Will open source go dark? Going dark != changing an open-source license; it means removing public access. In an AI model’s latent space, ideas can lose their attribution to the people who shared them. Those ideas may appear elsewhere as someone else’s “extremely smart” insights.
1
0
2
112
For decades, science and open source have relied on recognition for contributions. How can contributors get that recognition now?
1
0
1
41
Lukasz Kaczmarczyk retweeted
Fulfilling its public mission, the University of California decided to contract with Oxford University Press to make the first edition my engineering mechanics book public access. academic.oup.com/book/4888
0
1
11
593
Lukasz Kaczmarczyk retweeted
Some reflections on AI in math. I found it helpful to simply list out the various parts of what a mathematician does, why math is valuable, and where the funding comes from. From there, what happens if AI gets more capable?
youtu.be/LKrwhjOI8Nc
youtu.be/LKrwhjOI8Nc
54
303
2,541
285,062
Making Turkish Damascus :)
Playing with plasticity, cont. On the left is rotation. I use a trick from arxiv.org/html/2605.0457…: i.e. decompose rotation as a constant base plus a high-order correction; then shift the rotation so it doesn't pass 2pi, where the Hessian is singular.
Playing with plasticity, cont. On the left is rotation. I use a trick from arxiv.org/html/2605.0457…: i.e. decompose rotation as a constant base plus a high-order correction; then shift the rotation so it doesn't pass 2pi, where the Hessian is singular.
0
5
26
968
They said not to use high-order approximations for plasticity :)
For HPC, controlling the local block size is essential, but this is hard to achieve with plasticity. Is not an issue with “proximal” Galerkin. Here, log plastic strain is 3d order polynomial.
For HPC, controlling the local block size is essential, but this is hard to achieve with plasticity. Is not an issue with “proximal” Galerkin. Here, log plastic strain is 3d order polynomial.
1
1
42
1,498
Lukasz Kaczmarczyk retweeted
🏆 Announcing the winners of the 2026 MFEM Workshop Visualization Contest!
📺 Best simulation video:
Henrique Bergallo Rocha for "Plasma temperature and vorticity in a linear plasma device"
👉 More details: mfem.org/gallery
👉 All submissions: youtu.be/zptZrXhxJEU
📺 Best simulation video:
Henrique Bergallo Rocha for "Plasma temperature and vorticity in a linear plasma device"
👉 More details: mfem.org/gallery
👉 All submissions: youtu.be/zptZrXhxJEU
0
3
5
165
Ideas from statistics are finding their way into classical problems in continuum mechanics. The Bregman divergence is used in “proximal” Galerkin methods to handle unilateral constraints, such as those arising in plasticity.
1
7
85
7,050
These same ideas also allow us to define loss functions for LLMs. That is closely related to cross-entropy. But here, instead of finding the active set, we find proximal solutions for the next state using a latent variable.
youtube.com/watch?v=GlYgs6…
youtube.com/watch?v=GlYgs6…
0
0
2
287
Kiedy powstawały maszyny parowe i James Watt pracował nad ich udoskonalaniem, nikt jeszcze nie wiedział, jak działają. O tym, jak działa AI, wiemy dziś dużo więcej, niż wówczas wiedziano o termodynamice. To, że nie wiemy dokładnie, jak działa technologia, jest stanem naturalnym.
@trajektoriePL Maszyny parowe też mają dusze. Wielu się z tym zgodzi. Muszą być dostatecznie skomplikowane, tajemnicze i robić dużo hałasu. AI robi dużo hałasu.
0
0
2
230
That work is essential to provide safety evidence for power from existing AGRs and to build a future design tool and twins for HTGRs. Great to see how software tools made in @UofGEngineering are used :)
Amentum delivered two impressive presentations at the 8th EDF Nuclear Graphite Conference in Oxford. Patrick Loughnane and Samuel Primavesi presented their team’s work on the validation of the #MoFEM code for simulating structural integrity problems with nuclear graphite.
0
0
1
218
Ciekawe, jakby teraz zadać te dwa proste pytania.
"Gadające głowy". Reżyseria mistrz Kiślowski.
youtube.com/watch?v=jRXqf3…
0
0
0
188
"Gadające głowy". Reżyseria mistrz Kiślowski.
youtube.com/watch?v=jRXqf3…
youtube.com/watch?v=jRXqf3…
0
0
0
391
A new Zifeng Wang tutorial on implementing the Schrödinger equation and related topics is now live on the MoFEM webpage. @mofemjoseph
mofem.eng.gla.ac.uk/tutorial_schro…
mofem.eng.gla.ac.uk/tutorial_schro…
0
1
4
238
The trick I’m using is that instead of introducing a multiplicative decomposition, I treat plasticity as the topological evolution of an intermediate, non-conforming configuration, governed by the maximum dissipation (the thing related to the thing that makes us old).
I have a new solver for dissipative processes which, as in the movie Imperative, separates the sacrum (conservation laws) from the profanum, i.e. dissipative processes such as plasticity. The former are smooth (Newton solver); the latter are unilateral (gradient-constrained opti)
0
1
7
360
I have a new solver for dissipative processes which, as in the movie Imperative, separates the sacrum (conservation laws) from the profanum, i.e. dissipative processes such as plasticity. The former are smooth (Newton solver); the latter are unilateral (gradient-constrained opti)
Wow!! I found "Imperative" on YT. Amazing movie.
youtube.com/watch?v=TQ0ZU7…
1
1
8
996
Thus, two movies by Zanussi: Camouflage helped me become an academic, and Imperative taught me how to do mechanics. ;)
0
0
0
119
Our paper in Nano Energy;
A multiphysics model for triboelectric nanogenerator design with explicit surface roughness representation. Simulation in @mofemjoseph
sciencedirect.com/science/articl…
A multiphysics model for triboelectric nanogenerator design with explicit surface roughness representation. Simulation in @mofemjoseph
sciencedirect.com/science/articl…
0
1
0
201
Wow!! I found "Imperative" on YT. Amazing movie.
youtube.com/watch?v=TQ0ZU7…
youtube.com/watch?v=TQ0ZU7…
0
0
0
1,025