Researchers have introduced the fTNN, a novel tensor neural network approach designed to tackle fractional partial differential equations (PDEs) on bounded domains. This methodology focuses on solving the fractional Poisson equation and time-dependent fractional advection-diffusion equation, leveraging a geometry-adapted integration split that incorporates a spatially dependent near-field radius. The fTNN's subspace method enables efficient and accurate solutions to these complex equations, which are crucial in various fields such as physics and engineering. By employing a deterministic tensor neural network, the fTNN offers a robust framework for addressing fractional PDEs1. The development of the fTNN has significant implications for advancing numerical analysis and scientific computing, particularly in areas where fractional calculus plays a key role. So what matters to practitioners is that the fTNN's innovative approach can potentially enhance the accuracy and efficiency of simulations, leading to breakthroughs in fields like materials science and fluid dynamics.