A Day-to-Day Dynamical Approach to the Most Likely User Equilibrium Problem

• DOI: 10.1287/trsc.2024.0525
• Publication Date: 2024-07-08T14:17:46Z
• Authors: Jiayang Li; Qianni Wang; Liyang Feng; Jun Xie; Yu (Marco) Nie
• Keywords: Present day; Day to day

The lack of a unique user equilibrium (UE) route flow in traffic assignment has posed significant challenge to many transportation applications. maximum-entropy principle, which advocates for the consistent selection most likely solution, is often used address challenge. Built on recently proposed day-to-day discrete-time dynamical model called cumulative logit (CumLog), this study provides new behavioral underpinning (MEUE) flow. It been proven that CumLog can reach UE state without presuming travelers are perfectly rational. Here, we further establish always converges MEUE if (i) have no prior information about routes and thus, forced give all an equal initial choice probability or (ii) gather from same source such general proportionality condition satisfied. Thus, may be as practical solution algorithm problem. To put idea into practice, propose eliminate enumeration requirement original through iterative discovery scheme. We also examine versions four popular continuous-time models compare them with CumLog. analysis shows replicator dynamic only one potential some regularity. analytical results confirmed numerical experiments. History: This paper accepted Transportation Science Special Issue ISTTT25 Conference. Funding: research was funded by United States National Foundation’s Division Civil, Mechanical Manufacturing Innovation [Grant 2225087]. work J. Xie Natural Foundation China 72371205]. Supplemental Material: online appendix available at https://doi.org/10.1287/trsc.2024.0525 .