
Assistant Professor (Research) at @BrownUniversity Physics Informed ML | Travelling, Painting & Blogging https://xitter.cf/t.co/ZZNtEQ6ITV
Providence, USA
Joined May 2022
- Tweets19
- Following39
- Followers36
- Likes9
Somdatta Goswami retweeted
Somdatta Goswami retweeted
I am looking for 4 qualified PhD students in the various areas of mechanics and materials. Send your CV and one-page statement if you fit the requirements. @UUtah @UtahMech #mechanics @EPoromechanics @InterPoreTweets @AGU_GWHydro @FRC_IACM_Com @IACM_Community
Somdatta Goswami retweeted
I am happy to present our team's work on the role of EGFR in drug induced liver injury at the University of Pittsburgh, Pittsburgh Liver Research Center retreat tomorrow at 5.00 PM. Looking forward to interact with the attendees. @PLRC_PittLiver #plrc #pittsburgh
Somdatta Goswami retweeted
Exciting news for JHU CaSE and this research team led by @SURGroup aiming to advance scientific ML!
An interdisciplinary team of Hopkins Engineers led by @JHUCaSE's Michael Shields has received a @doescience award to research scientific machine learning focused on greater predictive capabilities for scientific simulations. engineering.jhu.edu/news/fac…
Somdatta Goswami retweeted
Replying to @argonne_lcf
@argonne_lcf Thanks a lot for this amazing two-week training project! I was so happy to be here, and I have definitely learned a lot of HPC techniques! Most importantly, I've met some wonderful people here! I feel very fortunate to have met you! @sgoswami26 @itshelenxu
Somdatta Goswami retweeted
The 2022 @argonne Training Program on Extreme-Scale Computing comes to a close today. With its 10th year now in the books, nearly 700 participants have attended ATPESC to learn how to use the world's most powerful supercomputers for scientific research.
extremecomputingtraining.anl…
Somdatta Goswami retweeted
ATPESC is underway! Celebrating its 10th anniversary, the two-week training course provides researchers the opportunity to learn the latest supercomputing tools and techniques from some of the world's leading #HPC experts.
Somdatta Goswami retweeted
Physics-Informed Neural Operators
deepai.org/publication/physi…
by @sgoswami26 et al.
#NeuralNetwork #PartialDifferentialEquation
Booking.com gave my husband and me a nightmare experience. I can now plainly discern a state of deterrence and deliberate harassment of travelers.
My experience buying a used car in the USA (1/3)
thetalkingsense.com/f/buying…
#BuyUsedCar #Learn2Drive
Somdatta Goswami retweeted
Neural operator learning of heterogeneous mechanobiological insults contributing to aortic aneurysms
deepai.org/publication/neura…
by @sgoswami26 et al.
#ComputerScience #Learning
A robust monolithic solver for phase-field modeling of fracture.
Check the work on: arxiv.org/abs/2111.10649
#isogeometricAnalysis #phasefield #brittlefracture
Spring is sooner recognized by flora than by us!
#SpringVibes #BrownUniversity
Have you quantified over-parameterization in surrogate models?
deepai.org/publication/on-th…
@kontolati and I have presented a detailed quantification of the effect of over-parametrization in manifold-based surrogates and deep operator networks.

ALT Focus on the x-axis of the plots. For manifold base surrogate (m-PCE): modest over-parameterization leads to better generalization, both within and outside of distribution, whereas aggressive over-parameterization leads to over-fitting. For Deep Operator Network (DeepONet): aggressive over-parameterized leads to better generalization.
Which surrogate model will give more accuracy when you have less training data than when you have a large training dataset? deepai.org/publication/on-th…
@kontolati and I have demonstrated the effect of training dataset size on the generalization accuracy of #mPCE, #DeepONet, and #FNO

























