Here you will find a code and related work that may be useful.
SQUIC is an L1-regularized maximum likelihood method for performant large-scale sparse precision matrix estimation. SQUIC is available as a shared library (libSQUIC) intended for Linux and Mac OS and with Python, R, and C++ APIs.
The DDSG technique is a grid-based function approximation method that combines High-Dimension Model Representation (HDMR), a variant of Dimensional Decomposition (DD), and adaptive Sparse Grid (SG). The combined approach enables a highly performant and scalable gird base function approximation method that can scale efficiently to high dimensions and on distributed architectures. This library is user-friendly and parallelized with MPI. The SG part of this algorithm uses the Tasmanian open-source sparse grid library is used.
Building Interpretable Climate Emulators for Economics
A linear model to emulate the complexities of physical non-linear dynamics in more sophisticated models, such as Earth System Models (ESMs) or Earth System Models of Intermediate Complexity (ESMIC), and to capture the variability of carbon cycle responses across these models. The focus is on two carbon cycle model configurations: serial and parallel reservoir configurations. The models include three reservoir classes: atmosphere (A), ocean (O), and land biosphere (L). The serial model (3SR) consists of three sequentially connected carbon reservoirs, with the atmosphere connected to the upper ocean (O1), and the ocean connected to the deep ocean (O2). The parallel model (4PR) introduces the land biosphere, where carbon from the atmosphere is divided into two parallel streams: land biosphere and ocean.