Algorithms, Tools and Methods Labs

The cross-domain Algorithms, Tools and Methods Labs (ATMLs) are interdisciplinary teams of scientists performing research and development on and providing training in generic, cross-domain areas. This also includes the development of original algorithms, tools and methods. Typically, an ATML addresses overarching topics such as numerical algorithms, machine learning or parallel performance tools, that are key to enabling computational & data-intensive science and for transforming data into knowledge. It may often work closely with the various Simulation and Data Labs (SDLs) but also has direct user contacts of its own. As for the SDLs, the work program of a typical ATML expert is split between research and support; however, education and training usually play a more central and visible role. Currently, JSC boasts 9 ATMLs, including Visualization, HPC Parallel Performance, Application Optimization, Accelerating Devices, Concurrency and Parallelism, Numerical and Statistical Methods, Scalable Machine Learning and Multi-Purpose Artificial Intelligence, Data Management and Analytics, Federation Technologies & Services.

ATML Numerical and Statistical Methods

ATML Parallel Performance

ATML Application Optimization and User Service Tools

ATML Visualization & Interactive HPC

ATML Deep Learning

SDL Applied Machine Learning

ATML Accelerating Devices

ATML Federation Technologies & Services

ATML Concurrency and Parallelism

ATML Advanced Time Integrators

ATML Data Management and Analytics