask a sharp question about cause and effect in a 2025 physics paper. Some processes allow events to form a cycle of causes while staying free of logical clashes. In all known cases, the order of events is set jointly by what every agent does. The authors ask whether one agent, acting alone, could fix her own order relative to others.

They show that the answer is yes. They build a process in which any party can, by her own action alone, place any other party in her future. They also give a causal inequality that this process breaks to the fullest extent. The result widens what kinds of causal structure physics may allow.

Catalog record

CREATORIlyass Mejdoub, Augustin Vanrietvelde
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE51 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-18540

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Unilateral determination of causal order in a cyclic process Ilyass Mejdoub, Augustin Vanrietvelde Abstract The recent years have seen interest into the possibility for (classical as well as quantum) causal structures that, while remaining logically consistent, feature a cyclic causal order between events, opening intriguing possibilities for new physics. In the cyclic processes displayed so far, the global causal order is determined jointly by the operations performed by agents at each event, a feature that can be certified through the introduction of causal games and (the violation of) causal inequalities. This raises the question of whether there exist processes in which an agent acting at a single event can unilaterally determine her causal ordering with respect to some other events. We answer this question in the affirmative, by introducing a process in which any party may be put in a position to pick, on her own, any other party to lie in her future. We certify this behaviour by displaying a related causal inequality that the process allows to maximally violate. I Introduction It is usually assumed that the causal order between events should be definite and acyclic, and that a scenario featuring a cyclic causal order would necessarily involve logical inconsistencies (e.g. of the grandfather paradox type), making its physical existence implausible. Over the past decade, however, it has been shown – through the development of the framework of superchannels chiribella2013quantum , or equivalently of process matrices oreshkov2012quantum ; araujo2017purification – that there exist processes that display a cyclic causal order while remaining logically consistent. This has led to various investigations of these processes’ classification, properties, physical plausibility, and potential uses. Among these exotic processes, the best-known are those involving the coherent (and possibly dynamical) control of causal order by a quantum system wechs2021quantum . However, there also exist processes that do not fall into this category, which we will call here processes beyond quantum control oreshkov2012quantum ; baumeler2014maximal ; baumeler2016space ; araujo2017quantum ; araujo2017purification ; 2017Quant…1…39A ; Baumeler:2021dnu . These allow for the violation of causal inequalities , the analogue of Bell inequalities for causal order oreshkov2012quantum ; branciard2015simplest ; this violation certifies the incompatibility of observed probability distributions with a definite causal order in a device-independent way. Interestingly, classical processes violating causal inequalities also exist, an example being the Lugano process (also called Baumeler-Wolff or Araújo-Feix) baumeler2014maximal ; baumeler2016space . In this context, it is natural to ask how far acausality can be taken: how causally exotic can a process be, while remaining logically consistent? A common feature of the processes beyond quantum control displayed up until now is that they can be conceived of as featuring a multilateral determination of causal order: part of the causal order (typically, the fact that one given event lies in the other events’ future) is determined by correlations between the choices of operations being performed at each event. It would seem intuitive that this ‘globalness’ of the determination of causal order is a necessary feature of processes beyond quantum control, ensuring that their causal cyclicity does not entail logical inconsistencies. In this paper, we show that this intuition is wrong: some processes feature unilateral determination of (part of the) causal order. By this, we mean more precisely that for each event A in the process, there exists a choice of operations being performed at other events such that the operation being performed at A determines, alone, which among all other events lies in A ’s causal future. We first formalise unilateral determination of causal order by introducing a causal game, played collaboratively by parties acting at each event, that encapsulates this notion. We show that this game satisfies a causal inequality: a bound on the winning probability of parties acting in any fixed causal order. Using a recently proposed method for building exotic processes, we then build a new process and prove that its use allows for a deterministic winning strategy in our causal game, violating the causal inequality and demonstrating unilateral determination of causal order. II A causal game for unilateral determination of causal order We now introduce a new causal game, aimed at pinning down the notion of unilateral determination of causal order. Denote by A_{1},...,A_{n} , with n\geq 5 , the participants to this game. Each A_{q} receives a binary input {I^{q}\in\{0,1\}} , picked uniformly at random, and can output a binary value O^{q}\in\{0,1\} . The referee also chooses uniformly at random a ‘sender’ party Q_{\textrm{send}}\in\{1,\ldots,n\} , and a ‘receiver’ party Q_{\textrm{rec}} , with Q_{\textrm{send}}\neq Q_{\textrm{rec}} . The label Q_{\textrm{send}} is communicated to all parties, but Q_{\textrm{rec}} is only communicated to the sender A^{Q_{\textrm{send}}} . The game is won if the receiver correctly guesses the input of the sender, i.e. \mathbb{P}(\textrm{success}):=\mathbb{P}(O^{Q_{\textrm{rec}}}=I^{Q_{\textrm{send}}})\,. (1) This should be contrasted with the form of a causal game for the Lugano process, as presented e.g. in Ref. vanrietvelde2022 . The Lugano process can also be used to win a game in which the referee picks at random a sender and receiver, the former of which has to communicate a message to the latter. However, this would require the identity of both the sender and receiver to be communicated by the referee to all parties. In contrast, here, only the sender’s identity is communicated to everyone, while the receiver’s identity is communicated to the sender alone. This makes the game starkly more difficult; in intuitive terms, a winning strategy now requires the ability for the sender to determine on her own which party will be given her input, as opposed to the identity of this party being determined jointly by the actions of all agents (as happens, for instance, in a strategy based on access to the Lugano process). In a definite causal order scenario, the game we introduced can’t be won with certainty, as we have the causal inequality \mathbb{P}_{\text{DCO}}(\text{success})\leq\dfrac{3}{4}\,, satisfied by all strategies featuring (possibly classical mixtures of) definite causal order. This probability can be computed as follows. If the sender happens to be in the causal past of the receiver, she transmits her input through the chain of subsequent parties and the receiver can output the correct answer. Otherwise, the receiver will have to guess the sender’s input, which will be correct with probability \dfrac{1}{2} . Hence the probability of success is \begin{split}&\mathbb{P}_{\text{DCO}}(\text{success})\\ &=\mathbb{P}(\text{$P_{s}$ before $P_{r}$})\mathbb{P}(\text{success}|\text{$P_{s}$ before $P_{r}$})\\ &\quad+\mathbb{P}(\text{$P_{r}$ before $P_{s}$})\mathbb{P}(\text{success}|\text{$P_{r}$ before $P_{s}$})\\ &=\dfrac{1}{2}\cdot 1+\dfrac{1}{2}\cdot\dfrac{1}{2}\\ &=\dfrac{3}{4}\,.\end{split} (2) By symmetry of the roles of P_{r} and P_{s} , we have \mathbb{P}(\text{$P_{s}$ before $P_{r}$})=\mathbb{P}(\text{$P_{r}$ before $P_{s}$})=\dfrac{1}{2} . In contrast, access to the process introduced in this paper allows the n parties to win the previous game with certainty, for any n\geq 5 . III Routed circuits In order to specify the new process, we will employ the framework of routed circuits vanrietvelde:phd . We very briefly and semi-formally recap the methods employed, and refer the reader to past work for the technical details. Here, a quantum system A is associated to a finite-dimensional Hilbert space \mathcal{H}_{A} , and a quantum channel of type A\to B is a completely positive trace-preserving (CPTP) map \mathcal{C}:\mathcal{L}(\mathcal{H}_{A})\to\mathcal{L}(\mathcal{H}_{B}) , where \mathcal{L}(\mathcal{H}) denotes the space of linear operators on \mathcal{H} . We start by introducing higher-order operations. A superchannel \mathcal{S} (sometimes also called a supermap) is a higher-order

Read the complete text