Granum: an agentic, GPU-accelerated discrete element program for soil

Granum is a discrete element program for soil written in Rust, with a double-precisionCPU engine, single-precision CUDA and wgpu engines, a desktop interface built around lab-oratory tests and staged models, and a Model Context Protocol server through which an AIagent can build, run and report models. It adds to the usual spheres, clumps, Hertz–Mindlincontacts and parallel bonds a breathing-aggregate model of clay and cemented clay. Onthe impact benchmarks of Chung and Ooi (2011), both engines match Hertz’s solutions towithin 0.1% and the oblique impacts of Kratos to within 0.2◦ above 15◦. On the DEM8penetration benchmark, the CUDA engine reproduces the results of eight open-source codes,including Yade, LIGGGHTS and MercuryDPM, for beds of 25,000 to 100,000 particles in 2.3to 12.7 minutes, faster than every CPU code but two to four times slower than the dedicatedGPU codes. The tests exposed, and the program now corrects, a loss of displacement insingle-precision position updates and an inefficient grid for a large sphere among small ones.In geotechnical use the GPU runs a triaxial test 16 times faster than 32 CPU threads, cali-brates an eleven-parameter cemented-clay model in hours, and simulates a 785,000-particleembankment in 2.5 hours.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-10-26
DOI
https://doi.org/10.5281/zenodo.23652514
Primary Topic
Geotechnical Engineering and Soil Mechanics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Granum: an agentic, GPU-accelerated discrete element program for soil

Sompote Youwai
1 links
Zenodo (CERN European Organization for Nuclear Research)
Geotechnical Engineering and Soil Mechanics
article

Granum: an agentic, GPU-accelerated discrete element program for soil

Sompote Youwai
article en

Abstract

Granum is a discrete element program for soil written in Rust, with a double-precisionCPU engine, single-precision CUDA and wgpu engines, a desktop interface built around lab-oratory tests and staged models, and a Model Context Protocol server through which an AIagent can build, run and report models. It adds to the usual spheres, clumps, Hertz–Mindlincontacts and parallel bonds a breathing-aggregate model of clay and cemented clay. Onthe impact benchmarks of Chung and Ooi (2011), both engines match Hertz’s solutions towithin 0.1% and the oblique impacts of Kratos to within 0.2◦ above 15◦. On the DEM8penetration benchmark, the CUDA engine reproduces the results of eight open-source codes,including Yade, LIGGGHTS and MercuryDPM, for beds of 25,000 to 100,000 particles in 2.3to 12.7 minutes, faster than every CPU code but two to four times slower than the dedicatedGPU codes. The tests exposed, and the program now corrects, a loss of displacement insingle-precision position updates and an inefficient grid for a large sphere among small ones.In geotechnical use the GPU runs a triaxial test 16 times faster than 32 CPU threads, cali-brates an eleven-parameter cemented-clay model in hours, and simulates a 785,000-particleembankment in 2.5 hours.

Zenodo (CERN European Organization for Nuclear Research)
Openalex Percentile: Top 19%
Geotechnical Engineering and Soil Mechanics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon