Gaussian Processes on Manifolds

• Generate ground truth -> python3 scripts/ground_truth.py <name>
• Generate samples -> ./build/src/examples/generate_target <name>
• FEM eigenfunction -> mpirun -np 1 ./build/src/examples/fem_laplace <name> <num-modes>
• Diffusion maps eigenfunction -> mpirun -np 1 ./build/src/examples/diffusion_laplace <name> <num-modes>
• Ambient solution -> ./build/src/examples/ambient_solution <name>
• FEM solution -> ./build/src/examples/fem_solution <name> <num-modes>
• Diffusion maps solution -> ./build/src/examples/diffusion_solution <name> <num-modes>
• Plot eigenfunction -> ./build/src/examples/plot_eigenfun <name> <fun-num>
• Plot solution -> ./build/src/examples/plot_solution <name>
• Plot embedding -> python3 scripts/plot_embedding.py <name> <num-modes>

• Add sigma method to Gaussian Process
• Use internal Gaussian LLT decomposition of the gram matrix for GP methods
• Complete PetscVector and test it with PetscMatrix for multiple processes
• Check MPI/OPENMP how they share the cores
• MPI with shared-memory: https://stackoverflow.com/questions/64631418/passing-a-pointer-to-mpi-win-allocate-shared-through-a-wrapper https://stackoverflow.com/questions/39912588/can-i-use-mpi-with-shared-memory https://stackoverflow.com/questions/38592854/how-to-send-the-data-in-an-eigenmatrixxd-with-mpi