Anton Pershin

@StevenTobias7 Thank you very much, Steve! It is really exciting to work on this topic!

Congratulations to my former student @TonyPershin on the publication of "Training a neural network to predict dynamics it has never seen". As ever a pleasure to collaborate with you! journals.aps.org/pre/abstract/1…

New open access paper in Acta Numerica: "Mixed Precision Algorithms in Numerical Linear Algebra", with Theo Mary doi.org/10.1017/s09624…

Protons are not fundamental particles, but dynamical systems made of quarks and gluons. What a proton looks like depends on 2 quantities: the momentum at which you probe (Q^2) and the fraction of the proton momentum carried by each quark or gluon inside the proton (x) 1/6

Excited for the start of the @NewtonInstitute workshop on "Advances in geophysical and astrophysical turbulence". This is part of the programme "Mathematical aspects of turbulence: where do we stand?" newton.ac.uk/event/tur/

'Automatic differentiation to simultaneously identify #nonlinear #dynamics and....' by @eigensteve Kadierdan Kaheman and J Nathan Kutz @UW @ME_at_UW @uwengineering @uwescience hits 1000 downloads! bit.ly/3CniLFk #machinelearning #AI #noise #openscience #networks #HPC

arxiv.org/abs/2205.05513… Recent Developments in Theories of Inhomogeneous and Anisotropic Turbulence J. B. Marston and S. M. Tobias

You might not believe it, but the following 6 machine learning books are fully free: - Deep Learning - Dive into Deep Learning - Machine Learning Engineering - Python Data Science Handbook - Probabilistic Machine Learning - Machine Learning Yearning Here are their links 🧵

New video from the Applied Mathematics seminar series featuring Yohann Duguet discussing Oblique Laminar-turbulent Interfaces in Plane Channel Flows: youtube.com/watch?v=waWMWL…

We have a paper in ICLR! The title is “Learned Coarse Models for Efficient Turbulence Simulation.” We wanted to see if we could train general-purpose ML models to predict turbulent dynamics accurately at low spatial and temporal resolutions (1/n).

It turned out that the laminarization probability is a handy tool for that. So we developed a simple yet efficient method for its estimation which allowed us to find optimal parameters of oscillations suppressing transition to turbulence

Our new paper (joint with Cédric Beaume, Tom Eaves and @StevenTobias7) about optimizing the control of transition to turbulence is out @JFluidMec! Minimal seeds, edge states and laminarization probability are all discussed there together bit.ly/3xWzBeb

Lead author and @UMD_IPST graduate student Keshav Srinivasan, with Tom Antonsen, Ed Ott and Michelle Girvan, describes "Parallel Machine Learning for Forecasting the Dynamics of Complex Networks" in the current @PhysRevLett cover story.

Beyond Graph Neural Networks with PyNeuraLogic by @GustavSourek buff.ly/3IvaNwh

Supervised causal induction: In this work, we learn to induce causal structure by treating the inference process as a black box and design a neural network architecture that learns the mapping from data to graph structures via supervised training.arxiv.org/abs/2204.04875

Many in the tech industry evaluate projects in a very short-termist way -- "what am I getting out of this right now?" The really valuable game to play is the long game, over a timeline of years or even decades. And the long game is always about relationships and knowledge.

Missed our #MachineLearning Workshop last week? We covered: ☀️solar power forecasts 🌡️climate scenarios 🌧️exploring rainfall patterns 🌍🌐#DestinationEarth & #ECMWFdigitaltwins And much more! Discover some of the sessions and view replays at stories.ecmwf.int/mlws2022/ #MLWS2022

New EPSRC Postdoctoral opportunity: National Fellowships in Fluid Dynamics (NFFDy). Open to any STEM discipline related to fluid dynamics. Expressions of interest are open until 3 May. Find out how to apply: orlo.uk/FfdMu