Conference Agenda
Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
Please note that all times are shown in the time zone of the conference. The current conference time is: 10th Sept 2026, 07:06:13pm CEST (Norway)
External resources will be made available 30 min before a session starts. You may have to reload the page to access the resources.
|
Daily Overview |
| Session | |
|
Thematic Session 6: Growth with Transitions, Tipping and Climate Shocks (HYBRID) Location: Auditorium M: Jan Mossin Session Chair: Alexandra Brausmann, University of Vienna Session Chair: Lucas Bretschger, ETH Zurich | |
    | |
Transversality Condition Matters: Ensuring Uniqueness of Deep Learning Solutions in Economics and Finance ETH Zurich, Switzerland Transversality is an important sufficient condition for identifying the solution in infinite horizon economic and financial models. Without such a condition, there exists a continuum of functions that satisfy the Hamilton-Jacobi-Bellman (HJB) functional equation. In this paper, we explore this manifold of solutions with numerical and analytical methods. Using a standard continuous-time model, we demonstrate that, without explicitly imposing the transversality condition, widely used numerical algorithms, including the (Deep) Galerkin-type methods, may converge to arbitrary points of this manifold, leading to significant and uncontrollable biases. Using an example of the AK-Ramsey model with logarithmic utility (a prototypical model for many environmental economics and financial mathematics applications), the paper demonstrates that the area of direct applicability of projection-type algorithms is narrower than one might expect based on contemporary literature. We propose a novel approach using a functional transformation of the original HJB equation to effectively incorporate the transversality condition, ensuring convergence to the actual value function. | |