Steve Sun
@Poromechanics
Followers
2K
Following
3K
Media
186
Statuses
904
Associate Professor of Civil Engineering at Columbia University. Former Sandia Research Scientist. Editor of IJNME. ML for physics and mechanics of material.
New York, USA
Joined May 2011
Thank you, @CUSEAS, for publishing the story "Understanding How Materials Behave by Watching Them Move" based on our paper in PNAS! #DIC #WARP #Differentialsimulation #inverseproblems #AI4Science
https://t.co/Lyl2DHScYU
engineering.columbia.edu
Columbia Engineers developed a machine learning framework that uses AI to infer how materials behave using inexpensive data.
0
0
4
Building models that accurately simulate how different materials behave under stress requires expensive data. In a new paper, @columbiaceem @Poromechanics describes a new approach. https://t.co/sIwX3oDVOJ
0
1
1
If you don't read enough, you risk reinventing the wheel. If you read too much, you risk paralysis. The art of balancing learning and creation is delicate, especially today, when short-term success is so easily conflated with lasting impact.
27
76
767
Hiring alert! Robotics and vis folks â take note đ
UBC Computer Science invites applications for up to two full-time tenure-track positions, with the following priority areas: visualization, robotics, reinforcement learning, data management, and data mining. Applications are due Wed Dec 10, 2025.
0
1
8
Mixing DataâDriven and PhysicsâBased Constitutive Models Using UncertaintyâDriven Phase Fields - Storm - 2025 - International Journal for Numerical Methods in Engineering - Wiley Online Library
onlinelibrary.wiley.com
There is a high interest in accelerating multiscale models using data-driven surrogate modeling techniques. Creating a large training dataset encompassing all relevant load scenarios is essential...
0
0
3
Link of the video of my Plenary lecture in COMPLAS 2025 "CONSTITUTIVE LAWS AS GENERATIVE GRAPHS AND TREES" https://t.co/BFCPzxDEKa
0
0
4
Analytical solution of Cook's membrane: https://t.co/aqmkfGUQZQ
0
0
1
I am speaking at SES Annual Technical Meeting 2025. Please check out my talk if you're attending the event! - via #Whova event app
0
1
7
Slides for plenary lecture for the 8th Biot conference. #poromechanics
linkedin.com
My plenary lecture at Biot's conference on computational poromechanics in cold regions. We would like to extend our gratitude to the Army Research Office's Earth Materials Program for the support we...
1
1
4
Discovering neural elastoplasticity from kinematic observations - PubMed https://t.co/gFozade0Vx
#MachineLearning
pubmed.ncbi.nlm.nih.gov
Inferring accurate and precise material models necessary for high-fidelity predictions has been a central challenge in constitutive modeling. Both traditional regression methods and modern machine-...
0
1
1
Our first-ever PNAS article has just been published! Congratulations to George Barkoulis Gavris for this wonderful work! #differentiablesimulations #modeldiscovery #opensource #expressivity #differentiability
https://t.co/YBk08KXiyS
pnas.org
Inferring accurate and precise material models necessary for high-fidelity predictions has been a central challenge in constitutive modeling. Both ...
1
0
11
Excited to deliver the plenary lecture for COMPLAS on Wednesday morning! See you there! https://t.co/zqmGLxjtXo
linkedin.com
Login to LinkedIn to keep in touch with people you know, share ideas, and build your career.
0
1
2
Our former associate research scientist, Prof. Yousef Heider, will join the University of Kassel as a full professor! Congratulations, Yousef! I am super excited to hear this news! https://t.co/rygy96uiWJ
uni-kassel.de
Zum 01. September wird Herr Prof. Dr.-Ing. habil. Yousef Heider die Leitung des neuen Fachgebiets KĂźnstliche Intelligenz in der Mechanik am Fachbereich Maschinenbau Ăźbernehmen. Die Dekanin Frau Prof....
0
0
2
Our latest work on mesh-based material models inferred from a domain-specific foundation model has been published by CMAME ! 90-day free access from the URL below. https://t.co/zLrJu1UOWB
0
0
1
I don't have a strong opinion about whether video models âunderstand the world.â But I do think the first bar should be checking whether you can recover consistent geometry from videoânot whether it makes accurate predictions of physics. (âAccurate physicsâ is not even
10
33
301
In a field long constrained by the rigidity of closed-form equations and the fragility of data-driven models, this paper offers a geometry-aware, data-efficient, and computationally scalable framework for modeling complex, path-dependent material behavior. 7/7
0
0
0