Toggle light / dark theme

The electrically readable complex dynamics of robust and scalable magnetic tunnel junctions (MTJs) offer promising opportunities for advancing neuromorphic computing. In this work, we present an MTJ design with a free layer and two polarizers capable of computing the sigmoidal activation function and its gradient at the device level. This design enables both feedforward and backpropagation computations within a single device, extending neuromorphic computing frameworks previously explored in the literature by introducing the ability to perform backpropagation directly in hardware. Our algorithm implementation reveals two key findings: (i) the small discrepancies between the MTJ-generated curves and the exact software-generated curves have a negligible impact on the performance of the backpropagation algorithm, (ii) the device implementation is highly robust to inter-device variation and noise, and (iii) the proposed method effectively supports transfer learning and knowledge distillation. To demonstrate this, we evaluated the performance of an edge computing network using weights from a software-trained model implemented with our MTJ design. The results show a minimal loss of accuracy of only 0.4% for the Fashion MNIST dataset and 1.7% for the CIFAR-100 dataset compared to the original software implementation. These results highlight the potential of our MTJ design for compact, hardware-based neural networks in edge computing applications, particularly for transfer learning.

In just over three years since its launch, NASA’s James Webb Space Telescope (JWST) has generated significant and unprecedented insights into the far reaches of space, and a new study by a Kansas State University researcher provides one of the simplest and most puzzling observations of the deep universe yet.

In images of the deep universe taken by the James Webb Space Telescope Advanced Deep Extragalactic Survey, the vast majority of the galaxies rotate in the same direction, according to research by Lior Shamir, associate professor of computer science at the Carl R. Ice College of Engineering. About two thirds of the galaxies rotate clockwise, while just about a third of the galaxies rotate counterclockwise.

The study— published in Monthly Notices of the Royal Astronomical Society —was done with 263 galaxies in the JADES field that were clear enough to identify their direction of rotation.

A Kansas State University engineer recently published results from an observational study in support of a century-old theory that directly challenges the validity of the Big Bang theory.

Lior Shamir, associate professor of computer science, used imaging from a trio of telescopes and more than 30,000 galaxies to measure the redshift of galaxies based on their distance from Earth. Redshift is the change in the frequency of waves that a galaxy emits, which use to gauge a galaxy’s speed.

Shamir’s findings lend support to the century-old “tired light” theory instead of the Big Bang. The findings are published in the journal Particles.

It is a deep question, from deep in our history: when did human language as we know it emerge? A new survey of genomic evidence suggests our unique language capacity was present at least 135,000 years ago. Subsequently, language might have entered social use 100,000 years ago.

Our species, Homo sapiens, is about 230,000 years old. Estimates of when language originated vary widely, based on different forms of evidence, from fossils to cultural artifacts. The authors of the new analysis took a different approach. They reasoned that since all human languages likely have a —as the researchers strongly think—the key question is how far back in time regional groups began spreading around the world.

“The logic is very simple,” says Shigeru Miyagawa, an MIT professor and co-author of a new paper summarizing the results.

A team of quantum computer researchers at quantum computer maker D-Wave, working with an international team of physicists and engineers, is claiming that its latest quantum processor has been used to run a quantum simulation faster than could be done with a classical computer.

In their paper published in the journal Science, the group describes how they ran a quantum version of a mathematical approximation regarding how matter behaves when it changes states, such as from a gas to a liquid—in a way that they claim would be nearly impossible to conduct on a traditional computer.

Over the past several years, D-Wave has been working on developing quantum annealers, which are a subtype of quantum computer created to solve very specific types of problems. Notably, landmark claims made by researchers at the company have at times been met with skepticism by others in the field.

Light-emitting diodes (LEDs) are widely used electroluminescent devices that emit light in response to an applied electric voltage. These devices are central components of various electronic and optoelectronic technologies, including displays, sensors and communication systems.

Over the past decades, some engineers have been developing alternative LEDs known as quantum LEDs (QLEDs), which utilize (i.e., nm-size semiconducting particles) as light-emitting components instead of conventional semiconductors. Compared to traditional LEDs, these quantum dot-based devices could achieve better energy-efficiencies and operational stabilities.

Despite their potential, most QLEDs developed so far have been found to have significantly slower response speeds than typical LEDs using inorganic III-V semiconductors. In other words, they are known to take a longer time to emit light in response to an applied electrical voltage.

There are moments in the history of human thought when a simple realization transforms our understanding of reality. A moment when chaos reveals itself as structure, when disorder folds into meaning, and when what seemed like an arbitrary universe unveils itself as a system governed by hidden symmetries.

The Bekenstein bound was one such revelation—an idea that whispered to us that entropy, information and gravity are not separate but rather deeply intertwined aspects of the cosmos. Jacob Bekenstein, in one of the most profound insights of modern physics, proposed that the entropy of any physical system is not limitless; it is constrained by its energy and the smallest sphere that can enclose it.

This revelation was radical: Entropy—long regarded as an abstract measure of disorder—was, in fact, a quantity deeply bound to the fabric of space and time. His bound, expressed in its simplest form, suggested that the total information that could be stored in a region of space was proportional to its energy and its size.

Not long to go now: After more than nine months on the International Space Station, two astronauts are a step closer to returning home following the launch of a crew swap mission on Friday.

A Falcon 9 rocket with a Crew Dragon fixed to its top blasted off from the Kennedy Space Center in Florida at 7:03 pm (2303 GMT), carrying a four-member team bound for the orbital outpost.

“We celebrate the countless individuals all over the world that have made this journey possible,” said astronaut Nichole Ayers, the designated pilot of the Crew-10 mission, just before launch.

Most cells in the body send out little messengers called extracellular vesicles that carry proteins, lipids, and other bioactive molecules to other cells, playing an important role in intercellular communication. But healthy cells are not the only ones that rely on extracellular vesicles. Cancer cells do, too. Small extracellular vesicles that are shed from tumor cells contribute to how cancer spreads to healthy tissue.

These small messengers could be a key to developing new cancer-fighting drugs and therapies, but it has been unclear how exactly the recipient cells absorb the extracellular vesicles and their cargo. Recent research used state-of-the-art imaging to observe the uptake of tumor-derived small extracellular vesicles by target cells. The results were published in Nature Communications on March 12, 2025.

“In recent years, extracellular vesicles have attracted attention as a carrier of intercellular signaling. However, the mechanism of their internalization by target cells has not been well understood. We wanted to elucidate the pathway and mechanism of internalization of extracellular vesicles by target cells,” said Kenichi G. N. Suzuki, a professor at the Institute for Glyco-core Research at Gifu University in Gifu and a chief at the Division of Advanced Bioimaging, National Cancer Center Research Institute in Tokyo, Japan.

Does autoimmunity underlie minimal change disease?

Tobias B. Huber, Nicola M. Tomas & team report a direct pathogenic role of anti-nephrin autoantibodies in the development of podocytopathy with a minimal change disease phenotype:

The electron microscopy image shows moderate podocyte foot process effacement (without electron-dense deposits) in the anti-nephrin rabbit.


Address correspondence to: Tobias B. Huber or Nicola M. Tomas, III. Department of Medicine, University Medical Center Hamburg-Eppendorf, Martinistrasse 52, D-20246 Hamburg, Germany. Phone: 49.40.7410.53908; Email: [email protected] (TBH); [email protected] (NMT).