Chang Liu

Metal is like a rope made of many strands: pull too hard and it snaps. Engineers have long sought a way to make it both strong and flexible at the same time. Typically, additives that increase strength make the metal brittle. But a team of materials scientists found a loophole: they added a small amount of nitrogen to the CrMnFeCoNi alloy, just one atom in a hundred, and got an unexpected result.

Previously, generative models for drug discovery worked like an artist painting without choosing a canvas. They learned from existing molecules and simply produced similar ones, without considering how the space of possible compounds is organized. SpaceGFN changes the approach: now chemical space is a construction set where the user specifies the parts and the assembly rules.

When a massive star runs out of fuel, it sheds its hydrogen envelope and explodes, like a final, desperate exhale. That’s exactly the kind of supernova (type Ib) astronomers observed. 62 days after the blast, in the cooling debris, they caught a faint infrared glow: it turned out to be the signature of carbon monoxide (CO).

Gold nanoparticles in liquid are like tiny living creatures: they constantly change their internal structure, as if breathing. Scientists managed to peek into this process using an electron microscope that takes snapshots every millisecond, and a neural network that cleans the images from noise.

Three satellites will drift apart in the heliocentric void, forming a triangle with sides of three million kilometers.

A match flares from both ends. First the head ignites, and a moment later the flame engulfs the wooden stick. Something similar may happen in some Type Ia supernovae — except the match is a white dwarf (a star the size of Earth), and the flame is a thermonuclear explosion.

17 supernovae from the ZTF DR2 sample lit up ahead of schedule. In the first hours and days after the explosion, they emitted more light than standard models predict. Astronomers compare this to a Bickford fuse: before the blinding flash tears across the sky, a precursor spark runs through the outer layers.

Holding in mind the exact position of a pen as it's slowly moved is a routine task for the brain. For artificial neural networks, it's nearly impossible. Standard recurrent networks like GRU and LSTM tend to break continuous motion into rigid steps: the pen doesn't glide smoothly but jumps from point to point.