To meet the challenge posed by rapidly growing datasets from surveys like DESI and Euclid, researchers from INAF, the University of Parma, and the University of Waterloo have developed Effort.jl, a new emulator that reproduces EFTofLSS outputs in a fraction of the time. Running on a laptop in minutes instead of demanding a supercomputer, Effort.jl preserves the accuracy of the full model, and in some cases even restores details usually trimmed to save computing resources.

"Imagine wanting to study the contents of a glass of water at the level of its microscopic components, the individual atoms, or even smaller: in theory you can. But if we wanted to describe in detail what happens when the water moves, the explosive growth of the required calculations makes it practically impossible," explains Marco Bonici, University of Waterloo researcher and first author of the study. "However, you can encode certain properties at the microscopic level and see their effect at the macroscopic level... This is what an effective field theory does."

Effort.jl trains a neural network to mimic the model's predictions, but with an innovative twist: it incorporates existing theoretical knowledge and parameter gradients into its learning process. This allows the emulator to achieve high precision using far fewer training examples, dramatically cutting computational costs.

Extensive validation on both simulated and real data shows Effort.jl's predictions closely match EFTofLSS, giving scientists confidence in its reliability. With its efficiency and precision, Effort.jl is set to become an essential tool for analyzing the vast astronomical datasets now arriving, enabling faster exploration of the Universe's large-scale architecture.

Research Report:Effort.jl: a fast and differentiable emulator for the Effective Field Theory of the Large Scale Structure of the Universe

Related Links
Sissa Medialab
Understanding Time and Space