UMBC ATOMS Lab

AI & Theory-Oriented Molecular Science Lab at @UMBC ⚛️ directed by @trjosephson

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

And here’s what it looks like to formally prove (from the axioms of mathematics!) that the kinematic equations follow from calculus-based definitions of motion. A neat finding: change the type of “time” from real to complex, and ¾ kinematic equations hold. (6/⚛️)

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

We also show how these features of Lean can be used to define, prove, and interoperate *scientific* concepts. We formally define “thermodynamic system” and “ideal gas” using Lean structures, then prove that several gas laws are satisfied by the ideal gas. (5/⚛️)

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Our fanciest proof is BET adsorption theory, which invokes infinite geometric series. In Lean, theorems and proof tactics can be imported from the mathlib database, which has >100k formally proved theorems, all of which are proved from the fundamental axioms of math. (3/⚛️)

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Here’s what it looks like to translate Langmuir’s adsorption theory into a machine-verified math proof. The premises must be spelled out precisely, and Lean catches all the mathematical edge cases (can’t divide by zero, so need to add extra constraints to our premises). (2/⚛️)

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Learn more about our lab at https://t.co/ceeaC4fXbI and follow us at @ATOMSresearch

[UMBC ATOMS Lab @ATOMSresearch]

Congratulations to Max, @ssharlin_ , Parivash, and Catherine on the manuscript on Chemical Theorem Proving in Lean!

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Our first manuscript from the ATOMS Lab! "Formalizing Chemical Theory using the Lean Theorem Prover." As mathematicians use theorem provers to formally verify math proofs, we can do the same for theories in the chemical sciences. https://t.co/QeHyje7RSG (1/⚛️)

[UMBC ATOMS Lab @ATOMSresearch]

Congratulations @RodAndre160038 on completing your REU project on modeling 1,4-dioxane adsorption. Thank you for your hard work and we are glad to have you on the team this summer!

[UMBC ATOMS Lab @ATOMSresearch]

Lab Picnic 081122

[UMBC ATOMS Lab @ATOMSresearch]

Congratulations @CharlieFoxML on his Master's thesis defense and Parivash on passing the PhD qualifying exam!

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Oh my goodness, just delighted that we got our first grant! "Simulation methods for competitive adsorption in Brønsted acidic zeolites" Thank you, @NSF! https://t.co/ht2EPZQGt0 @UMBC_CBEE @ATOMSresearch

[Volker Deringer @vl_deringer]

Meta-learning interatomic potentials: using an accurate #ML model to train another, much faster one, for accelerated #compchem simulations. Very excited to share this preprint describing work by @JoeMorrow3594, now available on @arxiv (https://t.co/kZRnzDHrRJ)! (1/4)

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Excited to be at #AIChEAnnual!
Tomorrow, my student @Fran_Nacion presents a poster "Generating Adsorption Equations Using Symbolic Regression" from 10:00-12:30 https://t.co/mm1Gt0A6cb

[Francesca (Nikki) Nacion @Fran_Nacion]

I am honored to have presented my first poster at the Annual American Institute of Chemical Engineers Student Conference. I couldn’t ask for a better mentor Dr. Tyler Josephson and a more supportive @UMBC_CBEE department! @umbcAIChE @ChEnected @trjosephson #AICHE2021

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Introducing ODEs in Python with a Lokta-Volterra model of predator-prey interactions😆

[UMBC ATOMS Lab @ATOMSresearch]

Getting into the Halloween spirit with the one and only: @trjosephson ! 🦖

[Tyler R. Josephson (atomslab.bsky.social) @trjosephson]

Introducing our new group website! Check it out! (and follow our student-run group Twitter account @ATOMSresearch)

[UMBC ATOMS Lab @ATOMSresearch]

The ATOMS Lab ⚛️ website is finally up and running! Come check it out: https://t.co/XgZar3QKC8

[CompChem Highlights @CCHighlights]

Bayesian optimization of nanoporous materials
https://t.co/jlJ1XjS1vd #compchem