Deep Global, Local and MultiStep solvers approximating the solutions of FBSDEs with jumps which are related to a class of (PIDEs) Partial Integro-Differential Equations.

A pytorch implementation of Backward Stochastic Differential Equation

Neural hedging policies (GRU, deep-set, no-trade-band) minimising CVaR, entropic and spectral risk under transaction costs, on fused CUDA kernels at billions of paths per second, with a deep BSDE pricer.