Les apps mobiles Blued et Finka sont maintenant absentes de l'AppStore d'Apple en Chine...
Grindr y avait déjà été censuré.
Ces trois apps sont destinées aux public homosexuel et bisexuel.

La Chine a choisi de les faire interdire de l'AppStore d'Apple et du PlayStore de Google, les deux multinationales ont choisi de respecter la loi et de bannir ces deux apps.
Mais silencieusement, sans réagir, sans communication, comme si c'était normal.

L'homosexualité n'est pas illégale en Chine, contrairement à de trop nombreux autres pays qui partagent une religion commune et une loi commune.
Mais petit à petit la Chine interdit ceci ou cela, comme la "Shangai Pride" suspendue en 2020.

Ces deux multinationales, Apple et Google, font assembler (assemblage final) leurs smartphone respectifs en Chine pour de nombreux marchés.
La Chine est aussi destinée à devenir leur premier marché, comme c'est déjà le cas pour l'automobile.

Je me souviens de Tim Cook donnant des leçons de chose à un Président des États-Unis pourtant ouvertement pour le mariage gay.
Le même allant porter la bonne parole dans des pays où ses propres mœurs étaient criminalisé, et là ne l'ouvrant absolument pas. Les Droits de l'Homme? Après le business, un peu de sérieux!
Toujours le même donnant à la Chine les clés pour surveiller tout ce qui est stocké dans iCloud, qu'on peut maintenant appeler Xi Cloud...

Je vous avouerais que je suis choqué et énervé, j'avais même préparé un autre texte (trop brutal, trop engagé), car entre ce que Apple et Tim Cook disent et font, il y a une abysse, incroyable!

Selon le Financial Times, les dirigeants d'Apple ont considérablement accéléré leur recherche d'un successeur pour Tim Cook. Le but serait de le remplacer dès 2026.
Cette volonté de le remplacer au plus tôt ne serait pas liée à des contre-performances à venir ce qui est une bonne nouvelle. On ne connaît en revanche pas la raison profonde, aucune piste n'étant évoquée.

Nous ne vous cachons pas nous réjouir de voir du sang neuf à la tête d'Apple et ne serions pas surpris que le fiasco d'Apple Intelligence ne soit la raison profonde de cette volonté de confier les rênes de la société à quelqu'un d'autre.

Researchers in Austria have entangled matter-based qubits with photonic qubits in a ten-ion system. The technique is scalable to larger ion-qubit registers, paving the way for the creation of larger and more complex quantum networks.

Quantum networks consist of matter-based nodes that store and process quantum information and are linked through photons (quanta of light). Already, Ben Lanyon’s group at the University of Innsbruck has made advances in this direction by entangling two ions in different systems. Now, in a new paper published in Physical Review Letters , they describe how they have developed and demonstrated a new method to entangle a string of ten ions with photons. In the future, this approach could enable the entanglement of sets of ions in different locations through light, rather than one ion at a time.

To achieve this, Lanyon and colleagues trapped a chain of 10 calcium ions in a linear trap in an optical cavity. By changing the trapping voltages in the trap, each ion was moved, one-by-one, into the cavity. Once inside, the ion was placed in the “sweet spot”, where the ion’s interaction with the cavity is the strongest. There, the ion  emitted a single photon when exposed to a 393 nm Raman laser beam. This beam was tightly focused on one ion, guaranteeing that the emitted photon – collected in a single-mode optical fibre – comes out from one ion at a time. This process was carried out ten times, one per ion, to obtain a train of ten photons.

By using quantum state tomography, the researchers reconstructed the density matrix, which describes the correlation between the states of ions (i) and photons (j).  To do so, they measure every ion and photon state in three different basis, resulting in nine Pauli-basis configurations of quantum measurements. From the density matrix, the concurrence (a measure of entanglement) between the ion (i) and photon (j) was found to be positive only when  i = j, and equal to zero otherwise. This implies that the ion is uniquely entangled with the photon it produced, and unentangled with the photon produced by other ions.

From the density matrix, they also calculate the fidelity with the Bell state (a state of maximum entanglement), yielding an average 92%. As Marco Canteri points out, “this fidelity characterizes the quality of entanglement between the ion-photon pair for i=j”.

This work developed and demonstrated a technique whereby matter-based qubits and photonic qubits can be entangled, one  at a time, in ion strings.  Now, the group aims to “demonstrate universal quantum logic within the photon-interfaced 10-ion register and, building up towards entangling two remote 10-ion processors through the exchange of photons between them,” explains team member Victor Krutyanskiy. If this method effectively scales to larger systems, more complex quantum networks could be built. This would lead to applications in quantum communication and quantum sensing.

The post Ten-ion system brings us a step closer to large-scale qubit registers appeared first on Physics World.

Selon Mac Gurman, Apple se désintéresse de la gamme Mac Pro.
Ces machines qui n'ont pas été renouvelées depuis 2023 et toujours dotées de processeur M2 seraient considérées comme inutiles par la société qui préfère miser dans le futur sur le Mac Studio qui deviendrait le haut de gamme professionnel que ce soit en puissance de CPU qu'en dotation RAM.

Il y a quelques années nous aurions fustigé cette position. Aujourd'hui, avoir un Mac Pro n'a plus de réel intérêt surtout quand deux ans plus tard un MacBook Pro en a pratiquement la même puissance brute pour un prix bien moindre.
C'est d'ailleurs un des effets de biais des puces Apple Silicon et de leur évolution. Elles tendent à pousser l'achat de machines milieu de gamme quitte à les renouveler plus fréquemment plutôt que de prendre une machine bien plus coûteuse et ultra puissante qui n'aura un intérêt que dans des domaines très pointus et qui sera surclassée deux ans plus tard par une machine coûtant e tiers de son prix.

Measuring blood flow to the brain is essential for diagnosing and developing treatments for neurological disorders such as stroke, vascular dementia or traumatic brain injury. Performing this measurement non-invasively is challenging, however, and achieved predominantly using costly MRI and nuclear medicine imaging techniques.

Emerging as an alternative, modalities based on optical transcranial measurement are cost-effective and easy to use. In particular, speckle contrast optical spectroscopy (SCOS) – an offshoot of laser speckle contrast imaging, which uses laser light speckles to visualize blood vessels – can measure cerebral blood flow (CBF) with high temporal resolution, typically above 30 Hz, and cerebral blood volume (CBV) through optical signal attenuation.

Researchers at the California Institute of Technology (Caltech) and the Keck School of Medicine’s USC Neurorestoration Center have designed a lightweight SCOS system that accurately measures blood flow to the brain, distinguishing it from blood flow to the scalp. Co-senior author Charles Liu of the Keck School of Medicine and team describe the system and their initial experimentation with it in APL Bioengineering.

The SCOS system consists of a 3D-printed head mount designed for secure placement over the temple region. It holds a single 830 nm laser illumination fibre and seven detector fibres positioned at seven different source-to-detector (S–D) distances (between 0.6 and 2.6 cm) to simultaneously capture blood flow dynamics across layers of the scalp, skull and brain. Fibres with shorter S–D distances acquire shallower optical data from the scalp, while those with greater distances obtain deeper and broader data. The seven channels are synchronized to exhibit identical oscillation frequencies corresponding to the heart rate and cardiac cycle.

When the SCOS system directs the laser light onto a sample, multiple random scattering events occur before the light exits the sample, creating speckles. These speckles, which materialize on rapid timescales, are the result of interference of light travelling along different trajectories. Movement within the sample (of red blood cells, for instance) causes dynamic changes in the speckle field. These changes are captured by a multi-million-pixel camera with a frame rate above 30 frames/s and quantified by calculating the speckle contrast value for each image.

Human testing

The researchers used the SCOS system to perform CBF and CBV measurements in 20 healthy volunteers. To isolate and obtain surface blood dynamics from brain signals, the researchers gently pressed on the superficial temporal artery (a terminal branch of the external carotid artery that supplies blood to the face and scalp) to block blood flow to the scalp.

In tests on the volunteers, when temporal artery blood flow was occluded for 8 s, scalp-sensitive channels exhibited significant decreases in blood flow while brain-sensitive channels showed minimal change, enabling signals from the internal carotid artery that supplies blood to the brain to be clearly distinguished. Additionally, the team found that positioning the detector 2.3 cm or more away from the source allowed for optimal brain blood flow measurement while minimizing interference from the scalp.

“Combined with the simultaneous measurements at seven S–D separations, this approach enables the first quantitative experimental assessment of how scalp and brain signal contributions vary with depth in SCOS-based CBF measurements and, more broadly, in optical measurements,” they write. “This work also provides crucial insights into the optimal device S–D distance configuration for preferentially probing brain signal over scalp signal, with a practical and subject-friendly alternative for evaluating depth sensitivity, and complements more advanced, hardware-intensive strategies such as time-domain gating.”

The researchers are now working to improve the signal-to-noise ratio of the system. They plan to introduce a compact, portable laser and develop a custom-designed extended camera that spans over 3 cm in one dimension, enabling simultaneous and continuous measurement of blood dynamics across S–D distances from 0.5 to 3.5 cm. These design advancements will enhance spatial resolution and enable deeper brain measurements.

“This crucial step will help transition the system into a compact, wearable form suitable for clinical use,” comments Liu. “Importantly, the measurements described in this publication were achieved in human subjects in a very similar manner to how the final device will be used, greatly reducing barriers to clinical application.”

“I believe this study will advance the engineering of SCOS systems and bring us closer to a wearable, clinically practical device for monitoring brain blood flow,” adds co-author Simon Mahler, now at Stevens Institute of Technology. “I am particularly excited about the next stage of this project: developing a wearable SCOS system that can simultaneously measure both scalp and brain blood flow, which will unlock many fascinating new experiments.”

The post Non-invasive wearable device measures blood flow to the brain appeared first on Physics World.

Rendering VT-100 émulant l'affichage DOS

J'écris des moteurs d'échecs, ou d'autres comme pour le Reversi/Othello, depuis plusieurs décennies.
Je suis en train de travailler sur un moteur sorti fin 1994 pour le Minitel, écrit en C-Ansi, Borland C 3.1 pour DOS à l'époque et gcc maintenant pour Posix dans un terminal VT-100, 32 bits pour Pentium [1] à l'époque et 64 bits pour x86 et ARM aujourd'hui.
Du vieux code monothread, c'est à noter.

Et pour m'amuser, sur mon Mac mini M4, je ne fais non pas tourner une version compilée pour ARM mais celle compilée pour x86 64 bits. Le même exact code, grâce à Rosetta 2.
Ce code x86 sort de gcc, avec l'option -O3 : optimisé mais en limitant les risques.

Je viens de m'apercevoir que si en moyenne un cœur Performance du M4 est 3,3 fois plus rapide que chaque cœur de mon MBP 15" 2017, ce qui devrait amener ce code x86 monothread à tourner moins de 3 fois plus vite sous ARM via Rosetta 2 (85% d'efficacité mesurée en 2020), il tourne en fait 4 fois plus vite maintenant en 2025!

Je soupçonne que Rosetta 2 s'est vu amélioré et pourrait utiliser les 16 registres généraux en plus de l'architecture ARM 64 bits pour optimiser le code x86 quand il le transcrit pour nos Mac Apple Silicon.
Ce code simple (non-vectoriel, non fpu) tourne 30% plus vite qu'il le devrait !

Une limitation du Amd64 (ou x86 64 bit, ou x86-64) créé par AMD, c'est que son architecture ne contient que 16 registres généraux (entiers et pointeurs), contre 32 pour ARM 64bits (AArch64).

Ça permet évidemment à nos Mac ARM d'émuler facilement les 16 registres du x86, mais en laisse beaucoup inutilisés, et je pense que Rosetta 2 en profite pour limiter les relectures après écriture dans la pile (les écritures étant probablement conservées).

Une autre optimisation peut être d'éliminer les copies de registres inutiles, puisque en ARM 64 bits les opérations ont souvent 3 opérandes (2 sources et 1 destination), contre 2 en x86 64 bits (1 source, 1 destination).
Par exemple a=b+c en x86 64 bits consiste à "copier" (créer un alias) b dans a puis à lui ajouter c (2 opérations), quand en ARM 64 bits, on ajoute b et c dans a en une seule opération.

L'aliasing de registres (ex-copie) est "gratuite" en x86 maintenant, mais nécessite de décoder une instruction de plus, donc plus de charge coté décodeur (front end) qui est déjà un problème en x86 avec des instructions de longueur variable et le travail pour créer les micro-instructions et macro-instructions depuis les Core™ 2.

Je vais vous revenir avec une comparaison avec du code ARM natif (même compilo mêmes options). Mais ce résultat est très surprenant, Apple fait un travail incroyable sur Rosetta 2 !

VR puzzle adventure Tin Hearts will bring its first act to Quest in February.

Developed by Rogue Sun and IPHIGAMES, Tin Hearts is a Lemmings-style game that explores the story of a fictional Victorian inventor, Albert Butterworth. Guiding toy soldiers through this Dickensian world with block-based puzzles, VR support arrived in a post-launch update on PS VR2 and Steam last year. Originally targeting a December 11 launch, that's now been delayed to February 12, 2026.

Detailed in a press release, publisher Wired Productions calls Act 1 a standalone episode where these tiny soldiers are appropriately dressed for the festive season in an attic filled with toys. Costing $5.99 for the first part, the publisher previously stated Acts 2, 3, and 4 will follow “in the coming weeks” on Quest. No specific release dates have been confirmed yet.

Originally released through a now delisted PC VR prologue on PC VR in 2018, we had positive impressions in our Tin Hearts VR preview two years ago. Stating it offers “some well-considered mechanics” that caught our attention, we believed it provides “enjoyable puzzles and an intriguing whimsical setting.”

Tin Hearts is out now in full on flatscreen platforms, PS VR2, and PC VR. Act 1 arrives on the Meta Quest platform on February 12, 2026.

As we enter the final stretch of the International Year of Quantum Science and Technology (IYQ), I hope you’ve enjoyed our extensive quantum coverage over the last 12 months. We’ve tackled the history of the subject, explored some of the unexplained mysteries that still make quantum physics so exciting, and examined many of the commercial applications of quantum technology. You can find most of our coverage collected into two free-to-read digital Quantum Briefings, available here and here on the Physics World website.

Over the last 100 years since Werner Heisenberg first developed quantum mechanics on the island of Helgoland in June 1925, quantum mechanics has proved to be an incredibly powerful, successful and logically consistent theory. Our understanding of the subatomic world is no longer the “lamentable hodgepodge of hypotheses, principles, theorems and computational recipes”, as the Israeli physicist and philosopher Max Jammer memorably once described it.

In fact, quantum mechanics has not just transformed our understanding of the natural world; it has immense practical ramifications too, with so-called “quantum 1.0” technologies – lasers, semiconductors and electronics – underpinning our modern world. But as was clear from the UK National Quantum Technologies Showcase in London last week, organized by Innovate UK, the “quantum 2.0” revolution is now in full swing.

The day-long event, which is now in its 10th year, featured over 100 exhibitors, including many companies that are already using fundamental quantum concepts such as entanglement and superposition to support the burgeoning fields of quantum computing, quantum sensing and quantum communication. The show was attended by more than 3000 delegates, some of whom almost had to be ushered out of the door at closing time, so keen were they to keep talking.

Last week also saw a two-day conference at the historic Royal Institution (RI) in central London that was a centrepiece of IYQ in the UK and Ireland. Entitled Quantum Science and Technology: the First 100 Years; Our Quantum Future and attended by over 300 people, it was organized by the History of Physics and the Business Innovation and Growth groups of the Institute of Physics (IOP), which publishes Physics World.

The first day, focusing on the foundations of quantum mechanics, ended with a panel discussion – chaired by my colleague Tushna Commissariat and Daisy Shearer from the UK’s National Quantum Computing Centre – with physicists Fay Dowker (Imperial College), Jim Al-Khalili  (University of Surrey) and Peter Knight. They talked about whether the quantum wavefunction provides a complete description of physical reality, prompting much discussion with the audience. As Al-Khalili wryly noted, if entanglement has emerged as the fundamental feature of quantum reality, then “decoherence is her annoying and ever-present little brother”.

Knight, meanwhile, who is a powerful figure in quantum-policy circles, went as far as to say that the limit of decoherence – and indeed the boundary between the classical and quantum worlds – is not a fixed and yet-to-be revealed point. Instead, he mused, it will be determined by how much money and ingenuity and time physicists have at their disposal.

On the second day of the IOP conference at the RI, I chaired a discussion that brought together four future leaders of the subject: Mehul Malik (Heriot-Watt University) and Sarah Malik (University College London) along with industry insiders Nicole Gillett (Riverlane) and Muhammad Hamza Waseem (Quantinuum).

As well as outlining the technical challenges in their fields, the speakers all stressed the importance of developing a “skills pipeline” so that the quantum sector has enough talented people to meet its needs. Also vital will be the need to communicate the mysteries and potential of quantum technology – not just to the public but to industrialists, government officials and venture capitalists. By many measures, the UK is at the forefront of quantum tech – and it is a lead it should not let slip.

The week ended with Al-Khalili giving a public lecture, also at the Royal Institution, entitled “A new quantum world: ‘spooky’ physics to tech revolution”. It formed part of the RI’s famous Friday night “discourses”, which this year celebrate their 200th anniversary. Al-Khalili, who also presents A Life Scientific on BBC Radio 4, is now the only person ever to have given three RI discourses.

After the lecture, which was sold out, he took part in a panel discussion with Knight and Elizabeth Cunningham, a former vice-president for membership at the IOP. Al-Khalili was later presented with a special bottle of “Glentanglement” whisky made by Glasgow-based Fraunhofer UK for the Scottish Quantum Technology cluster.

The post The future of quantum physics and technology debated at the Royal Institution appeared first on Physics World.

Significant progress towards answering one of the Clay Mathematics Institute’s seven Millennium Prize Problems has been achieved using deep learning. The challenge is to establish whether or not the Navier-Stokes equation of fluid dynamics develops singularities. The work was done by researchers in the US and UK – including some at Google Deepmind. Some team members had already shown that simplified versions of the equation could develop stable singularities, which reliably form. In the new work, the researchers found unstable singularities, which form only under very specific conditions.

The Navier–Stokes partial differential equation was developed in the early 19th century by Claude-Louis Navier and George Stokes. It has proved its worth for modelling incompressible fluids in scenarios including water flow in pipes; airflow around aeroplanes; blood moving in veins; and magnetohydrodynamics in plasmas.

No-one has yet proved, however, whether smooth, non-singular solutions to the equation always exist in three dimensions. “In the real world, there is no singularity…there is no energy going to infinity,” says fluid dynamics expert Pedram Hassanzadeh of the University of Chicago. “So if you have an equation that has a singularity, it tells you that there is some physics that is missing.” In 2000, the Clay Mathematics Institute in Denver, Colorado listed this proof as one of seven key unsolved problems in mathematics, offering a reward of $1,000,000 for an answer.

Computational approaches

Researchers have traditionally tackled the problem analytically, but in recent decades high-level computational simulations have been used to assist in the search. In a 2023 paper, mathematician Tristan Buckmaster of New York University and colleagues used a special type of machine learning algorithm called a physics-informed neural network to address the question.

“The main difference is…you represent [the solution] in a highly non-linear way in terms of a neural network,” explains Buckmaster. This allows it to occupy a lower-dimensional space with fewer free parameters, and therefore to be optimized more efficiently. Using this approach, the researchers successfully obtained the first stable singularity in the Euler equation. This is an analogy to the Navier-Stokes equation that does not include viscosity.

A stable singularity will still occur if the initial conditions of the fluid are changed slightly – although the time taken for them to form may be altered. An unstable singularity, however, may never occur if the initial conditions are perturbed even infinitesimally. Some researchers have hypothesized that any singularities in the Navier-Stokes equation must be unstable, but finding unstable singularities in a computer model is extraordinarily difficult.

“Before our result there hadn’t been an unstable singularity for an incompressible fluid equation found numerically,” says geophysicist Ching-Yao Lai of California’s Stanford University.

Physics-informed neural network

In the new work the authors of the original paper and others teamed up with researchers at Google Deepmind to search for unstable singularities in a bounded 3D version of the Euler equation using a physics-informed neural network. “Unlike conventional neural networks that learn from vast datasets, we trained our models to match equations that model the laws of physics,” writes Yongji Wang of New York University and Stanford on Deepmind’s blog. “The network’s output is constantly checked against what the physical equations expect, and it learns by minimizing its ‘residual’, the amount by which its solution fails to satisfy the equations.”

After an exhaustive search at a precision that is orders of magnitude higher than a normal deep learning protocol, the researchers discovered new families of singularities in the 3D Euler equation. They also found singularities in the related incompressible porous media equation used to model fluid flows in soil or rock; and in the Boussinesq equation that models atmospheric flows.

The researchers also gleaned insights into the strength of the singularities. This could be important as stronger singularities might be less readily smoothed out by viscosity when moving from the Euler equation to the Navier-Stokes equation. The researchers are now seeking to model more open systems to study the problem in a more realistic space.

Hassanzadeh, who was not involved in the work, believes that it is significant – although the results are not unexpected. “If the Euler equation tells you that ‘Hey, there is a singularity,’ it just tells you that there is physics that is missing and that physics becomes very important around that singularity,” he explains. “In the case of Euler we know that you get the singularity because, at the very smallest scales, the effects of viscosity become important…Finding a singularity in the Euler equation is a big achievement, but it doesn’t answer the big question of whether Navier-Stokes is a representation of the real world, because for us Navier-Stokes represents everything.”

He says the extension to studying the full Navier-Stokes equation will be challenging but that “they are working with the best AI people in the world at Deepmind,” and concludes “I’m sure it’s something they’re thinking about”.

The work is available on the arXiv pre-print server.

The post Neural networks discover unstable singularities in fluid systems appeared first on Physics World.

NASA’s Goddard Space Flight Center (GSFC) looks set to lose a big proportion of its budget as a two-decade reorganization plan for the centre is being accelerated. The move, which is set to be complete by March, has left the Goddard campus with empty buildings and disillusioned employees. Some staff even fear that the actions during the 43-day US government shutdown, which ended on 12 November, could see the end of much of the centre’s activities.

Based in Greenbelt, Maryland, the GSFC has almost 10 000 scientists and engineers, about 7000 of whom are directly employed by NASA contractors. Responsible for many of NASA’s most important uncrewed missions, telescopes, and probes, the centre is currently working on the Nancy Grace Roman Space Telescope, which is scheduled to launch in 2027, as well as the Dragonfly mission that is due to head for Saturn’s largest moon Titan in 2028.

The ability to meet those schedules has now been put in doubt by the Trump administration’s proposed budget for financial year 2026, which started in September. It calls for NASA to receive almost $19bn – far less than the $25bn it has received for the past two years. If passed, Goddard would lose more than 42% of its staff.

Congress, which passes the final budget, is not planning to cut NASA so deeply as it prepares its 2026 budget proposal. But on 24 September, Goddard managers began what they told employees was “a series of moves…that will reduce our footprint into fewer buildings”. The shift is intended to “bring down overall operating costs while maintaining the critical facilities we need for our core capabilities of the future”.

While this is part of a 20-year “master plan” for the GSFC that NASA’s leadership approved in 2019, the management’s memo stated that “all planned moves will take place over the next several months and be completed by March 2026″. A report in September by Democratic members of the Senate Committee on Commerce, Science, and Transportation, which is responsible for NASA, asserts that the cuts are “in clear violation of the [US] constitution [without] regard for the impacts on NASA’s science missions and workforce”.

On 3 November, the Goddard Engineers, Scientists and Technicians Association, a union representing NASA workers, reported that the GSFC had already closed over a third of its buildings, including some 100 labs. This had been done, it says, “with extreme haste and with no transparent strategy or benefit to NASA or the nation”. The union adds that the “closures are being justified as cost-saving but no details are being provided and any short-term savings are unlikely to offset a full account of moving costs and the reduced ability to complete NASA missions”.

Accounting for the damage

Zoe Lofgren, the lead Democrat on the House of Representatives Science Committee, has demanded of Sean Duffy, NASA’s acting administrator, that the agency “must now halt” any laboratory, facility and building closure and relocation activities at Goddard. In a  letter to Duffy dated 10 November, she also calls for the “relocation, disposal, excessing, or repurposing of any specialized equipment or mission-related activities, hardware and systems” to also end immediately.

Lofgren now wants NASA to carry out a “full accounting of the damage inflicted on Goddard thus far” by 18 November. Owing to the government shutdown, no GSFC or NASA official was available to respond to Physics World’s requests for a response.

Meanwhile, the Trump administration has renominated billionaire entrepreneur Jared Isaacman as NASA’s administrator. Trump had originally nominated Isaacman, who had flown on a private SpaceX mission and carried out spacewalk, on the recommendation of SpaceX founder Elon Musk. But the administration withdrew the nomination in May following concerns among some Republicans that Isaacman had funded the Democrat party.

The post NASA’s Goddard Space Flight Center hit by significant downsizing appeared first on Physics World.

Like any major endeavour, designing and fabricating semiconductor chips requires compromise. As well as trade-offs between cost and performance, designers also consider carbon emissions and other environmental impacts.

In this episode of the Physics World Weekly podcast, Margaret Harris reports from the Heidelberg Laureate Forum where she spoke to two researchers who are focused on some of these design challenges.

Up first is Mariam Elgamal, who’s doing a PhD at Harvard University on the development of environmentally sustainable computing systems. She explains why sustainability goes well beyond energy efficiency and must consider the manufacturing process and the chemicals used therein.

Harris also chats with Andrew Gunter, who is doing a PhD at the University of British Columbia on circuit design for computer chips. He talks about the maths-related problems that must be solved in order to translate a desired functionality into a chip that can be fabricated.

The post Designing better semiconductor chips: NP hard problems and forever chemicals appeared first on Physics World.

Positron emission tomography (PET) is used extensively within preclinical research, enabling molecular imaging of rodent brains, for example, to investigate neurodegenerative disease. Such imaging studies require the highest possible spatial resolution to resolve the tiny structures in the animal’s brain. A research team at the National Institutes for Quantum Science and Technology (QST) in Japan has now developed the first PET scanner to achieve sub-0.5 mm spatial resolution.

Submillimetre-resolution PET has been demonstrated by several research groups. Indeed, the QST team previously built a PET scanner with 0.55 mm resolution – sufficient to visualize the thalamus and hypothalamus in the mouse brain. But identification of smaller structures such as the amygdala and cerebellar nuclei has remained a challenge.

“Sub-0.5 mm resolution is important to visualize mouse brain structures with high quantification accuracy,” explains first author Han Gyu Kang. “Moreover, this research work will change our perspective about the fundamental limit of PET resolution, which had been regarded to be around 0.5 mm due to the positron range of [the radioisotope] fluorine-18”.

System optimization

With Monte Carlo simulations revealing that sub-0.5 mm resolution could be achievable with optimal detector parameters and system geometry, Kang and colleagues performed a series of modifications to their submillimetre-resolution PET (SR-PET) to create the new high-resolution PET (HR-PET) scanner.

The HR-PET, described in IEEE Transactions on Medical Imaging, is based around two 48 mm-diameter detector rings with an axial coverage of 23.4 mm. Each ring contains 16 depth-of-interaction (DOI) detectors (essential to minimize parallax error in a small ring diameter) made from three layers of LYSO crystal arrays stacked in a staggered configuration, with the outer layer coupled to a silicon photomultiplier (SiPM) array.

Compared with their previous design, the researchers reduced the detector ring diameter from 52.5 to 48 mm, which served to improve geometrical efficiency and minimize the noncollinearity effect. They also reduced the crystal pitch from 1.0 to 0.8 mm and the SiPM pitch from 3.2 to 2.4 mm, improving the spatial resolution and crystal decoding accuracy, respectively.

Other changes included optimizing the crystal thicknesses to 3, 3 and 5 mm for the first, second and third arrays, as well as use of a narrow energy window (440–560 keV) to reduce the scatter fraction and inter-crystal scattering events. “The optimized staggered three-layer crystal array design is also a key factor to enhance the spatial resolution by improving the spatial sampling accuracy and DOI resolution compared with the previous SR-PET,” Kang points out.

Performance tests showed that the HR-PET scanner had a system-level energy resolution of 18.6% and a coincidence timing resolution of 8.5 ns. Imaging a NEMA ²²Na point source revealed a peak sensitivity at the axial centre of 0.65% for the 440–560 keV energy window and a radial resolution of 0.67±0.06 mm from the centre to 10 mm radial offset (using 2D filtered-back-projection reconstruction) – a 33% improvement over that achieved by the SR-PET.

To further evaluate the performance of the HR-PET, the researchers imaged a rod-based resolution phantom. Images reconstructed using a 3D ordered-subset-expectation-maximization (OSEM) algorithm clearly resolved all of the rods. This included the smallest rods with diameters of 0.5 and 0.45 mm, with average valley-to-peak ratios of 0.533 and 0.655, respectively – a 40% improvement over the SR-PET.

In vivo brain PET

The researchers then used the HR-PET for in vivo mouse brain imaging. They injected ¹⁸F-FITM, a tracer used to image the central nervous system, into an awake mouse and performed a 30 min PET scan (with the animal anesthetized) 42 min after injection. For comparison, they scanned the same mouse for 30 min with a preclinical Inveon PET scanner.

After OSEM reconstruction, strong tracer uptake in the thalamus, hypothalamus, cerebellar cortex and cerebellar nuclei was clearly visible in the coronal HR-PET images. A zoomed image distinguished the cerebellar nuclei and flocculus, while sagittal and axial images visualized the cortex and striatum. Images from the Inveon, however, could barely resolve these brain structures.

The team also imaged the animal’s glucose metabolism using the tracer ¹⁸F-FDG. A 30 min HR-PET scan clearly delineated glucose transporter expression in the cortex, thalamus, hypothalamus and cerebellar nuclei. Here again, the Inveon could hardly identify these small structures.

The researchers note that the ¹⁸F-FITM and ¹⁸F-FDG PET images matched well with the anatomy seen in a preclinical CT scan. “To the best of our knowledge, this is the first separate identification of the hypothalamus, amygdala and cerebellar nuclei of mouse brain,” they write.

Future plans for the HR-PET scanner, says Kang, include using it for research on neurodegenerative disorders, with tracers that bind to amyloid beta or tau protein. “In addition, we plan to extend the axial coverage over 50 mm to explore the whole body of mice with sub-0.5 mm resolution, especially for oncological research,” he says. “Finally, we would like to achieve sub-0.3 mm PET resolution with more optimized PET detector and system designs.”

The post High-resolution PET scanner visualizes mouse brain structures with unprecedented detail appeared first on Physics World.

Static electricity is an everyday phenomenon, but it remains poorly understood. Researchers at the Institute of Science and Technology Austria (ISTA) have now shed new light on it by capturing an “image” of charge distributions as charge transfers from one surface to another. Their conclusions challenge longstanding interpretations of previous experiments and enhance our understanding of how charge behaves on insulating surfaces.

Static electricity is also known as contact electrification because it occurs when charge is transferred from one object to another by touch. The most common laboratory example involves rubbing a balloon on someone’s head to make their hair stand on end. However, static electricity is also associated with many other activities, including coffee grinding, pollen transport and perhaps even the formation of rocky planets.

One of the most useful ways of studying contact electrification is to move a metal tip slowly over the surface of a sample without touching it, recording a voltage all the while. These so-called scanning Kelvin methods produce an “image” of voltages created by the transferred charge. At the macroscale, around 100 μm to 10 cm, the main method is termed scanning Kelvin probe microscopy (SKPM). At the nanoscale, around 10  nm to 100  μm, a related but distinct variant known as Kelvin probe force microscopy (KPFM) is used instead.

In previous fundamental physics studies using these techniques, the main challenges have been to make sense of the stationary patterns of charge left behind after contact electrification, and to investigate how these patterns evolve over space and time. In the latest work, the ISTA team chose to ask a slightly different question: when are the dynamics of charge transfer too fast for measured stationary patterns to yield meaningful information?

Mapping the charge on the contact-electrified surface of a polymer film

To find out, ISTA PhD student Felix Pertl built a special setup that could measure a sample’s surface charge with KPFM; transfer it below a linear actuator so that it could exchange charge when it contacted another material; and then transfer it underneath the KPFM again to image the resulting change in the surface charge.

“In a typical set-up, the sample transfer, moving the AFM to the right place and reinitiation and recalibration of the KPFM parameters can easily take as long as tens of minutes,” Pertl explains. “In our system, this happens in as little as around 30 s. As all aspects of the system are completely automated, we can repeat this process, and quickly, many times.”

This speed-up is important because static electricity dissipates relatively rapidly. In fact, the researchers found that the transferred charge disappeared from the sample’s surface quicker than the time required for most KPFM scans. Their data also revealed that the deposited charge was, in effect, uniformly distributed across the surface and that its dissipation depended on the material’s electrical conductivity. Additional mathematical modelling and subsequent experiments confirmed that the more insulating a material is, the slower it dissipates charge.

Surface heterogeneity likely not a feature of static electricity

Pertl says that these results call into question the validity of some previous static electricity studies that used KPFM to study charge transfer. “The most influential paper in our field to date reported surface charge heterogeneity using KPFM,” he tells Physics World. At first, the ISTA team’s goal was to understand the origin of this heterogeneity. But when their own experiments showed an essentially homogenous distribution of surface charge, the researchers had to change tack.

“The biggest challenge in our work was realizing – and then accepting – that we could not reproduce the results from this previous study,” Pertl says. “Convincing both my principal investigator and myself that our data revealed a very different physical mechanism required patience, persistence and trust in our experimental approach.”

The discrepancy, he adds, implies that the surface heterogeneity previously observed was likely not a feature of static electricity, as was claimed. Instead, he says, it was probably “an artefact of the inability to image the charge before it had left the sample surface”.

A historical precedent

Studies of contact electrification studies go back a long way. Philippe Molinié of France’s GeePs Laboratory, who was not involved in this work, notes that the first experiments were performed by the English scientist William Gilbert clear back in the sixteenth century. As well as coining the term “electricity” (from the Greek “elektra”, meaning amber), Gilbert was also the first to establish that magnets maintain their electrical attraction over time, while the forces produced by contact-charged insulators slowly decrease.

“Four centuries later, many mysteries remain unsolved in the contact electrification phenomenon,” Molinié observes. He adds that the surfaces of insulating materials are highly complex and usually strongly disordered, which affects their ability to transfer charge at the molecular scale. “The dynamics of the charge neutralization, as Pertl and colleagues underline, is also part of the process and is much more complex than could be described by a simple resistance-capacitor model,” Molinié says.

Although the ISTA team studied these phenomena with sophisticated Kelvin probe microscopy rather than the rudimentary tools available to Gilbert, it is, Molinié says, “striking that the competition between charge transfer and charge screening that comes from the conductivity of an insulator, first observed by Gilbert, is still at the very heart of the scientific interrogations that this interesting new work addresses.”

“A more critical interpretation”

The Austrian researchers, who detail their work in Phys. Rev. Lett., say they hope their experiments will “encourage a more critical interpretation” of KPFM data in the future, with a new focus on the role of sample grounding and bulk conductivity in shaping observed charge patterns. “We hope it inspires KPFM users to reconsider how they design and analyse experiments, which could lead to more accurate insights into charge behaviour in insulators,” Pertl says.

“We are now planning to deliberately engineer surface charge heterogeneity into our samples,” he reveals. “By tuning specific surface properties, we aim to control the sign and spatial distribution of charge on defined regions of these.”

The post New experiments on static electricity cast doubt on previous studies in the field appeared first on Physics World.

“Global collaborations for European economic resilience” is the theme of  SEMICON Europa 2025. The event is coming to Munich, Germany on 18–21 November and it will attract 25,000 semiconductor professionals who will enjoy presentations from over 200 speakers.

The TechARENA portion of the event will cover a wide range of technology-related issues including new materials, future computing paradigms and the development of hi-tech skills in the European workface. There will also be an Executive Forum, which will feature leaders in industry and government and will cover topics including silicon geopolitics and the use of artificial intelligence in semiconductor manufacturing.

SEMICON Europa will be held at the Messe München, where it will feature a huge exhibition with over 500 exhibitors from around the world. The exhibition is spread out over three halls and here are some of the companies and product innovations to look out for on the show floor.

Accelerating the future of electro-photonic integration with SmarAct

As the boundaries between electronic and photonic technologies continue to blur, the semiconductor industry faces a growing challenge: how to test and align increasingly complex electro-photonic chip architectures efficiently, precisely, and at scale. At SEMICON Europa 2025, SmarAct will address this challenge head-on with its latest innovation – Fast Scan Align. This is a high-speed and high-precision alignment solution that redefines the limits of testing and packaging for integrated photonics.

In the emerging era of heterogeneous integration, electronic and photonic components must be aligned and interconnected with sub-micrometre accuracy. Traditional positioning systems often struggle to deliver both speed and precision, especially when dealing with the delicate coupling between optical and electrical domains. SmarAct’s Fast Scan Align solution bridges this gap by combining modular motion platforms, real-time feedback control, and advanced metrology into one integrated system.

At its core, Fast Scan Align leverages SmarAct’s electromagnetic and piezo-driven positioning stages, which are capable of nanometre-resolution motion in multiple degrees of freedom. Fast Scan Align’s modular architecture allows users to configure systems tailored to their application – from wafer-level testing to fibre-to-chip alignment with active optical coupling. Integrated sensors and intelligent algorithms enable scanning and alignment routines that drastically reduce setup time while improving repeatability and process stability.

Fast Scan Align’s compact modules allow various measurement techniques to be integrated with unprecedented possibilities. This has become decisive for the increasing level of integration of complex electro-photonic chips.

Apart from the topics of wafer-level testing and packaging, wafer positioning with extreme precision is as crucial as never before for the highly integrated chips of the future. SmarAct’s PICOSCALE interferometer addresses the challenge of extreme position by delivering picometer-level displacement measurements directly at the point of interest.

When combined with SmarAct’s precision wafer stages, the PICOSCALE interferometer ensures highly accurate motion tracking and closed-loop control during dynamic alignment processes. This synergy between motion and metrology gives users unprecedented insight into the mechanical and optical behaviour of their devices – which is a critical advantage for high-yield testing of photonic and optoelectronic wafers.

Visitors to SEMICON Europa will also experience how all of SmarAct’s products – from motion and metrology components to modular systems and up to turn-key solutions – integrate seamlessly, offering intuitive operation, full automation capability, and compatibility with laboratory and production environments alike.

For more information visit SmarAct at booth B1.860 or explore more of SmarAct’s solutions in the semiconductor and photonics industry.

Optimized pressure monitoring: Efficient workflows with Thyracont’s VD800 digital compact vacuum meters

Thyracont Vacuum Instruments will be showcasing its precision vacuum metrology systems in exhibition hall C1. Made in Germany, the company’s broad portfolio combines diverse measurement technologies – including piezo, Pirani, capacitive, cold cathode, and hot cathode – to deliver reliable results across a pressure range from 2000 to 3e-11 mbar.

Front-and-centre at SEMICON Europa will be Thyracont’s new series of VD800 compact vacuum meters. These instruments provide precise, on-site pressure monitoring in industrial and research environments. Featuring a direct pressure display and real-time pressure graphs, the VD800 series is ideal for service and maintenance tasks, laboratory applications, and test setups.

The VD800 series combines high accuracy with a highly intuitive user interface. This delivers real-time measurement values; pressure diagrams; and minimum and maximum pressure – all at a glance. The VD800’s 4+1 membrane keypad ensures quick access to all functions. USB-C and optional Bluetooth LE connectivity deliver seamless data readout and export. The VD800’s large internal data logger can store over 10 million measured values with their RTC data, with each measurement series saved as a separate file.

Data sampling rates can be set from 20 ms to 60 s to achieve dynamic pressure tracking or long-term measurements. Leak rates can be measured directly by monitoring the rise in pressure in the vacuum system. Intelligent energy management gives the meters extended battery life and longer operation times. Battery charging is done conveniently via USB-C.

The vacuum meters are available in several different sensor configurations, making them adaptable to a wide range of different uses. Model VD810 integrates a piezo ceramic sensor for making gas-type-independent measurements for rough vacuum applications. This sensor is insensitive to contamination, making it suitable for rough industrial environments. The VD810 measures absolute pressure from 2000 to 1 mbar and relative pressure from −1060 to +1200 mbar.

Model VD850 integrates a piezo/Pirani combination sensor, which delivers high resolution and accuracy in the rough and fine vacuum ranges. Optimized temperature compensation ensures stable measurements in the absolute pressure range from 1200 to 5e-5 mbar and in the relative pressure range from −1060 to +340 mbar.

The model VD800 is a standalone meter designed for use with Thyracont’s USB-C vacuum transducers, which are available in two models. The VSRUSB USB-C transducer is a piezo/Pirani combination sensor that measures absolute pressure in the 2000 to 5.0e-5 mbar range. The other is the VSCUSB USB-C transducer, which measures absolute pressures from 2000 down to 1 mbar and has a relative pressure range from -1060 to +1200 mbar. A USB-C cable connects the transducer to the VD800 for quick and easy data retrieval. The USB-C transducers are ideal for hard-to-reach areas of vacuum systems. The transducers can be activated while a process is running, enabling continuous monitoring and improved service diagnostics.

With its blend of precision, flexibility, and ease of use, the Thyracont VD800 series defines the next generation of compact vacuum meters. The devices’ intuitive interface, extensive data capabilities, and modern connectivity make them an indispensable tool for laboratories, service engineers, and industrial operators alike.

To experience the future of vacuum metrology in Munich, visit Thyracont at SEMICON Europa hall C1, booth 752. There you will discover how the VD800 series can optimize your pressure monitoring workflows.

The post SEMICON Europa 2025 presents cutting-edge technology for semiconductor R&D and production appeared first on Physics World.

IOP Publishing’s Machine Learning series is the world’s first open-access journal series dedicated to the application and development of machine learning (ML) and artificial intelligence (AI) for the sciences.

Part of the series is Machine Learning: Science and Technology, launched in 2019, which bridges the application and advances in machine learning across the sciences. Machine Learning: Earth is dedicated to the application of ML and AI across all areas of Earth, environmental and climate sciences while Machine Learning: Health covers healthcare, medical, biological, clinical and health sciences and Machine Learning: Engineeringfocuses on applied AI and non-traditional machine learning to the most complex engineering challenges.

Here, the editors-in-chief (EiC) of the four journals discuss the growing importance of machine learning and their plans for the future.

Kyle Cranmer is a particle physicist and data scientist at the University of Wisconsin-Madison and is EiC of Machine Learning: Science and Technology (MLST). Pierre Gentine is a geophysicist at Columbia University and is EiC of Machine Learning: Earth. Jimeng Sun is a biophysicist at the University of Illinois at Urbana-Champaign and is EiC of Machine Learning: Health. Mechanical engineer Jay Lee is from the University of Maryland and is EiC of Machine Learning: Engineering.

What do you attribute to the huge growth over the past decade in research into and using machine learning?

Kyle Cranmer (KC): It is due to a convergence of multiple factors. The initial success of deep learning was driven largely by benchmark datasets, advances in computing with graphics processing units, and some clever algorithmic tricks. Since then, we’ve seen a huge investment in powerful, easy-to-use tools that have dramatically lowered the barrier to entry and driven extraordinary progress.

Pierre Gentine (PG): Machine learning has been transforming many fields of physics, as it can accelerate physics simulation, better handle diverse sources of data (multimodality), help us better predict.

Jimeng Sun (JS): Over the past decade, we have seen machine learning models consistently reach — and in some cases surpass — human-level performance on real-world tasks. This is not just in benchmark datasets, but in areas that directly impact operational efficiency and accuracy, such as medical imaging interpretation, clinical documentation, and speech recognition. Once ML proved it could perform reliably at human levels, many domains recognized its potential to transform labour-intensive processes.

Jay Lee (JL):  Traditionally, ML growth is based on the development of three elements: algorithms, big data, and computing.  The past decade’s growth in ML research is due to the perfect storm of abundant data, powerful computing, open tools, commercial incentives, and groundbreaking discoveries—all occurring in a highly interconnected global ecosystem.

What areas of machine learning excite you the most and why?

KC: The advances in generative AI and self-supervised learning are very exciting. By generative AI, I don’t mean Large Language Models — though those are exciting too — but probabilistic ML models that can be useful in a huge number of scientific applications. The advances in self-supervised learning also allows us to engage our imagination of the potential uses of ML beyond well-understood supervised learning tasks.

PG: I am very interested in the use of ML for climate simulations and fluid dynamics simulations.

JS: The emergence of agentic systems in healthcare — AI systems that can reason, plan, and interact with humans to accomplish complex goals. A compelling example is in clinical trial workflow optimization. An agentic AI could help coordinate protocol development, automatically identify eligible patients, monitor recruitment progress, and even suggest adaptive changes to trial design based on interim data. This isn’t about replacing human judgment — it’s about creating intelligent collaborators that amplify expertise, improve efficiency, and ultimately accelerate the path from research to patient benefit.

JL: One area is  generative and multimodal ML — integrating text, images, video, and more — are transforming human–AI interaction, robotics, and autonomous systems. Equally exciting is applying ML to nontraditional domains like semiconductor fabs, smart grids, and electric vehicles, where complex engineering systems demand new kinds of intelligence.

What vision do you have for your journal in the coming years?

KC: The need for a venue to propagate advances in AI/ML in the sciences is clear. The large AI conferences are under stress, and their review system is designed to be a filter not a mechanism to ensure quality, improve clarity and disseminate progress. The large AI conferences also aren’t very welcoming to user-inspired research, often casting that work as purely applied. Similarly, innovation in AI/ML often takes a back seat in physics journals, which slows the propagation of those ideas to other fields. My vision for MLST is to fill this gap and nurture the community that embraces AI/ML research inspired by the physical sciences.

PG: I hope we can demonstrate that machine learning is more than a nice tool but that it can play a fundamental role in physics and Earth sciences, especially when it comes to better simulating and understanding the world.

JS: I see Machine Learning: Health becoming the premier venue for rigorous ML–health research — a place where technical novelty and genuine clinical impact go hand in hand. We want to publish work that not only advances algorithms but also demonstrates clear value in improving health outcomes and healthcare delivery. Equally important, we aim to champion open and reproducible science. That means encouraging authors to share code, data, and benchmarks whenever possible, and setting high standards for transparency in methods and reporting. By doing so, we can accelerate the pace of discovery, foster trust in AI systems, and ensure that our field’s breakthroughs are accessible to — and verifiable by — the global community.

JL:  Machine Learning: Engineering envisions becoming the global platform where ML meets engineering. By fostering collaboration, ensuring rigour and interpretability, and focusing on real-world impact, we aim to redefine how AI addresses humanity’s most complex engineering challenges.

The post Physicists discuss the future of machine learning and artificial intelligence appeared first on Physics World.