Much of the puzzling nature of quantum mechanics can be understood through one simple but profound fact: particles of the same kind are truly identical
1/9
Kimseyi memnun edemezsin. Az konuşsan gizemli, çok konuşsan geveze olursun. Sessiz kalsan burnu havada, fikrini söylesen ukala. Ne yapsan bir kulp takacak yer bulurlar zaten. O yüzden boş ver, kafana göre takıl.
Blog post: "The French have the Quantum Circuits" https://t.co/cP5hbTWl2B
André Schrottenloher just published a preprint showing how to construct quantum ECDLP circuits with costs similar to the ones in our zero knowledge proofs.
The explosion is over, but the consequences continue.
About twelve thousand years ago, a relatively normal star in the constellation Vela suddenly exploded, creating a strange point of light briefly visible to humans living near the beginning of recorded history.
The outer layers of the star crashed into the interstellar medium, driving a shock wave that is still visible today.
The featured image, taken piecemeal over 60 hours from the Khomas Region of Namibia, captures some of that filamentary and gigantic shock in visible light, with details highlighted by hydrogen (red) and oxygen (blue) emissions.
As gas flies away from the detonated star, it decays and reacts with the interstellar medium, producing light in many different colors and energy bands.
Remaining at the center of the Vela Supernova Remnant is a pulsar, a star as dense as nuclear matter that spins around more than ten times in a single second.
Image Credit & Copyright: José Mtanous
A 26-year-old Chicago real estate agent bought a box of unknown negatives at a thrift auction in 2007 for around $400.
He took it home and found thousands of street photographs taken by a French-American nanny who had carried a Rolleiflex around her neck for forty years and shown her work to no one.
She had lost the storage locker for unpaid rent. She died poor in 2009 not knowing her photographs were being seen.
Vivian Maier is now considered one of the greatest American street photographers of the 20th century.
California ground squirrels will chew on the shed skin of rattlesnakes and then lick themselves and their offspring.
This makes the rodents smell like venomous snakes to fool predators.
110 million DFT calculations: the dataset that cured a stubborn bias in materials ML
Predicting whether a new material is stable, or how it conducts heat, usually means running density functional theory (DFT): accurate quantum mechanics, but painfully slow. Machine learning interatomic potentials (MLIPs) promise to act as fast surrogates for DFT, yet they share a stubborn weakness. Most are trained on databases of relaxed, near-equilibrium structures, so when a real simulation pushes them toward distorted, far-from-equilibrium configurations, they degrade. The community even named the symptom: "systematic softening," a consistent underprediction of energies, forces, and vibrational frequencies.
Luis Barroso-Luque and coauthors attack the problem from the data side rather than the model side. They release OMat24, an open dataset of over 110 million DFT calculations spanning most of the periodic table, built deliberately around non-equilibrium structures generated through rattled Boltzmann sampling, short ab initio molecular dynamics, and re-relaxations of perturbed crystals. The whole design bet is on diversity: structures far from the comfortable energy minima where older datasets cluster.
The payoff is concrete. MLIPs pretrained on OMat24 top the Matbench-Discovery leaderboard, with F1 scores above 0.9 for stability and formation-energy errors near 18 to 20 meV/atom. More telling is that this data diversity largely eliminates systematic softening across very different model architectures, and models trained on OMat24 alone soften even less than fine-tuned variants. Within months of release, every leading model on the leaderboard had adopted the dataset.
For materials development, batteries, energy, or catalysis, the takeaway is that the limiting factor for trustworthy property prediction is often the training data, not a cleverer network. Building or reusing diverse, far-from-equilibrium data as a pretraining foundation can make screening pipelines reliable under the operating conditions that actually matter, instead of only at idealized minima.
Paper: Barroso-Luque et al., Nature Computational Science (2026) — journal license | https://t.co/6vgGBqye6S