Image from their link showing evidence of fabricated references
Holy smoke. What ultimately happened???
11.03.2026 19:01
π 206
π 78
π¬ 9
π 20
Image from their link showing evidence of fabricated references
Holy smoke. What ultimately happened???
with AI (or anything else), i.e. classical rentiers, billionaires, AI owners. Education will likely become off-screen, targeted to learn critical thinking, for the very few who can afford to (maybe; otherwise just parroting their clan's tacit knowledge); slop vocational tutorials for the masses.
(e.g.: mass layoffs for industry, or just too many AI-generated papers to compete in academia). Thus, we have a fast vicious circle in which the only antidote to slop is suppressed. It is a recipe for societal disaster.
The only safe positions will be those affording to be "not in competition" [..]
Here's my vision for the consequence of LLM flood. We are irreparably polluting the information wells with unfilterable bad data (verbose, sloppy, incorrect, hallucinated). That includes code. At the same time, automation will make "organic" knowledge appear inefficient, and be selected against [..]
Finally! It was about time that scientists outsourced research to AI and finally concentrate on bureaucracy
www.cnrs.fr/en/update/cn...
> [..] "a reductive view of science that we should no longer be condoning. This means we have to take action now to guarantee the quality and the ethics of the research assessment system".
Precedenti: la Spagna - anche sotto un regime fascista - ha risparmiato al suo popolo due guerre mondiali:
https://sociale.network/@peacelink/116170311336447513
#Sanchez
Take-home:
Intrinsic dimensionality provides a nonlinear, physically interpretable descriptor of conformational ensembles β complementary to RMSD, PCA, and tICA.
Useful for folding studies, CV design, and Markov state modeling.
Sequence-localized ID:
Sliding-window profiles reveal regions of coordinated flexibility and heterogeneity along the chain β mesoscale dynamical map.
Secondary-structure ID:
Helices, strands, coils show distinct dimensionality driven more by local structural context than global folding state.
Power of instantaneous ID:
In NTL9, ID detected a transient folding intermediate
(high ID + high RMSD)
β non-native but compact metastable state missed by averages.
ID vs RMSD:
Both separate folded/unfolded ensembles, but
π ID shows sharper separation with minimal overlap.
Because ID measures degrees of freedom, not deviation from a reference.
Projection matters:
β’ Distances / backbone dihedrals β global folding motions
β’ Side-chain Ο angles β local plasticity
Ο-based ID is higher in unfolded due to heterogeneous side-chain space.
Case study: villin folding trajectories.
Surprisingly: folded state shows higher ID than unfolded (for backbone/dist projections).
Why?
Compact globule supports many small-amplitude modes;
unfolded chain moves along fewer large-scale directions.
Three temporal ID metrics:
β’ Instantaneous ID (per-frame)
β’ Averaged ID (mean over time)
β’ Overall ID (global manifold)
Instantaneous ID reveals transitions & metastable states.
Three spatial analysis modes:
β’ Whole molecule
β’ Sliding sequence windows
β’ Secondary-structure segments
β Enables localization of flexibility and dynamical complexity along the protein.
Key idea: ID complements standard observables.
β’ RMSD β distance from reference
β’ PCA/tICA β linear embeddings
β’ ID β effective dimensionality of motion
It captures nonlinear manifold structure and temporal heterogeneity.
Workflow in MDIntrinsicDimension:
1οΈβ£ Load MD trajectory
2οΈβ£ Compute rigid-bodyβinvariant internal coordinates
(distances, dihedrals, etc.)
3οΈβ£ Estimate ID with modern manifold estimators
Outputs: overall ID + time-resolved ID.
Intrinsic dimension (ID) = minimal number of independent collective coordinates needed to describe the sampled conformational manifold.
Unlike raw DOF counts, ID discounts correlated or frozen motions β a compact measure of dynamical complexity.
MD simulations generate massive, high-dimensional trajectories.
But how many effective degrees of freedom does a biomolecule really explore?
π This paper introduces MDIntrinsicDimension: a Python tool to estimate the intrinsic dimension (ID) of molecular motions directly from MD.
Our paper on Intrinsic Dimension estimation for MD is out in JCIM. Tweetorial follows.
pubs.acs.org/doi/10.1021/...
Judging from the literature, chemistry, physics and engineering are now officially subfields of AI.
youtu.be/aoag03mSuXQ?...
Another amazingly researched one from veritasium. This on the Xz compromise. With thoughts on the social and security implications of open source, and a deserved homage to RMS. And an actual detailed explanation on the .so vector.
Our work presented today at the #PhysicsOfBiomolecules workshop in Bressanone sites.unimi.it/bressanone/! @sdimarco.bsky.social will show our attempts to make replica exchange molecular dynamics simulations more effective for sampling #RNA conformations
On this International Day of Women and Girls in Science I (as in previous years) take an Implicit Association Test to remind myself of my potential biases and with the goal of remembering to take them into account
You too can do this at:
implicit.harvard.edu/implicit/
![Comic. Panels up to the 10-year point are grayed out. New panels since the Ten Years comic, which chronicles the first ten years of PERSON 1's journey with cancer: (1) [two people in bed] PERSON 1 (woman): One more chapter? PERSON 2 (man): Donβt we both have to get up early? PERSON 1: Nnnnnggggh PERSON 2: Sure, good point. (2) [many people wearing masks, walking while looking at graphs on their phones] (3) [birds landing on people] PERSON 2 in beanie and scarf: Hah! They like *my* seeds best. PERSON 1 in scarf holding phone with a bird sitting on it: Wait, how do I take a picture of this one? (4) [two people rowing boats with tree landscape] (5) [Person 1 carries overflowing stack of things to Person 2 in bed] PERSON 1: I brought you honey lemon tea, more pillows, a cinnamon roll, Tylenol, another blanket, aβ PERSON 2: It was just Appendicitis, Iβm reallyβ PERSON 1: *It is my turn to take care of you and I am going to do it right!* (6) [Two people in car] (7) [still in car) PERSON 1: Oh my god. PERSON 2: Oh my god. (8) [car driving] PERSON 1: Pull over! PERSON 2: I am! (9) [both people get out of car] (10) [Large colored panel of aurora borealis over water with both people looking on] (11) [Person 1 sits against tree while Person 2 lies on the ground] PERSON 1: Fifteen years. No sign of the cancer. (12) I *am* having some weird symptoms. Joint pain. Fatigue. I think Iβm losing my close-up vision. PERSON 2: Yeah. Me too. (13) PERSON 2: I think weβre getting old. (14) PERSON 1: I guess thatβs okay. PERSON 2: Itβs all I wanted.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:cz73r7iyiqn26upot4jtjdhk/bafkreihwhjblp5lyiwp63ouar7ocfea7evmckytoeo7rkfuoo4gw5rnrra)
Comic. Panels up to the 10-year point are grayed out. New panels since the Ten Years comic, which chronicles the first ten years of PERSON 1's journey with cancer: (1) [two people in bed] PERSON 1 (woman): One more chapter? PERSON 2 (man): Donβt we both have to get up early? PERSON 1: Nnnnnggggh PERSON 2: Sure, good point. (2) [many people wearing masks, walking while looking at graphs on their phones] (3) [birds landing on people] PERSON 2 in beanie and scarf: Hah! They like *my* seeds best. PERSON 1 in scarf holding phone with a bird sitting on it: Wait, how do I take a picture of this one? (4) [two people rowing boats with tree landscape] (5) [Person 1 carries overflowing stack of things to Person 2 in bed] PERSON 1: I brought you honey lemon tea, more pillows, a cinnamon roll, Tylenol, another blanket, aβ PERSON 2: It was just Appendicitis, Iβm reallyβ PERSON 1: *It is my turn to take care of you and I am going to do it right!* (6) [Two people in car] (7) [still in car) PERSON 1: Oh my god. PERSON 2: Oh my god. (8) [car driving] PERSON 1: Pull over! PERSON 2: I am! (9) [both people get out of car] (10) [Large colored panel of aurora borealis over water with both people looking on] (11) [Person 1 sits against tree while Person 2 lies on the ground] PERSON 1: Fifteen years. No sign of the cancer. (12) I *am* having some weird symptoms. Joint pain. Fatigue. I think Iβm losing my close-up vision. PERSON 2: Yeah. Me too. (13) PERSON 2: I think weβre getting old. (14) PERSON 1: I guess thatβs okay. PERSON 2: Itβs all I wanted.
Fifteen Years
xkcd.com/3172/
chemistry coloring books
one page from the coloring book
chemistry coloring books
chemistry coloring book in the lab
As some of you may know, I draw coloring books in my spare time for kids and adults. Last year, I did a series on Chemistry Nobel Prizes. ALL 117 of them :D
Now, for a series of very good reasons: F**k Amazon !!!
You can now buy them from Lulu bookstores: www.lulu.com/search?sortB...
π’ We are #hiring!
β«οΈHead, Advanced #Genomics Technologies Core Facility
β«οΈHead, #Histopathology Core Facility
β«οΈResearch Associate on #genAI applied to #DrugDiscovery
β«οΈ Senior Spatial #Omics Technician
...& many more!
Β β‘οΈ https://bit.ly/irbjobs
#IRBJobs #ScienceJobs #JobOffer #OncoSky #AcademicSky
The deluge - or even the possibility - of LLM-written papers, written in plausible prose but actually unfounded, is definitely eroding my trust in what is published.
Will it be worth reading papers any more? Even assuming it is physically possible to pre-screen them.
I made a map of 3.4 million Bluesky users - see if you can find yourself!
bluesky-map.theo.io
I've seen some similar projects, but IMO this seems to better capture some of the fine-grained detail