Neural implicit solution formula for efficiently solving Hamilton-Jacobi equations (Q6889655)

In the abstract, the authors say the following: The paper presents an implicit solution formula for the Hamilton-Jacobi partial differential equation (HJ PDE), which is derived using the method of characteristics and is shown to coincide with the Hopf and Lax formulas in the case where either the Hamiltonian or the initial function is convex. The formula provides a simple and efficient numerical approach for computing the viscosity solution of HJ PDEs, bypassing the need for the Legendre transform of the Hamiltonian or the initial condition, and the explicit computation of individual characteristic trajectories. A deep learning-based neural network method is then proposed to learn this implicit solution formula. Building upon this neural network, an algorithm is developed that approximates the characteristic curves piecewise linearly for state-dependent Hamiltonians. Experimental results show that the proposed method delivers highly accurate solutions, even for nonconvex Hamiltonians, and exhibits remarkable scalability, achieving computational efficiency for problems up to 40 dimensions. \N\NAs an important direction for future work, the authors point out that a comprehensive analysis is needed to confirm whether the proposed formula describes the viscosity solution of HJ PDEs in nonconvex problems and the convergence analysis is also needed for the proposed minimization problem lying in the neural network method for the nonconvex case.