Lightnews — Scholar-powered news

Gerald Monard @gmonard.bsky.social · 16d

yes indeed

1

Gerald Monard @gmonard.bsky.social · 24d

Pretty sure it would translate to French Universities the same way.

Jed Brown @jedbrown.org · 24d

"Do Universities Investing In Technology Transfer Via Patenting Lose Money?"

This study reaches a conclusion I've believed to be true since seeing how tech transfer offices work. The paper calls for closing tech transfer offices and instead open sourcing all innovations.
ttb.sk/clanky/do-un...

ABSTRACT Substantial investments are made in
universities patenting new developments to pursue
a return. To gauge the impact of the holistic costs
of patenting at universities, this study provides a
new methodology for quantifying the investment
in intellectual property (IP) that includes not only
technology transfer staff costs but also direct and
opportunity faculty-related costs. It then uses the
novel methodology and publicly accessible data on
an average American research university case study.
The results found all component costs were higher
than the IP-related income, with the opportunity cost
for writing patents instead of grants being more than
33 times the income realized through IP protection.
Overall, the case study university loses over $9
million per year on IP with a negative ROI of -97.6%.
Research universities have opportunities to increase
research income >10% by ignoring IP. It is clear that
Bayh-Dole Act and similar national legislation, is
harming university economics. It can be concluded
that as generally practiced in the U.S. now, it is not
rational to continue to support university technology
transfer by patents. Instead, to improve the economic
bottom lines of universities, as well as increase the
good that research and development does for society,
universities can open source all innovations.

3 3

Gerald Monard @gmonard.bsky.social · Jul 25

I see I am not alone here...

Reposted by Gerald Monard

HGPU group @hgpu.bsky.social · Jul 20

Dissecting the NVIDIA Blackwell Architecture with Microbenchmarks

#CUDA #PTX #HPC #Performance #Benchmarking

hgpu.org?p=30053

Dissecting the NVIDIA Blackwell Architecture with Microbenchmarks

The rapid development in scientific research provides a need for more compute power, which is partly being solved by GPUs. This paper presents a microarchitectural analysis of the modern NVIDIA Bla…

hgpu.org

1 4

Reposted by Gerald Monard

Jan Řezáč @jrezac.bsky.social · Jul 15

PM6-ML, our semiempirical quantum-mechanical #CompChem method with machine learning correction (see paper: pubs.acs.org/doi/10.1021/...), is now also available as an Atomic Simulation Environment (ASE) calculator.

github.com/Honza-R/PM6-...

3 18

Reposted by Gerald Monard

Science Magazine @science.org · Jul 2

Exclusive: An international scientific competition widely credited with spurring the development of artificial intelligence for biology appears to be on its deathbed. scim.ag/44ukS90

Exclusive: Famed protein structure competition nears end as NIH grant money runs out

Agency silent on funding renewal for contest that inspired creation of AIs that predicted how proteins would fold

scim.ag

4 33 81

Reposted by Gerald Monard

Cole Group @colegroupncl.bsky.social · Jun 19

Do you run biomolecular simulations on #HPC? Rob Welch's write-up "Engineering Supercomputing Platforms for Biomolecular Applications" describes his experiences on various HPC testbeds as part of the ExaBioSim #ExCALIBUR project and is a must-read!: arxiv.org/abs/2506.15585

Engineering Supercomputing Platforms for Biomolecular Applications

A range of computational biology software (GROMACS, AMBER, NAMD, LAMMPS, OpenMM, Psi4 and RELION) was benchmarked on a representative selection of HPC hardware, including AMD EPYC 7742 CPU nodes, NVID...

arxiv.org

2 5 9

Gerald Monard @gmonard.bsky.social · Jun 9

Cornelis is back on track with Omni-Path. It's cool to see this. This article outlines the consequences of having different numbers of ports per switch in building large clusters. It's not because one switch is cheaper than the full set for a complete (non-blocking) network is!

The Register @theregister.com · Jun 9

Omni-Path is back on the AI and HPC menu in a new challenge to Nvidia's InfiniBand

After a five-year hiatus, Cornelis' interconnect returns at 400Gbps, with Ethernet support next Five years after Intel spun off its Omni-Path interconnect tech into Cornelis Networks, its 400Gbps CN5000 line of switches and NICs is finally ready to do battle with its long-time rival, Nvidia's InfiniBand.…

dlvr.it

3

Reposted by Gerald Monard

Jan H. Jensen @janhjensen.bsky.social · Jun 5

ChatGPT can now not only write, but also execute, RDKit code
www.linkedin.com/posts/nichol... #compchem

ChatGPT can now use RDKit! | Nicholas Runcie | 15 comments

🔥 Blog Post 🔥 ChatGPT now has access to RDKit. You can ask the models to write and execute cheminformatics code directly in the chat. I have briefly played with these models, and my highlights are: ...

www.linkedin.com

3 14

Reposted by Gerald Monard

Greg Landrum @greglandrum.bsky.social · Jun 1

This week I have updated and revised an old blog post showing how to perform extended Hueckel calculations with the #RDKit. This is a fun one for me because it involves work I did back in grad school. :-)
greglandrum.github.io/rdkit-blog/p...

Doing extended Hueckel calculations with the RDKit – RDKit blog

Including an exploration of charge variability across conformers

greglandrum.github.io

1 5 18

Reposted by Gerald Monard

ajdecon @ajdecon.org · May 27

Periodic offer: If you’re interested in careers in high performance computing or machine learning infra, I’m always happy to chat. Especially if you’re dealing with a recent layoff.

That’s not the same as a guarantee my thoughts will be useful! 😅 But still, happy to help where I can.

5 10

Reposted by Gerald Monard

ctdotxmt.bsky.social @ctdotxmt.bsky.social · May 21

riscv.epcc.ed.ac.uk/community/wo...

#hpc #supercomputing #sc25 #risc_v #riscv @riscv.org.web.brid.gy

International workshop on RISC-V for HPC (RISCV-HPC) at SC25

Important dates

riscv.epcc.ed.ac.uk

1 2

Reposted by Gerald Monard

HPC Guru @hpcguru.bsky.social · May 20

Microsoft Discovery helped discover a novel coolant for immersion cooling of data centers in approximately 200 hours — a process that traditionally would have taken months or years

venturebeat.com/ai/microsoft...

#AI #HPC

Microsoft just launched an AI that discovered a new chemical in 200 hours instead of years

Microsoft launches Discovery platform that uses agentic AI to compress years of scientific research into days, transforming R&D across pharmaceuticals, materials science, and semiconductor industries.

venturebeat.com

1 3

Reposted by Gerald Monard

Olexandr Isayev 🇺🇦 🇺🇸 @olexandr.bsky.social · Apr 29

Long & windy road of academic publishing! Few journal rejections and two years (!!!) after preprint, AIMNet2 paper was just published @chemsocrev.rsc.org With 69 citations to it as of now, it's immediately part of 2025 HOT🌶️ Article collection. pubs.rsc.org/en/content/a... #chemsky #compchem

3 9 51

Reposted by Gerald Monard

Chanda Prescod-Weinstein 🌌 @chanda.blacksky.app · Feb 26

Also, and I realize this is specialized information, but: the earth is round, not flat.

David Burbach 🌹 @dburbach.bsky.social · Feb 25

Kind of amazing, though not really, that DOD's #1 rocket contractor: A) doesn't understand the jet stream and B) doesn't understand closed airspace over military ranges in NV

8 44 290

Reposted by Gerald Monard

Orange Dick Tater Hater @realbiscuitspaw.bsky.social · Feb 1

430 5.8K 32K

Reposted by Gerald Monard

Liz Neeley @lizneeley.bsky.social · Feb 1

The news cycle right now - even just for science & higher ed - is overwhelming. We have to do it, but fortunately, we don’t have to do it alone.

My team is focused on tracking what’s happening to research funding, jobs & salaries, shared datasets, university responses & safety of our communities

15 270 1.4K

Reposted by Gerald Monard

Jan @dermewes.bsky.social · Dec 26

#GFN2-xTB disrupted my #compchem workflows because it's 1000x faster than #DFT and general. Unfortunately, this comes at the cost of limited accuracy, but this is a choice. Any method can have only 2 of these 3: Accuracy (few kcal/mol), speed, and/or generality (applicable across chemical space) (1)

4 7 42

Reposted by Gerald Monard

Rich Miller @techjournalism.bsky.social · Dec 9

Microsoft says its new data centers will use a closed-loop cooling system that requires no fresh water, shifting away from evaporative cooling designs. New projects in Phoenix and Mt Pleasant, Wisconsin, will pilot zero-water evaporated designs in 2026. www.microsoft.com/en-us/micros...

1 3 16

Reposted by Gerald Monard

FX Coudert @nanoporous.net · Dec 10

For colleagues who knew him, from the international community of adsorption, porous materials, thermodynamics and molecular simulation: Prof. Alain Fuchs (1953–2024) has passed away last week-end.
psl.eu/en/news/prof...

4 4 28

Gerald Monard @gmonard.bsky.social · Dec 10

Très triste nouvelle. 😢

Reposted by Gerald Monard

Andrew Jones (hpcnotes) @hpcnotes.bsky.social · Nov 27

There are 1 key things to learn about programming in C - the first is pointers - and the second is counting from 0

15 8 160

Reposted by Gerald Monard

Andy Aydın-Aitchison @andy8chi.bsky.social · Nov 21

Academics wanting to 'verify' their Bluesky profile as genuine might give an ORCID identifier in their profile and link back to their Bluesky profile from ORCID.

#AcademicSky

28 280 630

Reposted by Gerald Monard

NERSC @nersc.bsky.social · Nov 21

🏆 Congrats to the team behind the project "Breaking the Million-Electron and 1 EFLOP/s Barriers: Biomolecular-Scale Ab Initio Molecular Dynamics Using MP2 Potentials" for winning the prestigious Gordon Bell Prize at #SC24!
@nersc.bsky.social is proud to have supported this exciting work.

Scientists accepting the Gordon Bell Prize

1 6 40

Gerald Monard @gmonard.bsky.social · Nov 21

Targeted street advertising #SC24 (no wonder it's expensive)

3