This is the Jax implementation of MaxMode algorithm, based on the works of:
A. F. López-Lopera, F. Bachoc, N. Durrande and O. Roustant (2018), "Finite-dimensional Gaussian approximation with linear inequality constraints". SIAM/ASA Journal on Uncertainty Quantification, 6(3): 1224–1255.
F. Bachoc, A. Lagnoux and A. F. López-Lopera (2019), "Maximum likelihood estimation for Gaussian processes under inequality constraints". Electronic Journal of Statistics, 13 (2): 2921-2969.
F. Bachoc, A. F. López-Lopera and O. Roustant (2022), "Sequential construction and dimension reduction of Gaussian processes under inequality constraints". Journal on Mathematics of Data Science
López-Lopera, A., Bachoc, F. and Roustant, O., 2022. "High-dimensional additive Gaussian processes under monotonicity constraints". Advances in Neural Information Processing Systems, 35, pp.8041-8053.
Since it is based on Jax, it supports training and inference on GPU/TPU natively.