GenSBI is a JAX-based library for simulation-based inference (SBI) using state-of-the-art generative methods, currently centred on optimal-transport flow matching and diffusion models. It offers a flexible, high-performance toolkit for inference problems where the likelihood is intractable, featuring transformer-based architectures (SimFormer, Flux1, and Flux1Join), built-in posterior calibration diagnostics (SBC, TARP, L-C2ST), and a high-level recipes API. Built on JAX and Flax NNX, it runs seamlessly on CPU, GPU, and TPU. Installation pip install gensbi # with GPU (CUDA 12) support: pip install gensbi[cuda12] For other installation options
Amerio, Aurelio
This repository is for the paper "Dynamics of Solar Atmospheric Heating: Establishing Causality Between Photospheric Flux Cancellation and Chromospheric Responses via Deep Learning and Multi-Instrument Observations" by Ahmed Ibrahim Abdalla. It is associated with manuscript number AAS79712. This reproducibility package contains the custom Python scripts (including the SDB-ResNet deep learning framework, 3D PFSS extrapolations, dynamic self-calibration algorithms, and Si IV spectroscopic extraction) required to reproduce the figures and kinematic analysis presented in the manuscript.
Abdalla, Ahmed
Here you will find codes and necessary files used to perform Random Forest machine learning analysis, bootstrapping analysis, data pruning analysis, and plotting for the Manuscript Assessing early-stage net CO2 fluxes from soils and maize plants during a short-term enhanced weathering mesocosm trial. "ML_N2O_v3.R" is used to run the Random Forest machine learning analysis. "ml_Functions.R" contains the functions needed to run "ML_N2O_v3.R" "Bootstrap Code.py" is used to run the bootstrapping function and make figures for the bootstrap analysis. "Run 1 CO2_Plotter.py" and "Run 2 CO2_Plotter.py" are used to make figures to plot the fluxes and c
Chiaravalloti, Isabella, Zhang, Shuang, Planavsky, Noah
Initial release of the MATLAB code and processed data supporting: Hould Gosselin, G., Rutter, N., Mann, P. J., Marsh, P., and Sonnentag, O.: Fine-scale spatial variability of winter CO2 and CH4 fluxes in Arctic tundra derived from snowpack gradient measurements, Biogeosciences (in review), egusphere-2025-5637. Contents Processed snowpack CO2 and CH4 concentration profiles, concentration gradient models, and per-profile flux inputs and outputs for 119 microsites along five transects (T1-Shrub, T2-Mixed, T3-Tussock, T4-Forest, T5-Inlet) at Trail Valley Creek, Northwest Territories, Canada, sampled 20 to 25 March 2024. Processing scripts for con
Gabriel Hould Gosselin
Description This repository contains the complete codebase for the manuscript "Human-induced vegetation shifts drive global declines in urban vegetation transpiration." The software implements a hybrid physics–deep learning framework (CNN-PM) for estimating urban vegetation transpiration (VTu), a six-model comparison framework for systematic architecture and input ablation analysis, a multi-source geospatial data preprocessing pipeline, an XGBoost-SHAP trend attribution module, three evapotranspiration partitioning modules (uWUE, Pérez-Priego and TEA) and a two-dimensional flux footprint module (FFP). All code is written in Python and built o
Chen, Han
Abstract Solar energetic particle (SEP) events are rare, high-impact space weather phenomena that pose serious radiation hazards to astronauts and polar-route aircraft crews, damage satellite electronics, and disrupt satellite communications and navigation systems. Forecasting SEPs allows preventive measures to be taken before an event's onset. This task is difficult because SEP events, defined as periods when the flux of > 10 MeV protons exceeds 10 proton flux units (pfu), are extremely rare and make up less than 1% of the samples in our dataset. We present a stage-by-stage study of a novel preprocessing pipeline for SEP forecasting (normali
EskandariNasab, MohammadReza, Woodhouse, Garrett, Hamdi, Shah Muhammad, Filali Boubrahimi, Soukaina
This archive accompanies the manuscript “Neural Network Imputation of the Pitch-Angle-Resolved Medium-Energy Electron Flux Data in LEO.” It provides the materials required to demonstrate and verify the inference process of eight trained neural network models developed for imputing missing local pitch angle electron flux measurements from the Medium-Energy Electron Detector (MEED) aboard the Fengyun-3E satellite. The archive includes trained models for the E7 and E8 energy channels (E7, 280–410 keV; E8, 410–600 keV) and the PA1, PA2, PA8, and PA9 local pitch angle bins (0°–20°, 20°–40°, 140°–160°, and 160°–180°, respectively), together with th
Chen, Jiali, Zou, Hong, Ye, Yuguang, Sun, Jiachen, Li, Tianhao
In this revision, the authors would like to share the training function utilized for splitting the dataset by days, trying to avoid that some specific operational/meteorological conditions are not included in the training dataset because of random splitting. The specific function utilized for Sim2Real inference has been included, considering the additional performance metrics required for this step.
Díaz Alonso, Sergio
The code provided here was used to produce the results in the submitted manuscript, titled 'Target-dependent Uncertainty Analysis in Air-Sea Heat Flux Predictions using a Convolutional Neural Network in the Mediterranean Sea' , in the journal Ocean-Land-Atmosphere Research.
Ghani, Mahmud Hasan
Reproduction package and permanent record of the Anulum entry to the Fusion Equilibrium Challenge (NeurIPS 2026 competition, Codabench 17739, organised by Sophelio): reconstruction of the poloidal flux map of a tokamak plasma, and of q95 and normalised beta, from coil currents and Thomson-scattering profiles alone, without any magnetic measurement. DIII-D: an ensemble of 69 multilayer perceptrons predicting principal coordinates of the flux map, with gradient-boosted tree heads, networks dedicated to the X-points and to the submitted scalars, and a map-finishing step. MAST (zero-shot): an equal-temperature reconstruction from the coil set and
Šotek, Miroslav
This is the initial release of FluxNet, the repository accompanying "Deep Learning Ice Shelf Basal Melt Rates via Differentiable Physics".
Bente, Kim
Science is approaching a critical thermal boundary. As high-performance computing, artificial intelligence, and advanced infrastructure scale, semiconductor power densities are breaching the $1,000\text{ W/cm}^2$ thresholds established by the DARPA ICECool and ARPA-E COOLERCHIPS benchmarks. Conventional computational fluid dynamics (CFD) packages struggle to model these extreme boundaries, often facing mathematical divergence, software crashes, or requiring massive supercomputing overhead. The Microchannel-CHF-Solver is a high-precision finite-difference thermal-scalar solver designed specifically to resolve Critical Heat Flux (CHF) instabili
Schramm, Daniel
