On the Expressiveness of TPTL and MTL over ω-Data Words

Abstract: Metric Temporal Logic (MTL) and Timed Propositional Temporal Logic (TPTL) are prominent extensions of Linear Temporal Logic to specify properties about data languages. In this paper, we consider the class of data languages of non-monotonic data words over the natural numbers. We prove that, in this setting, TPTL is strictly more expressive than MTL. To this end, we introduce Ehrenfeucht-Fraisse (EF) games for MTL. Using EF games for MTL, we also prove that the MTL definability decision problem ("Given a TPTL-formula, is the language defined by this formula definable in MTL?") is undecidable. We also define EF games for TPTL, and we show the effect of various syntactic restrictions on the expressiveness of MTL and TPTL.