research Efficient Boltzmann Sampling Consistency models with importance sampling for efficient, unbiased Boltzmann distributions. Diffusion Samplers for Molecular Energies Diffusion-based generative samplers for molecular energy functions. Constrain-KL for Variational Autoencoders A constrained optimisation approach for setting exact KL targets in beta-VAE training. numerical-methods Deflation Techniques for Nonlinear PDEs Sparse Jacobian-Newton deflation in Julia for finding multiple nonlinear PDE solutions. teaching Teaching and Peer Tutoring Tutorials and mentoring for first-year mathematics and numerical analysis students.
