Diptarka Hait

Thrilled to start my independent career as an Asst. Prof. @ChemColumbia & an ARS @FlatironInst. We'll develop new methods for modeling excited states, with an eye on practical photocatalysis. Other areas of quantum #compchem are also of interest, stay tuned for more!

GPU accelerated density fitting CASSCF from Ruiyan @mtzlab published at @JCIM_JCTC. Time for high-throughput, high-accuracy multireference simulations of ground and excited states! Was a privilege to contribute. #compchem pubs.acs.org/doi/10.1021/ac…

@StacyMalaker @YaleChem Congratulations!

Five years ago, I shared a picture of CRB on my very first day at @YaleChem. Today, almost exactly 5 years later I'm walking into the same building with a new title (and a lot of construction)! I’m thrilled to share that I am officially an Associate Professor with tenure!! 🥂🎉

Yale Chemistry invites applications for tenure-track assistant professor positions in the following research areas: chemical biology, energy sciences, and physical or theoretical chemistry, to start July 1. Consider applying or sharing with your networks: bit.ly/48xZzX2.

@KCarsch Congratulations!

We're looking forward to working on alkylation chemistry, supported by the ACS Petroleum Research Fund! cen.acs.org/acs-news/progr…

Please consider applying if it seems like a good fit or forward to someone who you think might be interested. I'm happy to chat about the institute, initiative, or my own research interests (electronic structure, photocatalysis, homogenous catalysis, etc.)

Please check out the initiative's webpage (simonsfoundation.org/flatiron/initi…) to learn more about us and our research, as well as the interfolio posting (apply.interfolio.com/173399) for further details. The deadline to apply is Dec 5 2025.

The Initiative for Computational Catalysis @FlatironInst invites applications at the postdoctoral level for Flatiron Research Fellowships. Fellows work alongside research scientists to develop new methods & software for studying electronic structure & dynamics of catalysis.

Team: @_jdep_ , Martin, @toddjmartinez & me. Code: github.com/mtzgroup/mlp_g… Data: zenodo.org/records/170429… Paper: pubs.acs.org/doi/10.1021/ac… @JCIM_JCTC #compchem

Geodesics on machine learned PES therefore offers a cheap & reliable route to finding TSs on ab initio PES. We invite the community to apply geodesic construction for finding transition states and would be curious to know what difficulties/limitations are encountered.

Identifying transition states often requires hundreds/thousands of ab initio calculations. We report a more efficient route by finding the shortest path (“geodesic”) between reactant and product on a machine-learned potential energy surface, without any ab initio calculations.

This has been a challenging problem that vexed me for a while. Congratulations to Richard (@mhg_group) and Leo (@FlatironInst) for leading this interesting work that opens up lots of fun applications (and challenges for even heavier elements). pubs.acs.org/doi/full/10.10…

Predicting accurate XPS for shells with nonzero angular momentum is challenging because of spin-orbit coupling induced splitting. We report a combination of OODFT and NOCI in @JCIM_JCTC (link below) that is accurate to ~0.2 eV vs expt. for L-edge of 3rd period elements.#compchem

@myphysicsvoyage Congratulations!

Thank you!!

@quekpottheories Congratulations!

Thrilled to share that I start @EPFL @ICepfl this week as an Asst Prof of Computer Science and Physics! We'll work on quantum algorithms and learning theory, and probe the computational limits of observing physical phenomena. 💻⚛️Towards a physical theory of computing!

@chenru_duan Thanks, Chenru! I saw your recent React-OT paper but still need to fully internalize all the math therein. Applications geodesics to optimal transport would be very interesting, please let me know if you discover anything along these lines.

@fullCIalchemist This is quite interesting finding! Seems very relevant to how we can set up optimal transport as well.

Locating transition states is often a #compchem challenge. We find that geodesics (shortest length paths) on ML PES can yield excellent initial geometries for transition state optimization on DFT PES (even compared to pure DFT frozen string method/FSM!). Preprint below!

Code and data at doi.org/10.5281/zenodo…. Hopefully it proves useful to folks in their research!

Geodesic construction on MLPs is thus a cheap and effective way to initiate DFT transition state optimizations. arxiv.org/abs/2507.17968 is the associated preprint (with lots more context). This is my last Stanford project, together with @_jdep_, Martin Stöhr & @toddjmartinez.