Multiscale ML & Constructor Tech @ ICLR 2025
31 January, 2025
Constructor proudly supports the Machine Learning Multiscale Processes workshop at ICLR 2025 in Singapore!
The ICLR workshop is a unique and infinitely ambitious gathering that unites physics at every scale, from quantum to cosmology, keynoted by the Nobel laureate Chairman of the Constructor Group Strategic Advisory Board.
Who else aims to solve high–temperature superconductors, cancer, fusion power, and weather prediction simultaneously?
The fundamental laws of nature, the Standard Model of Physics, and its most practical aspect, quantum mechanics, are well–established.
In theory, these equations govern everything from the behavior of a hydrogen atom to Earth's climate.
However, the challenge lies in complexity. Accurately computing even a relatively simple system of 100 atoms remains beyond the reach of modern computers.
Some of the most significant scientific breakthroughs have stemmed from advances in addressing scale transitions: renormalization, density functional theory, the discovery of the Higgs boson, multiscale models for complex chemical systems, climate modeling, and protein folding.
These achievements are celebrated for their profound impact but are also specific to their respective domains and not easily transferable to other systems. We believe AI has the potential for a universal solution.
Make sure to submit your work until 10 February 2025!
If you solve scale transition, you solve science (and get a Nobel prize).
Quickly reproducible reviewable research
Peer reviewers are a crucial component of a scientific conference. Their decisions form the event program. Their feedback guides the direction of the research. Reviewers work almost entirely on pure enthusiasm and love of science. As event organizers, we greatly appreciate their time and effort, and our goal is to make the review process smooth and convenient for them.
One aspect is code and data. Traditionally, the authors just prepare an archive and send it along with the paper. The problem with this approach is that to run this code, a reviewer must set up a local environment with the correct versions of the correct libraries and data files in the correct locations. The effort is too much. At best, reviewers just look into the code when they have technical questions.
At the workshop, you will experiment with bringing the review to the next level. We give the authors access to the Constructor Research Platform. Once the deadline passes, their projects will be duplicated, anonymized, and shared with reviewers.
The result is that reviewers can run the code and analyze the data in one click inside their web browser without the need to set up or download anything, thus enchanting the review quality and reviewer experience.
If you’d like a personal demo of the Constructor Research or want to learn more about how it can accelerate your own research, don’t hesitate to get in touch.
We look forward to seeing you in April!
