In our daily life, we perceive many different events around us with different participants involved, such as a woman dancing in a theater, or a man enjoying an ice cream. To produce and understand the sentences that express these events, it is crucial to identify the number of participants involved and the role that each one plays. As in a drama, we have to identify the role that each character has to understand the story.

Agents, patients and experiencers

In some events, there are two characters involved, as in the girl kicks the ball, with the girl and the ball; in others, instead, there is only one, as in the woman dances or the man falls down, with the woman or the man. Each of these participants has a different role. For example, the woman is an agent, because she is doing the action, just the opposite of the man, who is a patient, because he is the one affected. In the case of the girl has fun, this participant has the role of experiencer, the one who experiences the enjoyment.

In psycholinguistics, the mechanisms involved in the processing of agent, patient, and experiencer roles have been investigated through different experiments such as reading sentences while participants’ brain activity or eye-movements are recorded. In these studies, it has been found that agent and patient categories play a crucial role in language comprehension and production. These two roles are used when reading, listening to, and producing sentences, and different mechanisms are involved in their processing (e.g., ¹²³⁴⁵). Therefore, they form two different categories within the event role repertoire ⁶. They are two different characters within the drama. However, as I will discuss, there is no experimental evidence for a third character, the experiencer role.

One well-established result regarding the processing of agents and patients comes from experiments where participants are asked to read sentences while their brain activity is recorded by means of electrodes, known as EEG studies. Results of these studies show that the processing of patients elicited an N400 component (a negative brain response that occurs around 400 milliseconds after a stimulus is processed) compared to the processing of agents. This effect is associated with the processing of unexpected elements. For example, when processing the day was breezy so the boy went outside to fly, it is expected to read next a kite based on the preceding context; hence, if an airplane is read instead, as this is unexpected, then an N400 effect appears ⁷. Something similar happens with the processing of agents and patients: agents are expected, whereas patients unexpected.

Across languages, speakers tend to interpret the first ambiguous noun phrase of a sentence as an agent instead of as a patient, due to the agent preference ⁸: agents are preferred and are cognitively more salient than patients. Imagine that someone reads the sentence Mary falls down. Before encountering the verb, the reader cannot know whether Mary is an agent or a patient. However, as agents are preferred over patients, the reader interprets that it is an agent by default. Once the verb is reached, there are two options: confirm that Mary is an agent (the expected role) or recategorize it as a patient (the unexpected role). In Mary falls down, it is the second option, Mary is categorized as a patient, because the verb selects this role. This recategorization is reflected in an N400 component. This result shows that agents and patients have different processing mechanisms and that they form two different categories. Furthermore, this effect has been largely found in different languages such as, Basque (Isasi-Isasmendi et al., 2024), Spanish ⁹, German (Haupt et al., 2008), Hindi (Bickel et al., 2015), or Aïwoo (Sauppe et al., 2023), supporting its robustness.

Therefore, these experimental findings show that there is a preference toward agents over patients. The agent is the protagonist of the show, while the patient plays a secondary role.

What about experiencers?

Linguists have proposed that there are other roles within the event role repertoire, for example, the experiencer role ¹⁰.This role corresponds to the participant who experiences a mental state denoted by the verb. Participants of events like loving someone or enjoying an ice cream have this role. The experiencer is the participant who loves or enjoys.

But does experimental evidence show that the experiencer role is a specific role category within the event role repertoire, as agent and patient roles do?

In Sánchez-López (2026), I addressed this question. I investigated whether speakers use a third category, the experiencer role, when processing language. In three reading eye-tracking experiments and in one EEG experiment, I asked Spanish native speakers to read sentences, while their eye-movements and their brain activity were recorded.

In the first two eye-tracking experiments, I compared de processing of sentences with two participants with different combinations of agent, experiencer, and patient roles: (i) María golpea a Juan “Mary hits John”, with an agent (María) and a patient (Juan); (ii) María asusta a Juan “Mary frightens John”, with an agent (María) and an experiencer (Juan); and (iii) María ama a Juan “Mary loves John”, with an experiencer (María) and a patient (Juan).

These two experiments revealed that Spanish native speakers produced higher fixation times on verbs selecting agent-experiencer roles (ii), than on the other two sentence types (i, iii), independently of whether the subject was animate or inanimate (e.g., María/La situación asusta a Juan “Mary/The situation frightens John”). Interestingly, no processing differences were found between sentences selecting agent-patient (i) and experiencer-patient (iii). These findings show that the combination of agent-experiencer involves higher processing cost than other combinations.

But why is the agent-experiencer combination costlier than the others?

A possible explanation could be that the experiencer role is not a specific category within the event role repertoire. As Dowty ¹¹ proposed, agents and experiencers are both grouped within the same category, the one of the proto-agent. Based on this proposal, there are only two roles in the event role repertoire, proto-agent and proto-patient, and proto-agents always become subjects. Hence, in sentences like Mary frightens John, both participants are proto-agents. Since they are both possible candidates to become subject, a competition for subject assignment appears. This competition is reflected in higher processing cost compared to those sentences without competition, sentences with only one proto-agent participant (e.g., Mary hits John or Mary loves John).

To better understand this competition effect, imagine that the proto-agent is the hero of the story and the proto-patient is the villain. In a drama, if there is a hero, there is also a villain. If there are two heroes and no villain in the story, it would be unusual, and probably, it would take more time to understand the story, since this is not expected. Hence, the same happens with sentences like Mary frightens John, which have two proto-agents; this combination is unexpected and it entails higher processing cost than when there is only one proto-agent and one proto-patient.

In another reading experiment using both eye-tracking and EEG techniques, I compared the processing of the three roles, agent, experiencer, and patient again to determine whether this experiencer role behaves like the agent role or not. In these experiments, I used sentences with only one participant involved: (iv) María baila “Mary dances” with an agent, (v) María disfruta “Mary has fun” with an experiencer, and (vi) María se cae “Mary falls down” with a patient. The use of these sentences allowed a more direct comparison of the processing of these roles, since the three roles are subjects. The structure of all sentences is the same (one participant and the verb), and only the event role category changes.

Results revealed that participant looked more at sentences with agents and experiencers than at sentences with patients. Moreover, I found that participants exhibited a preference to interpret the first ambiguous noun phrase (i.e., María) as an agent/experiencer; hence, when they realized that it was a patient instead, an N400 component appeared. But this effect only appeared with patients, not with experiencers, showing that participants processed similarly agents and experiencers, and both different from patients.

What does this evidence show?

These results align with the hypothesis that there are only two role categories within the event role repertoire (Dowty, 1991): proto-agent and proto-patient. Therefore, when the sentences Mary dances or Mary has fun are processed, in both cases, Mary is interpreted as a proto-agent, whereas in Mary falls down, she is interpreted as a patient instead.

All in all, this evidence suggests that agents and experiencers belong to the same role category, the one of the proto-agent. Therefore, when we process language, it seems that we do not differentiate between agents and experiencers, rather we process them similarly. Experiencers are grouped together with agents, being the same character in the event, both proto-agents.

So, proto-agents are the protagonists of the story, the ones that spectators prefer to pay attention to, whereas proto-patients are the secondary characters who complement the story. Regarding experiencers, they do not play a different character in the story, instead, they are also the protagonists.

The post Doing and experiencing in language: how similar are they? appeared first on Mapping Ignorance.

Triplet-triplet annihilation

The most developed route to upconversion relies on a process called triplet-triplet annihilation. A light-absorbing molecule known as a sensitizer first captures a low-energy photon and passes that energy to a second type of molecule, the annihilator. The annihilator settles into a triplet excited state, an unusually long-lived condition that allows it to persist until it meets a second annihilator molecule in the same state. When two such molecules collide, one drops back to its ground state while the other is pushed up into a higher-energy singlet state, from which it emits a photon of higher energy than either of the photons that started the process.

A long-standing problem with this scheme is that ordinary sensitizer molecules lose a portion of the absorbed energy internally, as the initially excited state converts into the lower-energy triplet state that gets passed along. This loss caps how large an energy jump the whole system can ultimately achieve. One promising way around the problem is to replace the conventional sensitizer with an organic radical, a molecule carrying a single unpaired electron. That unpaired electron gives radicals unusual electronic and magnetic behavior and can make the conversion from absorbed light into a usable excited state considerably more efficient, since it opens transitions that are otherwise very weak in conventional molecules.

A single hydrogen to control it all

To work out why, the investigation combined light-based spectroscopy, electron spin resonance (a technique that detects unpaired electrons), and quantum chemical calculations. After the radical absorbs red light, its excitation transfers within about ten picoseconds (ten trillionths of a second) to the attached anthracene unit, creating a long-lived excited state that still carries some radical character. This internal handoff stretches the effective excited-state lifetime to more than 250 nanoseconds, roughly ten times longer than the 27 nanoseconds measured for the radical unit on its own. A longer-lived excited state gives the sensitizer far more opportunity to encounter another molecule and hand off its energy before simply decaying back to the ground state. The calculations independently pointed to the same conclusion, showing that energy transfer to a nearby annihilator proceeds far more readily through this long-lived triplet-like state than directly from the radical’s first excited state.

A mix needed

The work also tested an idea proposed in earlier research, that a single radical-containing molecule of this kind might act as both sensitizer and annihilator on its own, a so-called single-component upconversion system. The new experiments argue against this. Solutions containing only the radical molecule, across a range of concentrations, showed no detectable upconverted light, consistent with the radical’s higher excited states relaxing away too quickly for the necessary emission to compete.

Efficient upconversion instead required mixing the radical sensitizer with its hydrogen-added counterpart, or with the well-established annihilator 9,10-diphenylanthracene. Because the two molecules share nearly the same molecular framework, this two-molecule mixture is described as a “pseudo-single-component” system. Shining red light at 658 nanometres onto the radical paired with its hydrogenated twin produced blue emission at 450 nanometres, an apparent energy gain of about 0.9 electronvolts, with 7 percent of absorbed photon pairs converted into upconverted light. Pairing the same radical sensitizer with 9,10-diphenylanthracene instead produced blue emission at 437 nanometres, a slightly larger energy gain of about 1.0 electronvolt, with 12 percent efficiency. Both figures mark a large improvement over earlier all-organic radical sensitizers, which had reached efficiencies of only a few tenths of a percent, and the upconverted emission remained stable over hours of continuous illumination.

A practical strategy for photon upconversion

Beyond these specific molecules, the study illustrates how a minimal change in molecular structure, adding or removing a single hydrogen atom, can completely redirect a molecule’s photochemical role, switching it from an energy donor to an energy acceptor. Combining optical measurements, spin resonance, and computation made it possible to trace exactly why this switch occurs at the level of electronic structure, rather than simply observing that it happens.

The findings offer a clearer picture of how organic radicals can be harnessed for photon upconversion and point toward a practical strategy for building more efficient materials from radicals and their closed-shell relatives together. Such materials could eventually help capture more of the solar spectrum, drive light-powered chemical synthesis, or improve the efficiency of light-emitting technologies, by putting to use photons that would otherwise be wasted.

Author: César Tomé López is a science writer and the editor of Mapping Ignorance

Disclaimer: Parts of this article may have been copied verbatim or almost verbatim from the referenced research paper/s.

The post A new route to photon upconversion appeared first on Mapping Ignorance.

Author: Subir Sarkar, Emeritus Professor, Department of Physics, University of Oxford

According to our best understanding, the universe is expanding – and is doing so at an accelerating rate. This is believed to be caused by something called the “cosmological constant”, which was first proposed by Albert Einstein in his theory of general relativity. In recent decades, it has become better known as dark energy, which is believed to make up about 70% of the universe.

Crucial to this realisation were studies of Type Ia supernovae – exploding white dwarf stars. These are thought to emit a specific amount of light, which allows astronomers to determine their distances very accurately and thereby track the expansion of the universe. This work was awarded the 2011 Nobel prize in physics.

The accelerating expansion of the universe is thought to be due to negative pressure, an unusual property of dark energy that allows it to overcome the attractive force of gravity. Yet the exact nature of dark energy remains a puzzle. It cannot be explained by our best theory for the fundamental building blocks of the universe – known as the standard model of particle physics.

However, it is an integral part of another theory, devised to explain how the large-scale universe came to be. This is the standard model of cosmology, also called the lambda cold dark matter (ΛCDM) model.

Other observational evidence supporting the existence of dark energy includes fluctuations in the temperature of the cosmic microwave background (CMB) – the “afterglow of the Big Bang” – and the imprint of baryon acoustic oscillations (BAO) – sound waves from the early universe – in the distribution of galaxies.

But the ΛCDM model has come under increasing scrutiny as new data has emerged. A conference hosted by the UK’s Royal Society debated emerging cracks in the standard cosmological model. It has been noted that the model’s past successes may partly be due to “confirmation bias” – the unconscious tendency to favour information that supports our existing beliefs while giving less attention to information that challenges them.

Flawed framework?

Most cosmologists remain unconvinced by such criticism and are instead persuaded by the multiple lines of evidence that the universe’s expansion is accelerating.

However, all such analyses, for example of the CMB or BAO, are conducted using a special solution of Einstein’s equations called the FLRW (Friedmann-Lemaitre-Robertson-Walker) framework. This is a description of space-time that assumes the “cosmological principle”, namely that the universe is, on average, homogeneous and isotropic (looks the same in all directions).

However, our recent study found evidence to falsify the FLRW framework, presenting results showing that the universe is asymmetric, or lopsided. We believe this presents a fundamental challenge to dark energy by undermining the very basis for standard analyses of cosmological data.

The observation that the universe may be lopsided is known as the cosmic dipole anomaly. This anomaly cannot be fixed by tweaking parameters in the ΛCDM model. It requires starting again from scratch to build a model of the universe based on observations, rather than on a philosophical principle.

The Nobel prize winning astrophysicist Jim Peebles has said that the cosmic dipole anomaly is about as well established as another cosmological irregularity known as the Hubble tension. However, he notes, the cosmic dipole anomaly has received only a fraction of the interest heaped on the Hubble tension.

With my collaborators, Mohamed Rameez and his PhD student Animesh Sah at the Tata Institute of Fundamental Research, Mumbai, we have also shown that the acceleration of the expansion of the universe inferred from Type Ia supernovae is not the same in all directions.

The signal from supernovae is mainly aligned with a temperature difference in the afterglow of the Big Bang – the cosmic microwave background. This temperature difference, called the CMB dipole anisotropy, is believed to be due to the local motion of the solar system.

‘Tilted observers’

The acceleration inferred from supernovae cannot therefore be due to dark energy. Rather, it is probably an illusion, because we are “tilted observers”. This is because of Earth’s specific location in the cosmos, where our neighbouring galaxies are participating in a large-scale but localised streaming motion called the bulk flow.

The relationship between the intrinsic brightness of Type Ia supernovae and their distance has been crucial to the idea of dark energy. However, this relationship may not be as robust as was first assumed.

Astronomers in South Korea found evidence that the ages of the white dwarf stars that produce Type Ia supernovae (their progenitors) affect their brightness. However, this has been challenged by another team.

When we corrected for the dependence of the intrinsic brightness on the progenitor age, the indication is that the universe is decelerating, rather than accelerating.

This is just as is expected for a universe without dark energy. But it will come as a surprise to most cosmologists who believe that there is compelling independent evidence for ΛCDM, in particular from studies of the CMB. But CMB temperature fluctuations are not influenced by dark energy. Rather, they are related to the spatial curvature of the universe, the baryon density, and the dark matter density.

Whether dark energy is still a valid inference thus depends on whether the distribution of matter in the universe is indeed isotropic – exactly the same in all directions. Observations suggest otherwise. If these findings are confirmed, they could signal a paradigm shift in cosmology.

This article is republished from The Conversation under a Creative Commons license. Read the Original article.

The post Cracks in the foundations of today’s cosmological model appeared first on Mapping Ignorance.

This device was built ¹ to explore fundamental physics and realize a previously unobserved behavior, but it could also open doors for new applications in areas that bridge optics and electronics, including sensing, imaging and telecommunications. The phenomenon could help improve how signals are sent through and between devices, as well as create new opportunities to store more information in those signals.

“This electrical device that we manufactured at the Lurie Nanofabrication Facility has the potential to turn into something that measures different aspects of light,” said Yiming Gong, who helped lead the project as a doctoral student in the U-M Department of Physics. “But this originates from a very fundamental level of physics, which is the interference between different optical absorption processes.”

The researchers showed they could induce the orderly flow of electrons through a semiconductor using two different colors of light. By rotating the polarization of the two optical fields—the direction that the light waves wiggle—the researchers could also control the direction of the electronic current.

“This isn’t the way things normally work. When you think about electrons moving through a material, they’re moving because you’ve applied an electrical field and they actually bounce around and drift across the materials. Here, using light, you can actually sort of squirt the electrons in a specific direction without applying an electric field,” said U-M physicist Steven Cundiff, senior author of the team’s report.

Although previous work had shown the ability of light to get electrons flowing with light alone, this work goes a step further and shows the light gets electrons flowing in a narrow beam in a specific direction.

“The light no longer merely switches the current on; it also aims it,” Cundiff said.

Cundiff likens this to a lighthouse, which sweeps a beam of light by rotating its lamp. Here, the beam is made of electrons and that beam can be rotated by rotating the polarization of the two phase-coherent optical fields.

The phenomenon relies on quantum interference, which arises when two colors of light drive different absorption pathways to the same final state. The light transfers energy to the semiconductor material in discrete, or quantized, packets called photons, which mobilize the material’s electric charge carriers.

In the set-up, the incident light is absorbed by two routes at once, and the routes can be thought of as overlapping ripples, Cundiff said. Those ripples line up and enhance each other for electrons traveling a certain direction, but cancel each other out for electrons moving in other directions.

A collaborator on an earlier project, J.E. Sipe of the University of Toronto, had predicted it would be possible to create such an “electron lighthouse.” Now, working with the Lurie Nanofabrication Facility, Gong was able to realize such a device.

“The LNF is an amazing facility,” Gong said, adding it was a painstaking process to meld the device’s materials together in a way that didn’t introduce any extraneous electric fields.

“That was the biggest puzzle to solve for me, because there isn’t a standard way to do that. So I worked with the LNF staff to play around with different recipes and temperatures to come up with a manufacturing process.”

The post Electron lighthouse appeared first on Mapping Ignorance.

For decades, Alzheimer’s research has been dominated by a seemingly intuitive idea: the disease is largely explained by the accumulation of abnormal proteins in the brain, primarily beta-amyloid and tau. However, while this view has been useful, it has fallen short in explaining certain fundamental aspects. One of the most striking is that Alzheimer’s affects women far more than men. This is not a minor difference: roughly two-thirds of cases occur in women, and the clinical progression tends to be faster and more severe. The work by López-López and colleagues addresses this imbalance from a different angle. Instead of focusing exclusively on protein accumulation, it shifts attention to the brain’s immune system—specifically to a molecular pathway known as interferon signaling. What the study proposes is that the brain’s immune response, far from being a mere bystander, could be a central driver of the disease, and one that operates differently in men and women ¹.

To understand the importance of this idea, it helps to briefly consider what interferons are. They are molecules that form part of the body’s first line of defense against infections, especially viruses. They activate genetic programs that help cells detect and neutralize threats. Under normal conditions, this system is essential for survival. The problem arises when it becomes chronically or inappropriately activated, as is the case in several inflammatory diseases ². So, the study begins with an analysis of human brain tissue from Alzheimer’s patients. Using transcriptomics (techniques that measure the activity of thousands of genes at once) the authors identified a clear pattern: in the brains of women with Alzheimer’s, genes associated with the interferon response were much more strongly activated than in men (Figure 1). This pattern was not explained by differences in disease severity, since both groups showed comparable levels of neuropathological damage. Instead, it points to an intrinsic difference in how the brain responds to the disease. This finding is significant because it suggests that women not only develop Alzheimer’s more often, but may do so through a partially distinct biological pathway. In other words, the same clinical diagnosis may conceal different underlying mechanisms depending on sex.

To move beyond correlation and explore causality, the researchers turned to animal models. They used genetically modified mice (APP/PS1), widely employed in Alzheimer’s research because they develop amyloid plaques similar to those seen in patients. Once again, the same pattern emerged: female mice showed much stronger activation of the interferon pathway, along with increased brain inflammation, neuronal damage, and functional impairment. An especially interesting point is that these differences were not limited to amyloid plaque levels. Indeed, female mice did have more plaques, but the study shows that neuronal damage and inflammation cannot be explained solely by amyloid accumulation. This supports a growing idea in the field: that Alzheimer’s toxicity depends not only on protein buildup, but also on how the brain responds to it. In this context, microglia (the brain’s resident immune cells) become key players. Cell-level analyses revealed that these cells are the main drivers of interferon activation. In essence, microglia detect signals of damage or stress and launch an inflammatory response which, if sustained over time, can ultimately harm neuronal tissue. To test whether this pathway is not only associated with disease but actually contributes to it, the authors performed direct manipulation experiments acutely activating interferon signaling in healthy animals using an immune stimulus. To do that, they did intraperitoneal administration of polyinosinic:polycytidylic acid (poly(I:C)) to induce inflammation, but also they performed crossings between the APP/PS1 mice and a microglia-specific Rela conditional knockout line (AD hiIFN). The result was striking: within hours to days, the brains of these animals began to show changes resembling Alzheimer’s pathology, including altered neuronal activity, synaptic disruption, and markers of axonal damage. This results suggest that activation of this pathway is not a secondary phenomenon, but may be sufficient to induce key features of the disease. In parallel, the researchers performed the inverse experiment: they chronically enhanced interferon signaling in microglia within APP/PS1 mice (an Alzheimer’s disease model). This led to a worsening of neurodegenerative features, even though, paradoxically, amyloid plaque levels did not increase—and in some cases even decreased. This last finding is particularly revealing because it challenges a widespread assumption: that fewer plaques necessarily means a healthier brain. Here, the opposite happens!! The animals had less amyloid burden, yet more neuronal damage and worse function. This points to inflammation, and specifically interferon signaling, as an independent driver of degeneration.

Next, the key question is whether this pathway can be targeted therapeutically. To address this, the study explores a pharmacological intervention aimed at the cGAS–STING pathway (a molecular system that senses cellular damage and triggers interferon production). Using a specific inhibitor (STING inhibitor C-176), the researchers demonstrated by RNA-seq and immunohistochemistry that the treatment was able to reduce interferon signaling in the brain. The results were consistent: inflammation decreased, neuronal integrity was better preserved, and cognitive performance improved in memory tests (very important from the therapeutic and functional perspective). More importantly, these effects were far more pronounced in females, which aligns with the idea that this pathway is more active in them and therefore more impactful when blocked. Notably, these improvements occurred without significant changes in amyloid plaque levels. Once again, the conclusion is clear: modulating the immune response may be as important as, or even more important than, targeting protein aggregates. Taken together, this work proposes a shift in how Alzheimer’s is understood. Rather than viewing it solely as a disease of protein accumulation, it suggests it is also—critically—a disorder of immune dysregulation. More specifically, one in which excessive interferon signaling plays a central role, particularly in women.

The practical implications are twofold. First, it opens the door to therapies that modulate immune pathways in the brain, something already feasible in other diseases using drugs that target similar mechanisms. Second, it highlights the importance of incorporating sex as a biological variable in both research and treatment design. What works in men may not work the same way in women, and vice versa. As with any preclinical study conducted in mice we must be cautious when extrapolating to humans: results in animal models do not always translate directly, and manipulating the immune system carries inherent risks, given its essential role in protecting against infections. However, the value of this work lies not only in its potential clinical applications, but in the conceptual framework it offers.

By identifying interferon signaling as a key factor in sex-specific vulnerability to Alzheimer’s, the study provides a mechanistic explanation for a long-observed but poorly understood phenomenon. Perhaps more importantly, it suggests that intervening in this pathway could alter disease progression even without changing classical amyloid pathology. In a field where many clinical trials have failed by focusing solely on plaque removal, this perspective is not just novel, it may be necessary to explore it in the future.

The post The hidden role of the immune system in female Alzheimer’s appeared first on Mapping Ignorance.

Flexible by design

The material at the center of this approach is a layered compound called TaCo₂Te₂. Like graphite, its layers are held together by weak van der Waals forces, which makes it easy to combine with other materials without demanding a perfect chemical match. Unlike most layered crystals, though, its atoms are not arranged in a perfectly regular pattern. At room temperature they undergo a small, periodic shift known as a Peierls distortion, a phenomenon first described in the 1930s by the physicist Rudolf Peierls, who showed that in some materials the electrons and the surrounding lattice can lower their combined energy by nudging atoms slightly out of their ideal positions. When TaCo₂Te₂ is heated above about 523 kelvin, or 250 degrees Celsius, this distortion disappears and the crystal adopts a more symmetric structure, although traces of the original pattern linger as fleeting, temperature-driven fluctuations.

These fluctuations are not random. Calculations of how the crystal vibrates, backed up by Raman spectroscopy and electron diffraction experiments, show that the lattice is especially unstable along one crystallographic direction, meaning atoms can shift more easily there than in the perpendicular direction. This directional softness turns out to be the key ingredient that controls how a second material grows on the surface.

Heating TaCo₂Te₂ also sets its atoms in motion across the surface. Tantalum atoms drift away, while cobalt and tellurium reorganize into a thin crystalline layer whose composition resembles cobalt telluride. Rather than clumping into isolated islands, this new layer spreads sideways first and only then thickens, producing a smooth, continuous interface with the crystal beneath it, a layer-by-layer growth style that electron microscopy confirms is energetically favorable.

Directionally locked

The most striking result is that the new crystal does not settle into just any orientation. Even though its symmetry differs from that of the underlying TaCo₂Te₂, it becomes directionally locked. Along one in-plane direction, the atomic spacings of the two materials match closely, giving a tight, well-ordered registry. Along the perpendicular direction the mismatch is far larger, but this happens to be exactly the direction where TaCo₂Te₂ is most flexible, thanks to its lattice instability. Instead of rotating to relieve that mismatch, as symmetry-mismatched interfaces typically do, the interface absorbs it through a gentle, ongoing structural adjustment. The result is a crystal that stays aligned while only one direction bears most of the strain.

That flexible direction does not settle into a simple repeating pattern either. Instead, it forms what is called a one-dimensional incommensurate superlattice: the repeating atomic spacings of the two crystals never line up exactly, yet they stay locked together over long distances. Physicists have long recognized this kind of arrangement as an efficient way to accommodate mismatched lattice spacings without piling up large numbers of defects, since the strain is spread out through a slowly changing modulation rather than concentrated at sharp boundaries. Diffraction measurements and a geometric analysis of the two lattices both confirm that no perfectly repeating, shared lattice can account for what is observed, so the incommensurate description fits best.

Directional lattice instability is key

To check that the lattice instability is truly responsible for this behavior, the same experiment was repeated with a closely related compound, TaNi₂Te₂, which has nearly the same crystal structure but lacks the structural instability. Under similar heating, a surface layer still forms, but its orientation varies from spot to spot, and rotational disorder sets in. This comparison shows that matching crystal symmetry is not enough on its own to produce a locked, well-ordered interface. The directional lattice instability is the extra ingredient that guides the growth.

The local reconstruction effect

One more effect helps stabilize the interface. Even above the temperature where the bulk crystal should be fully symmetric, the atomic layers right beneath the interface partly slip back into the low-temperature distorted structure. This local reconstruction stays confined to the surface rather than spreading through the whole crystal, and it appears to strengthen the bond between the two materials, keeping the aligned structure stable even at high temperature.

As we have just seen, structural instabilities, usually considered as a liability because they signal that a crystal is on the verge of changing shape, can instead become useful design tools. A substrate whose atoms rearrange more easily in certain directions can guide the growth of otherwise incompatible materials, producing well-aligned interfaces without extensive surface treatment. This idea could widen the range of material combinations available for future electronic, optical, and quantum devices, where the atomic arrangement at an interface often determines how well the whole device performs.

Author: César Tomé López is a science writer and the editor of Mapping Ignorance

Disclaimer: Parts of this article may have been copied verbatim or almost verbatim from the referenced research paper/s.

The post Rewriting the rules of heteroepitaxy with a flexible crystal appeared first on Mapping Ignorance.

 

A giant mirror to create “sunlight on demand” was just approved by the United States Federal Communications Commission (FCC), despite opposition from astronomers and the public, and real safety concerns.

The FCC approved the company Reflect Orbital to test one satellite, named Earendil-1, as a means of reflecting the sun’s rays back to Earth for extra solar energy and wide-area lighting. The light is expected to cover an area about five kilometres wide, and will require repointing every four minutes.

And this is just the start. Reflect Orbital plans to have more than 50,000 satellites in action by 2035, which they claim will be used across agricultural, emergency response and other industrial sectors.

There are many problems with this proposal, including impacts these satellites will have on human health and safety, as well as on astronomy and the low-Earth environment.

Flashes during mirror repointing could disrupt pilots and drivers. The light could also disrupt circadian rhythms of plants, animals and humans. Sensitive detectors in research telescopes, as well as star-tracking cameras on lower altitude satellites, could be overloaded and fried.

The FCC said that the “risks of harm raised on the record regarding Reflect Orbital’s solar reflector are unrelated to the Commission’s role in authorizing use of radiofrequency spectrum.”

‘Weird space stuff’

Satellite proposals for “emergent space activities” in low-Earth orbit are becoming increasingly outlandish. The proposals have become so weird, in fact, that the FCC recently published a document called “Spectrum Abundance for Weird Space Stuff.”

“Once the province of science fiction,” this document states, “American companies are now upgrading, relocating and servicing satellites; manufacturing pharmaceuticals in space; building private inhabitable spacecraft; and conducting private robotic missions to the surface of the Moon.”

Millions of orbital AI data centres are also planned. Corporations seem to be scrambling to launch anything that might persuade investors throw money at them: space advertising, hotels for billionaires, artificial meteor showers, space burials for cremated remains, solar-powered infrared beams to power data centres and a variety of orbital missiles.

The phrase “weird space stuff” is refreshingly truthful. So, how did we get here?

SpaceX controls orbit

There are close to 11,000 SpaceX Starlink satellites currently in orbit above our heads. Anyone who wants to launch into low-Earth orbit needs to carefully consider SpaceX operations, or directly co-ordinate with them.

Otherwise they risk collisions, like the near-miss between a Starlink and Chinese satellite in December 2025.

Even the Artemis I launch in 2022 and Artemis II launch in 2026 had small “cutout” windows in their launch timing to avoid satellites, including those belonging to Starlink.

Co-ordination is good. Forcing it because one corporation has effectively occupied low-Earth orbit is not. Indeed, the 1967 Outer Space Treaty, which was signed by more than 100 countries including the United States, China and Russia, states that, “outer space is not subject to national appropriation by claim of sovereignty, by means of use or occupation.”

Whether SpaceX’s extensive use of Earth orbits violates this principle is now being tested in real time.

Copycat megaconstellations

In February, SpaceX filed with the FCC for one million more satellites, for “AI data centers.”

One million. That is 40 times as many satellites as have ever been launched — for a single megaconstellation consisting of completely untested technology that may not even work in space.

Not only did the FCC accept SpaceX’s filing, but they did so at ludicrous speed. Scientists worldwide then had just 30 days to model the effects with woefully incomplete information on masses, sizes, compositions and orbital distributions.

At the time of writing, four other copycat AI data centre proposals have been filed by rival companies, for tens of thousands of satellites each. And SpaceX just proposed another 100,000 satellites to interface with the million AI data centres that it already asked for.

Solar energy from space

The U.S. Federal Communications Commission was originally set up to regulate radio broadcasts. But it is now being asked to evaluate many non-radio effects, including orbital safety, which it may not have the required expertise for. It would make sense to move some of this evaluation to the U.S. Office of Space Commerce. However, recent budget cuts make that infeasible.

Consequently, the FCC will soon be asked to judge a daunting range of satellite proposals. They include a cluster of proposals to gather solar energy from space.

One idea is to send solar power down to earth through high energy beams. These could change atmospheric chemistry and kill birds and other wildlife that stray into the beam.

They would also require no-fly zones around receiving stations for airplanes and also satellites on lower altitude orbits (such as the orbits SpaceX just requested for 100,000 more Starlink satellites).

Environmental costs

While many of these projects claim to solve environmental problems by creating clean energy or capturing it in space, they function as a form of greenwashing.

The solar energy generated is only clean if you ignore the environmental costs of building, launching, maintaining and burning satellites up in Earth’s atmosphere. The daily operations of all these proposed systems will have huge environmental consequences.

There are companies that have tested plans for removing space debris from orbit. This is helpful for avoiding Kessler Syndrome — a runaway chain reaction of collisions. But where will the debris go once removed from orbit?

It will fall into Earth’s atmosphere, where it will deposit metal and possibly impact Earth’s surface. It is unclear who is responsible for any resulting damage or deaths.

For all humankind

The majority of satellites in orbit today are American, and the main federal agency regulating satellites is not set up to do that well. We are now seeing the consequences.

While outer space is effectively infinite, low-Earth orbit most definitely is not. Satellites orbit the Earth around once every 90 minutes. This means the collision potential between two objects in orbit is large.

The many satellites and rocket bodies that have burned up in Earth’s atmosphere over the last few years have already measurably altered it. Preliminary studies show that using Earth’s atmosphere as a crematorium for tens of thousands of satellites will have devastating effects on ozone and other atmospheric chemistry.

Astronomy is also under threat from some of the “weirder” ideas like space mirrors, solar sails and diffuse sky brightening from orbital debris.

An innovation challenge

We are not here to argue against satellites. Indeed, they provide a wide range of beneficial services to science and society. But each satellite comes with a cost that must be taken into account.

Ultimately, this is an innovation challenge. Unfettered growth and exploitation of any environment comes with serious consequences, including to the long-term sustainability of the operations that depend on that environment.

Doing more with less is the engineering challenge that needs to be met if we want to continue to use satellites in orbit.

 

This article is republished from The Conversation under a Creative Commons license. Original article.

The post A giant space mirror to test ‘sunlight on demand.’ appeared first on Mapping Ignorance.

An international team led by Centro de Astrobiología (CAB, CSIC-INTA) researcher Izaskun Jiménez-Serra has now identified ¹ the first sugar in interstellar space: erythrulose. This molecule is the only possible four-carbon ketose, and on Earth it is commonly found in raspberries and sunless tanning products. Erythrulose was detected toward the molecular cloud G+0.693−0.027, located near the centre of our Galaxy, the Milky Way. The discovery was made possible by ultra-sensitive, broadband spectroscopic surveys carried out with the 40-m Yebes radio telescope and the 30-m telescope of the Institute for Radio Astronomy in the Millimeter Range (IRAM).

The team identified 12 spectral lines matching the laboratory spectrum of erythrulose measured at the University of the Basque Country. The study also shows that this sugar is at least eight times more abundant than similar three-carbon sugars, none of which were detected in the same region. “This finding was unexpected, as the prevailing view in astrochemistry is that interstellar molecules grow in size through the sequential addition of carbon atoms”, says Izaskun Jimenez-Serra.

Working in collaboration with chemists from the University of Extremadura and Radboud University (the Netherlands), the CAB team discovered that erythrulose can form within interstellar ices from simpler two-carbon alcohols and aldehydes.

Based on the abundance of erythrulose measured in the G+0.693−0.027 molecular cloud, the researchers estimate that between 0.5 and 50 million tonnes of this sugar could have reached Earth’s surface during the Late Heavy Bombardment, which occurred approximately 4.1 to 3.8 billion years ago. The presence of erythrulose in interstellar space therefore provides an alternative source of sugars that may have contributed to the emergence of the first metabolic and replication processes on the early Earth.

“The detection of erythrulose is very exciting because it opens up the possibility of discovering in space other sugars such as ribose, which is part of RNA, and other important molecules for the origin of life,” says Carlos Briones, co-author of the study.

The post Sugar in interstellar space appeared first on Mapping Ignorance.

What if we could stop blood cancers before they even start? New research reveals ¹ that chronic inflammation dramatically reshapes the bone marrow environment years before leukemia emerges, potentially offering a window for early intervention.

The Hidden Life of the Bone Marrow

Every second, your bone marrow produces millions of new blood and immune cells through delicate cooperation between blood stem cells and supportive stromal cells. Over time, aging, inflammation, or genetic mutations can disrupt this partnership, allowing mutated stem cells to silently expand.

This leads to a condition called clonal hematopoiesis, or CHIP, which affects roughly 10–20% of adults over 60 years old. Despite the lack of symptoms, the condition increases blood cancer risk tenfold and doubles the risk of cardiovascular disease. In some cases, it can progress to myelodysplastic syndrome, where bone marrow fails to produce healthy blood cells, and up to 30% of these advance to acute myeloid leukemia—an aggressive and often fatal blood cancer.

Uncovering the Culprit: Inflammation

Researchers conducted an in-depth molecular analysis of human bone marrow samples from healthy donors, people with CHIP, and patients with myelodysplastic syndrome (leukaemia). Using advanced techniques including single-cell sequencing and biopsy imaging, they built a high-resolution map of the bone marrow environment across these different stages.

The results revealed a striking cellular transformation beginning well before clinical disease appears. Researchers identified inflammatory stromal cells instead of normal stem-cell-supporting cells. Unlike healthy stromal cells, these inflammatory cells release signaling molecules that attract and activate certain immune cells (T cells). In turn, these T cells amplify the inflammatory signal, creating a self-reinforcing loop that sustains chronic inflammation, suppresses healthy blood formation, and promotes abnormal blood vessel development.

Surprisingly, the mutated blood cells do not directly drive inflammation. Instead, the inflammatory network in the surrounding microenvironment becomes dominant, replacing much of the bone marrow’s regenerative structure. As diseased stem cells also fail to trigger production of important signaling molecules, this could explain why bone marrow function deteriorates.

This discovery indicates inflammation has a central force in early blood disease stages and establishes the bone marrow microenvironment as a critical therapeutic target. Based on these findings, anti-inflammatory treatments may help preserve bone marrow function in older adults with CHIP. In addition, the distinct molecular signatures of these inflammatory cells may serve as biomarkers to identify at-risk individuals long before clinical symptoms emerge, possibly preventing progression to leukemia.

What I find particularly interesting about the research is that it shows yet another example of how deadly chronic infammation can be. I would not be surprised if the incidence of this disease were to increase in the next years, as we are most living under chronic stress conditions, which is one of the biggest contributors to chronic inflammation. To me, this is more of a health issue, it is a social issue and to get rid of it, we need much more than antiinflammatory drugs.

The post Inflammation-driven changes in bone marrow: An early indicator of leukemia? appeared first on Mapping Ignorance.

The Martini force field

One of the main tools for this kind of study is molecular dynamics simulation, which calculates how atoms move over time by applying the basic laws of physics. Simulating every single atom of a protein, its surrounding water, and a supporting material gives rich detail, but is extremely demanding on computing power, especially when many attachment geometries need comparing. A widely used shortcut is coarse-grained modeling, in which small clusters of atoms are grouped into single, simplified particles. This keeps the essential physical behavior of the system while making the calculations far cheaper. The most established coarse-grained approach for biological molecules is the Martini force field, first developed in 2004 to describe fatty, membrane-forming molecules and later extended to proteins and, more recently, to a more accurate third generation of the model. Martini has since become a standard tool for simulating large biological assemblies that would otherwise be out of reach.

Introducing MartiniSurf

A team of researchers now introduces MartiniSurf ¹, an open-source software package that automates the preparation of these coarse-grained simulations for biomolecules attached to surfaces. Building such a system by hand normally involves several separate programs and many manual steps, which makes it hard to reproduce results or compare different setups fairly. MartiniSurf instead combines the whole process into a single, streamlined workflow. Starting from the structure of a protein or a DNA strand, it converts the molecule into its coarse-grained form, builds a model of the supporting surface, positions the molecule in a chosen orientation, adds water and dissolved ions, and produces all the files needed to run the simulation with GROMACS, a molecular dynamics program that has been developed and refined for decades. Because every step is controlled by explicit settings rather than manual adjustments, two researchers using the same settings should obtain the same starting system, making comparisons across studies far more reliable.

Restraining forces or linkers

A central feature of the software is its flexibility in representing how a biomolecule is attached. In one mode, selected regions of the molecule are held near the surface through simple restraining forces, without modeling the physical connector itself. In another mode, an explicit molecular linker is placed between the biomolecule and the surface, so that the linker’s own length and flexibility can influence the simulation. Because the properties of such linkers are known to affect how active and stable an attached enzyme remains, this second option offers a more realistic picture whenever the necessary coarse-grained description of the linker is available.

Variety of supporting surfaces

The supporting surfaces themselves can also be varied widely. The software can generate simple flat reference surfaces or draw on previously developed coarse-grained models of real materials such as graphene, graphite, carbon nanotubes, and agarose-like gels, the type of porous material commonly used to immobilize enzymes in the laboratory. Properties such as surface chemistry, electric charge, particle spacing, and added chemical groups can all be adjusted. Recent studies have shown that surfaces built without care can cause water molecules near the interface to line up in unrealistic, almost frozen patterns, an artifact rather than genuine physical behavior. The software includes options that let users introduce controlled irregularity into the surface to reduce this problem, while staying consistent with the wider Martini modeling framework.

In practice

To demonstrate how the approach works in practice, the study presents examples involving both proteins and DNA. For an enzyme attached to an agarose-like surface, the software reproduces several different attachment arrangements, including a complete system containing the enzyme together with its natural helper molecule and reaction substrate. These examples show how changing the attachment point or adding chemical groups to the surface can be explored systematically, without needing to reconstruct every model from scratch by hand.

The approach is also extended to DNA tethered to a graphene-like surface, comparing an electrically neutral surface with a positively charged one. The coarse-grained simulations reproduce the same general behavior previously reported in much more computationally demanding all-atom studies: the DNA strand remains mostly upright above the neutral surface but bends progressively toward the surface when it carries a positive charge, drawn in by electrostatic attraction. This comparison is not meant to reproduce exact experimental values, but rather to confirm that the automated setup captures the right physical trends, providing a check on the reliability of the workflow.

Reproducibility

Every simulation generated by the software automatically records the settings, software versions, and modeling choices used to build it, so that another researcher can repeat or directly compare a given calculation later. Importantly, the software does not introduce new physical models or new force-field parameters of its own. Instead, it draws on already validated Martini tools and organizes them into a consistent, automated pipeline.

Overall, MartiniSurf addresses a practical bottleneck in computational chemistry rather than proposing a new simulation method. By automating the construction of surface-bound protein and DNA systems, it makes coarse-grained molecular dynamics studies faster to set up, easier to reproduce, and simpler to scale across many attachment geometries. This kind of tool should make it easier to systematically study how surface properties, attachment geometry, and molecular flexibility shape the behavior of immobilized biomolecules, supporting the design of better catalysts, biosensors, and other biotechnology applications.

Author: César Tomé López is a science writer and the editor of Mapping Ignorance

Disclaimer: Parts of this article may have been copied verbatim or almost verbatim from the referenced research paper/s.

The post Overcoming bottlenecks in bio-interface simulations: The MartiniSurf Approach appeared first on Mapping Ignorance.

Chimeric antigen-specific receptor (CAR) T-cell immunotherapy targeting B-cell maturation antigen (BCMA; BCMA-CAR T cells) has produced remarkable clinical responses in advanced multiple myeloma (MM), with a substantial fraction of patients achieving complete remission ¹. However, many patients relapse within months, largely due to insufficient CAR T-cell persistence, a recognized mechanism of acquired resistance ². A major barrier to durable responses is the tumor microenvironment (TME), which is profoundly reshaped by tumor metabolic reprogramming ³. In fact, many tumor types, specially MM, are highly dependent on glutamine (Gln) metabolism for cell survival and proliferation ⁴. Thus, Gln “addiction” depletes local Gln levels and availability, impairing antitumor T-cell effector functions and promoting immune escape.

Based on these observations, a team of researchers hypothesized ⁵ that Gln deprivation within the TME limits the efficacy of CAR T-cell therapies. Indeed, antigen-specific T cells fail to proliferate and to produce interferon gamma (IFN-γ) under Gln-restricted conditions. They therefore investigated whether enhancing Gln uptake via overexpression of the Human SLC1A5 (ASCT2) amino acid transporter could improve the metabolic fitness and the antitumor activity of BMCA-CAR T cells in MM.

To assess the functional impact of ASCT2 overexpression in T cells, the authors infected CD8⁺ and CD4⁺ T cells from mice with an expression vector containing Asct2. To assess the impact of enhanced Gln uptake on T-cell function, CD4⁺ and CD8⁺ T cells from mouse spleens were stimulated with anti-CD3/CD28 beads at varying Gln concentrations. Interestingly, Asct2 overexpression significantly improved T-cell proliferation and IFN-γ production. Similar enhancements were observed in CD8⁺ OT-I (these cells contained a transgenic TCR designed to recognize ovalbumin (OVA) peptide 257-264 on MHC class I, modeling thus cytotoxic T-cell responses) and CD4⁺ OT-II T cells (these cells contained a transgenic TCR designed to recognize OVA peptide 323-339 on MHC class II, modeling the helper T-cell responses). Thus, Asct2-overexpressing OT-I T cells showed increased cytotoxicity against B16OVA cells. B16OVA cells are a well established in vitro model for cancer immunotherapy, expressing ovalbumin in order to facilitate strong immune responses to tumor antigens. The authors found that, even at low Gln levels, Asct2-overexpressing CD4⁺ OT-II T cells were more efficient lysing B16OVA cells.

Next, to assess the potential of Asct2 overexpression to enhance ACT therapy in vivo, B16OVA tumor-bearing mice were treated with either control or Asct2 CD8⁺ OT-I T cells. Interestingly, 14 days after tumor injection they found lower levels of Gln in the interstitial fluid, confirming a Gln-deficient TME. Then, mice were subcutaneously injected with B16OVA cells and, 7 days later, treated with control or Asct2 CD8+ OT-I T cells. Remarkably, mice treated with Asct2 CD8⁺ OT-I T cells exhibited a delay in tumor growth as evidenced by tumor weight and survival rate analysis. Moreover, mice treated with Asct2 CD8+ OT-I T cells showed a higher number of IFN-γ–producing antigen-specific T cells in the spleen compared with control T cells.

Given the well-established Gln “addiction” of MM ⁶ and the consequent Gln depletion in the bone marrow (BM), the authors next evaluated whether Asct2 overexpression in BCMA-CAR T cells could improve Gln uptake and potentially benefit CAR T-cell immunotherapy in MM. Because MM expresses BCMA, they generated a murine anti–Bcma-CAR T-cell therapy effective against MM in vivo. They tested whether Asct2 overexpression could improve their efficacy in vivo and observed that Bcma-CAR cells exhibited robust proliferation and high IFN-γ secretion when stimulated with Bcma. Besides, subsequent co-culture experiments with Bcma-expressing MM cells confirmed that Bcma-CAR T cells selectively secreted IFN-γ, and displayed cytolytic activity against MM cells. Replacement of the GFP reporter with ASCT2 yielded BCMA-CAR/ASCT2 T cells with enhanced Gln transport capacity. In coculture assays, ASCT2-expressing BCMA-CAR T cells exhibited superior cytolytic activity against MM cells, particularly under low-Gln conditions.

To evaluate if Asct2 overexpression could enhance ACT immunotherapy with Bcma-CAR T cells in vivo, they used a murine model in which mice were challenged with MM cells and treated IV with CD8⁺ and CD4⁺Bcma-CAR T cells (1:1 ratio) or Asct2 Bcma-CAR T cells. Interestingly, Asct2 Bcma-CAR T cells significantly improved overall survival (OS) in mice challenged with tumors, achieving 90% survival at day 100 after treatment. Besides, fourteen days after CAR T-cell transfer, there was a significant reduction in tumor cells in the BM of both CAR T-cell–treated mice groups, being more significant in the BM of mice treated with Asct2 Bcma-CAR T cells compared with those treated with Bcma-CAR T cells. Moreover, there was a higher number of infiltrating CAR T cells in both the spleen and BM of mice receiving Asct2 Bcma-CAR T cells. Then, they next evaluated the antitumor effect of the Bcma-CAR T cells in a genetic MM preclinical model that closely mimics common genetic alterations observed in human MM and, similarly, treatment with Asct2 Bcma-CAR T cells significantly enhanced OS compared with conventional Bcma-CAR T cells.

Finally, the authors explored the clinical relevance of ASCT2 expression in MM. RNA-seq analysis of BM aspirates of untreated patients with MM compared ASCT2 expression levels in various B-cell subtypes obtained from healthy donors as previously described, showed a marked upregulation in MM cells compared with normal B-cell subsets, suggesting a potential prognostic relevance of ASCT2 expression in MM. Next, to explore the clinical relevance of ASCT2 overexpression in human CAR T cells, they created 2 lentiviral constructs encoding an antihuman BCMA-CAR. They prepared a conventional BCMA-CAR and the ASCT2 BCMA-CAR construct coexpressing ASCT2. Notably, ASCT2 BCMA-CAR T cells showed greater proliferation and higher IFN-γ production upon BCMA stimulation at various Gln concentrations, and they had a higher proportion of naïve stem cell memory and central memory subsets. To further demonstrate the importance of ASCT2, they performed cytotoxicity assays with BCMA-CAR T cells against tumor cells with different ASCT2 levels. Silencing ASCT2 in MM cells reduced Gln uptake and confirmed the functional importance of this transporter in tumor–immune metabolic competition.

In summary, Asct2 overexpression enhances T-cell and CAR T-cells metabolic fitness and effector function under Gln-restricted conditions. In both B16OVA melanoma and MM in vivo tumor models, this strategy improved antitumor immunity and increased overall survival. These findings support metabolic engineering of CAR T cells as a rational approach to overcome nutrient deprivation in the TME an to enhance durable tumor control.

The post Enhanced BCMA – CAR T-cell therapeutic efficacy in multiple myeloma appeared first on Mapping Ignorance.

A new study applies ¹ the idea to the simulation of acoustic waves of geophysical interest, like seismic imaging subsurface exploration. All of these applications rely on the Helmholtz equation, named after the nineteenth-century physicist Hermann von Helmholtz, which describes how a wave spreads out once its frequency is fixed. Higher frequencies mean more rapid oscillation, and more rapid oscillation is far harder to compute accurately. Conventional numerical methods handle this by using extremely fine computational grids, which quickly become slow and expensive as the frequency rises.

Neural networks offer an appealing alternative because they need no such grid, but they come with a stubborn flaw. Left to their own devices, they learn smooth, slowly changing patterns long before they learn rapidly oscillating ones, a well-documented tendency known as spectral bias, or low-frequency bias. For wave problems, where the interesting behavior is exactly the rapid oscillation, this bias means standard PINNs can take a very long time to train and often still miss fine detail.

Gabor-embedded PINN

The new approach addresses this by building oscillation into the network from the start, using Gabor functions. A Gabor function pairs a wave-like oscillation with a Gaussian envelope that fades toward zero away from its center, so the oscillation stays confined to a small region of space. These functions were introduced in 1946 by the Hungarian-British engineer Dennis Gabor, who later won the Nobel Prize in Physics for inventing holography, as a way to describe signals that are localized in both time and frequency; the same mathematical shape turns out to suit wavefields just as well. Instead of asking the network to invent every oscillation from nothing, the method hands it oscillations that already look like waves and asks it only to work out how they should be positioned and combined across space.

The central innovation is in how that positioning is learned. Rather than having the network output the wavefield directly, it is trained to output a transformed set of coordinates, and the Gabor functions applied in that transformed space produce the final oscillating solution. This shifts the burden of representing the wave’s fine structure into a coordinate mapping, which turns out to be much easier for the network to learn smoothly. Earlier attempts to combine Gabor functions with neural networks needed extra networks or additional parameters to control the position and strength of each function, adding complexity and, at times, instability during training. The new formulation avoids this: it needs no additional trainable parameters beyond what a conventional network of the same size would already have.

An absorbing boundary

A second contribution deals with the edges of the simulated region. Real waves can travel forever, but any simulation must stop somewhere, and naive boundaries send waves bouncing back into the domain as unwanted reflections. The established solution, first devised in 1994 for electromagnetic wave simulations and since adopted across many wave-based numerical methods, is a perfectly matched layer: a boundary region that gradually absorbs outgoing waves instead of reflecting them. Applying this technique inside a neural network training scheme usually calls for cumbersome complex-number arithmetic; the study works around this by splitting the calculation into separate, real-valued components and by supplying an approximate mathematical formula for the wave that would exist in a uniform medium, which keeps the training efficient without sacrificing accuracy near the boundary.

Fewer training steps, better results

The method was tested across velocity structures of increasing complexity, from a simple layered medium to benchmark models resembling real, geologically folded terrain, and across a range of frequencies. In every case it was compared against an otherwise identical standard network trained the same way. It consistently reached accurate results in far fewer training steps, produced smaller errors, and stayed stable even when the network’s size or the random starting values of its weights were changed, an area where conventional networks, and earlier Gabor-based attempts by other researchers, tended to falter. The gap was largest precisely where standard methods struggle most: at higher frequencies.

Notably, these improvements did not come from making the network bigger. They came from embedding a piece of mathematical knowledge about how waves behave directly into the network’s structure, freeing its limited capacity to focus on learning spatial relationships rather than reinventing basic oscillations over and over. Because the Helmholtz equation also governs electromagnetic waves and other oscillatory phenomena, the same strategy could plausibly extend beyond acoustics, provided the physical equations involved are adapted accordingly. Combining mathematical structure with machine learning can produce methods that are simultaneously faster, more accurate, and less computationally demanding than relying on general-purpose neural networks alone.

Author: César Tomé López is a science writer and the editor of Mapping Ignorance

Disclaimer: Parts of this article may have been copied verbatim or almost verbatim from the referenced research paper/s.

The post Gabor-embedded PINN for overcoming spectral bias in high-frequency acoustic simulations appeared first on Mapping Ignorance.