ongoing Differentiable Programming Using differentiable programming for neural implicit modeling, optimization, control and inverse design in science and engineering. Ultra-low Energy Loss in Biomolecular Motor Figure out the underlying mechanism of ~100% efficiency for bacterial flagellar motor. Physics Constrained Learning Impose physics priors into the ML workflow (data, architecture, loss function, optimization). done AI for Applied Physics/Mechanics Figure out the potentials of AI in multiferroic composites, complex fluid flows and near-field radiative heat transfer. Nanoscale Thermal Devices Advances near-field radiative heat transfer applications, dramatically enhancing energy efficiency in thermal rectifiers and thermophotovoltaic systems.