Niklas Rindtorff

PhD student at AITHYRA with @AlexanderTong7. Feedback:

[Niklas Rindtorff @Niklas_TR]

Introducing Strong Stochastic Flow Maps

TLDR: Stochastic Flow Maps where we learn the stochastic solution path.

Work led by Sam McCallum, @zwblasingame, with Timothy Herschelll, @AlexanderTong7, and @JamesFosterBath

Arxiv: https://t.co/Hy8WWZOnjE
Code: https://t.co/PMe6RoqyZA

[Niklas Rindtorff @Niklas_TR]

@AlexiGlad @aman_gif congrats! how does this relate to https://t.co/R7PdIdgXmG ?

[Niklas Rindtorff @Niklas_TR]

@JeffreyASachs high end: https://t.co/2OUZp1GFzW
solid: https://t.co/2s50sZy8E8
low end: https://t.co/rRjepkQZaL

[Caleb Watney @calebwatney]

Surreal endnotes section to have in an official government report

[Niklas Rindtorff @Niklas_TR]

@johannes_hage congratulations!

[Austin Cheng @auhcheng]

💠Introducing Clari, a generative model to predict organic crystal structure in seconds.

Clari’s speed makes it feasible to do virtual screening of organic molecules on the basis of solid-state properties.

co-first with Alston Lo @alston1o and Luka Mucko @srboljubbosanac

[Niklas Rindtorff @Niklas_TR]

@PolitlcsUK @dieworkwear collar gap alert

[Niklas Rindtorff @Niklas_TR]

@shruthibadri @aye_kaash Are there good options for people in the EU/US to order a copy?

[Frank Noe @FrankNoeBerlin]

My superstar first PhD student @schoeneberglab Johannes who became an experimentalist just published this awesome whole cell simulation accompanied by super resolution microscopy in Cell. So proud!

https://t.co/EZa39Gt2LH

[Niklas Rindtorff @Niklas_TR]

@MaikTWehmeyer @balderton @IndexVentures @dig_ventures @tigerglobal @VisionariesVC @ycombinator congratulations!

[Niklas Rindtorff @Niklas_TR]

@dvruette @jdeschena @himanshustwts seems like they simplified the approach a bit since, leaving the original decoder's causal attention as is in the encoder. I wonder wether they chose to work with a smaller sized decoder to increase speed - could not find any details on decoder vs. encoder param count.

[Corin Wagen @CorinWagen]

maaaaan

[Niklas Rindtorff @Niklas_TR]

@FrancoisRozet @zwblasingame @AlexanderTong7 @JamesFosterBath Ty! We are fairly sure you can get rid of the g-network entirely and write down the diffusion, at least approximately, in closed form. There are a couple other aspects we're looking to improve in the ergonomics of training SSFMs in the coming weeks. cc @zwblasingame @sgmccallum

[Niklas Rindtorff @Niklas_TR]

Introducing Strong Stochastic Flow Maps

TLDR: Stochastic Flow Maps where we learn the stochastic solution path.

Work led by Sam McCallum, @zwblasingame, with Timothy Herschelll, @AlexanderTong7, and @JamesFosterBath

Arxiv: https://t.co/Hy8WWZOnjE
Code: https://t.co/PMe6RoqyZA

[Niklas Rindtorff @Niklas_TR]

@msalbergo @zwblasingame @AlexanderTong7 @JamesFosterBath I will turn this tweet into a Slack sticker

[Niklas Rindtorff @Niklas_TR]

ty! Effectively we condition on polynomial approximation of the brownian path instead of a single brownian increment. Introduced in https://t.co/fRwlvzJ5sI. This does increase the size of the inputs: The higher the fidelity of the approximation, the higher the dimensionality of the conditioning signal.

[Niklas Rindtorff @Niklas_TR]

Shoutout to @PatrickKidger's diffrax.
Tagging people that might be interested @StefanoErmon @volokuleshov @adityagrover_ @ssahoo_ @sedielem @karsten_kreis @RickyTQChen @sitanch @gottapatchemall @ZhengyangGeng @ArashVahdat @emiel_hoogeboom @TimSalimans @ValentinDeBort1 @poolio @jaschasd @nmboffi @msalbergo @peholderrieth @k_neklyudov @mmbronstein @PPotaptchik @DidiKieran @LucaEyring @HannesStaerk @GabriCorso

[Niklas Rindtorff @Niklas_TR]

This approach raises a couple new, exciting, questions: For example, can we apply this approach to learn Stochastic Flow Maps of Underdampened Langevin Dynamics - to speed up the current workhorse of molecular simulations? Can we use this approach to run protein binder design more effectively? Can we extend this approach to other consistency objectives, such as Lagrangian and Eulerian?