Time functions as utilities

From MaRDI portal



Abstract: Every time function on spacetime gives a (continuous) total preordering of the spacetime events which respects the notion of causal precedence. The problem of the existence of a (semi-)time function on spacetime and the problem of recovering the causal structure starting from the set of time functions are studied. It is pointed out that these problems have an analog in the field of microeconomics known as utility theory. In a chronological spacetime the semi-time functions correspond to the utilities for the chronological relation, while in a K-causal (stably causal) spacetime the time functions correspond to the utilities for the K^+ relation (Seifert's relation). By exploiting this analogy, we are able to import some mathematical results, most notably Peleg's and Levin's theorems, to the spacetime framework. As a consequence, we prove that a K-causal (i.e. stably causal) spacetime admits a time function and that the time or temporal functions can be used to recover the K^+ (or Seifert) relation which indeed turns out to be the intersection of the time or temporal orderings. This result tells us in which circumstances it is possible to recover the chronological or causal relation starting from the set of time or temporal functions allowed by the spacetime. Moreover, it is proved that a chronological spacetime in which the closure of the causal relation is transitive (for instance a reflective spacetime) admits a semi-time function. Along the way a new proof avoiding smoothing techniques is given that the existence of a time function implies stable causality, and a new short proof of the equivalence between K-causality and stable causality is given which takes advantage of Levin's theorem and smoothing techniques.


This work imports the idea of utility function, originally developed in the field of microeconomics, to the relativistic space-time framework and proves some important results which relate this concept with the causality properties of space-time. Let us give some definition: We denote, as usual in relativistic physics, the causal and chronological relations by \(\leq\) and \(\ll\). A semi-time function is defined to be a continuous map \(t:M\to \mathbb{R}\) such that \(x\ll y \Rightarrow t(x)<t(y)\). A time function is defined to be a continuous map \(t:M\to \mathbb{R}\) such that \(x<y \Rightarrow t(x)<t(y)\). A temporal function is defined to be a \(C^1\) time function with a past directed time-like gradient. The Seifert relation is \(J^{+}_S = \bigcap_{g'>g} J^{+}_{g'}\), where \(g'>g\) means that the time-like cone of \(g'\) contains the causal cone of \(g\), and \(J^{+}_{g'} \subset M \times M\) is the causal relation for the space-time \((M,g')\). The Seifert relation is closed, reflexive and transitive, and furthermore, it is antisymmetric if and only if the space-time is stably causal [\textit{E. Minguzzi}, Classical Quantum Gravity 25, No.~1, Article ID 015010 (2008; Zbl 1132.83003)]. The \(K^{+}\) relation is instead the smallest closed and transitive relation which contains the causal relation [\textit{R. D. Sorkin} and \textit{E. Woolgar}, Classical Quantum Gravity 13, No.~7, 1971--1993 (1996; Zbl 0966.83022)], as a consequence \(J^{+}\subset K^+\subset J^{+}_S\). A space-time is \(K\)-causal if the relation \(K^{+}\) is antisymmetric. It must be mentioned that, in a previous work, the author proved the equivalence between \(K\)-causality and stable causality, and the equality \(K^+ = J^+_S\) in stably causal space-times [\textit{E. Minguzzi}, Commun. Math. Phys. 290, No.~1, 239--248 (2009; Zbl 1182.53065)] (this result is not used in this work where he offers a different proof). A preorder \(R\) on a set \(X\) is defined to be a reflexive and transitive relation on \(X\). An antisymmetric preorder is a partial order. Totality means \(x\leq_R y\) or \(y\leq_R x\). A utility on set \(X\) endowed with a preorder \(R\) is defined to be a function \(u : X\to \mathbb{R}\) with the strictly isotone property \[ \text{``}x \sim_R y \Rightarrow u(x) = u(y)\text{}\text{ and }\text{``}x <_R y \Rightarrow u(x)< u(y).\text{} \] As a first step, the author relates the concept of utility with that of time function by proving the following: Theorem. In a \(K\)-causal space-time the family of continuous \(K^{+}\)-utilities coincides with the family of time functions. Then the author takes advantage of Levin's theorem [\textit{V. L. Levin}, Sov. Math., Dokl. 28, 715--718 (1983); translation from Dokl. Akad. Nauk SSSR 273, 800--804 (1983; Zbl 0554.90012)], which concerns the existence of continuous utility functions on a topological space, to prove the following: Theorem. A space-time is \(K\)-causal if and only if it admits a time function (as a consequence time functions are always \(K^{+}\)-utilities). In this case, denoting by \(\mathcal{A}\) the set of time functions, the partial order \(K^{+}\) can be recovered from the time functions, that is \[ (x,y) \in K^{+}\Leftrightarrow\forall t \in \mathcal{A}, \;t(x)\leq t(y). \] In particular every \(K\)-causal space-time admits a time function. Using the equivalence between the existence of time functions and stable causality, he reobtains that \(K\)-causality coincides with stable causality. From the previous theorem, he also obtains that a chronological space-time in which \(\overline{J^{+}}\) is transitive admits a semi-time function. The author ends the paper by introducing, for every time (temporal) function \(t\), a total preorder, on \(M\) given by \[ T^{+}[t] = \{(p,q)\in M\times M: t(p)\leq t(q)\}, \] and proving the following result, which shows that the Seifert relation can be recovered from the set of time or temporal functions allowed by the space-time: Theorem. Let \(\mathcal{A}\) and \(\mathcal{B}\) be respectively the set of time functions and the set of temporal functions allowed by a space-time. In every space-time, \[ K^{+}\subset J^{+}_S\subset \bigcap_{t\in \,\mathcal{A}}\, T^{+}[t] \subset \bigcap_{t\in \,\mathcal{B}}\, T^{+}[t]. \] In a stably causal space-time, \[ K^{+}= J^{+}_S= \bigcap_{t\in \,\mathcal{A}}\, T^{+}[t] = \bigcap_{t\in \,\mathcal{B}}\, T^{+}[t]. \] Then, the author argues that since, in general, the identity \(J^{+}_S=J^+\) does not hold (but it holds in causally simple space-times, and globally hyperbolic space-times that are causally simple), the causal relation \(J^+_S\) is physically more significant than \(J^+\).



Cites work









This page was built for publication: Time functions as utilities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q625479)