Computational research
SIML, simulation, and optimization: understanding how systems learn, adapt, and make decisions.
Explore this workDeep SIML Labs
We develop simulation and optimization methods, and we’re working on physical implementations and new materials.
Less data to move and store creates an opportunity for lower energy use. Studio turns your existing enums into compact, verifiable payloads and code you can build with.
Rust · Python · C++ · Email sign-in · Invited beta
Explore what Studio doesOur work
Three connected areas of research, with a shared aim: learn more from every experiment.
SIML, simulation, and optimization: understanding how systems learn, adapt, and make decisions.
Explore this workBuilding physical implementations and connecting models with measurement and experiment.
Explore this workInvestigating material behavior, sensing, and actuation through modelling and physical research.
Explore this workSelected research
Explore our latest computational studies, methods, and research prototypes.
On one NVIDIA RTX 3060, SIML CUDA advanced 8,100,736 interacting cells and Triton advanced 6,022,368 under the same 6 GB memory limit in an MCF-7 signaling benchmark.
Read the studyA controlled binary lattice study on one RTX 3060: 12.08× faster resident updates than a tuned UInt8 CUDA control at the reported workload, with sustained population tests and a small-workload loss.
Read the studyResearch together
We welcome collaborators working on learning systems, physical implementations, and materials.