God made the volume, but the devil made the interface. Wolfgang Pauli. Robustness was in mind when designing the approach for handling damage at interfaces. Also, it should scale well. The applications are composites, but the main application is to model chip reliability.
25.01.2026 23:37
π 0
π 0
π¬ 0
π 0
Redirecting
Our new paper has been published.
Conservative data-driven finite element framework with adaptive hp-refinement for diffusion problems with material uncertainty.
doi.org/10.1016/j.cm...
15.01.2026 00:18
π 0
π 0
π¬ 0
π 0
This is for anyone who thinks they don't need to at least practice due diligence that output from an LLM is not copyright infringement or (obvious) plagiarism. (I contend it should be understood as both, even when the emitted text is not near-verbatim, because the mechanism is identical.)
08.01.2026 16:09
π 38
π 20
π¬ 3
π 0
Fun fact?
In the 1950sβ60s, the Polaris missile program (submarine-launched ballistic missiles, first (PC) computers), the adjoint method was used for optimal controlβthe same method later branded as backpropagation, now honoured with a Nobel Prize.
???
23.09.2025 13:55
π 0
π 0
π¬ 0
π 0
Seeking robustness while preserving high accuracy is a tricky business. We are on the right track, but not yet where I would like us to be. We are testing the Higham method for projecting onto a cone of symmetric positive-definite matrices.
05.08.2025 20:21
π 2
π 0
π¬ 0
π 0
Another of the above.
Griffith energy criterion, maximal crack release energy, and crack orientation through J-integral.
06.07.2025 18:44
π 1
π 0
π¬ 0
π 0
In #MoFEM (mofem.eng.gla.ac.uk), we like to break things and are not bad at this.
06.07.2025 17:53
π 4
π 0
π¬ 1
π 0
Preprint on our new paper on "Conservative data-driven finite element formulation", with MoFEM at the core of it.
arxiv.org/pdf/2506.18206
24.06.2025 09:42
π 1
π 0
π¬ 0
π 0
Today was perfect weather for a swim in Loch Lomond. It was my second time, but I am already addicted to open water swimming.
21.06.2025 17:47
π 2
π 0
π¬ 0
π 0
Exascale computing for research and the implications of quantum computing, AI and Net Zero
YouTube video by The Foundation for Science and Technology
Edinburgh event: #Exascale computing for research and the implications of quantum computing, AI and Net Zero
youtube.com/watch?v=DzbC...
#HPC #AI via @likask.bsky.social
31.05.2025 10:33
π 10
π 6
π¬ 0
π 0
First MoFEM work in Ukrainian on a block solver for GPU. :)
31.05.2025 13:40
π 1
π 0
π¬ 0
π 0
Exascale computing for research and the implications of quantum computing, AI and Net Zero
YouTube video by The Foundation for Science and Technology
Interesting discussions on HPC, AI, and computing. From my perspective, we often solve not the problems we want, but the ones we can. Also, it makes no sense to build infrastructure solely for AI; the same infrastructure can be used for other things.
youtu.be/DzbC4KG3XwU?...
31.05.2025 07:11
π 3
π 1
π¬ 0
π 0
This year, we are introducing a robust hybrid mixed solver, tailored for large-scale applications and emerging hardware architectures. This will complement the high-fidelity MoFEM solver with implicit integration of crack front evolution.
05.05.2025 11:33
π 1
π 0
π¬ 0
π 0
We have the pleasure of running a workshop on modelling cracks in MoFEM for our colleagues from EDF and Amentum from 8 to 9 May. We have been working together on this topic for over 16 years.
05.05.2025 11:33
π 2
π 0
π¬ 1
π 0
My talk at mfem.org/seminar (Lawrence Livermore National Laboratory) on the Mixed Finite Element formulation and MoFEM. Thanks to @tzanio.bsky.social
for the invitation. It was a great pleasure.
16.04.2025 14:31
π 5
π 1
π¬ 1
π 0
π
The next talk in our πππ @πππ‘π seminar series will be on Tuesday, April 8th, 9:00-10:30am PDT.
π Dr. Εukasz Kaczmarczyk (@likask.bsky.social) from University of Glasgow will present: "Mixed finite element formulation for solid mechanics problems".
π Sign up and learn more at: mfem.org/seminar.
29.03.2025 20:39
π 0
π 1
π¬ 0
π 1
MoFEM accoring OpenHub. There are 53 developers, but according to the estimate, it will take a single professional programmer 73 years to write our code!
openhub.net/p/mofem
28.03.2025 18:29
π 0
π 0
π¬ 0
π 0
Glasgow University can be also brutal.
21.03.2025 09:51
π 2
π 0
π¬ 0
π 0
Jak kaΕΌΔ
dziΔkowaΔ za kupienie komercyjnej usΕugi, to za pomoc bΔdΔ
ΕΌΔ
daΔ caΕej naszej miedzi? "Prawdziwi" patrioci sΔ
OK z tym?
10.03.2025 11:04
π 0
π 0
π¬ 0
π 0
Let's pray, mechanics. Mechanics have one dogma, and that is objectivity.
Before the lecture in the old church.
06.03.2025 16:08
π 6
π 1
π¬ 1
π 0
BTW Bandwith for L2 and H-div can be a priori determined, and does not depend on the mesh. Made in #MoFEM
26.02.2025 20:07
π 1
π 1
π¬ 0
π 0
Made in #MoFEM
26.02.2025 20:04
π 0
π 0
π¬ 0
π 0
Energetic spaces. The same mesh, 4th order, different space. You can observe how space (not the base itself) renders matrix structure and sparsity.
26.02.2025 20:03
π 3
π 0
π¬ 1
π 1
Keir Starmer, "Nigel Farage didn't even turn up to parliament today"
"Nigel Farage is fawning over Putin"
"That's not patriotism"
25.02.2025 19:16
π 638
π 149
π¬ 23
π 6
SΕucham poranka Tok FM z WrΓ³blem. Parfianowicz mΓ³wi, ΕΌe Rosjanie planujΔ
na sto lat do przodu ;) OgΓ³lnie ekspertΓ³w cechujΔ
zawsze mocne opinie. Btw WyciΔte z kontekstu, ale tak widaΔ ostrzej.
18.02.2025 09:20
π 0
π 0
π¬ 0
π 0
mofem.eng.gla.ac.uk/mofem/html/r...
16.02.2025 12:47
π 3
π 1
π¬ 0
π 0
Putin wanted to stop NATO from expanding, and it ended up with Sweden and Finland joining it. It would be funny if Trump wanted to annex Canada, and it ended up joining the EU.
15.02.2025 00:45
π 3
π 0
π¬ 0
π 0
π£π£ LAST FEW DAYS REMAINING to apply for fully funded PhDs for September 2025 Entry via the ExaGEO DLA ππ»π§ π§βπ www.exageo.org
UG and PG degree holders are welcome to apply!
13.02.2025 09:45
π 1
π 2
π¬ 0
π 1