For 50 years, the standard model of particle physics has been very successful in describing subatomic phenomena. In the past quarter of a century, this was challenged by a mismatch between its predictions and precision measurements of the anomalous magnetic moment of the muon, aμ. This disagreement was eventually reconciled, first through a determination in an ab initio lattice calculation1 of the most uncertain theoretical contribution, the leading-order hadronic vacuum polarization (LO-HVP), a μ LO-HVP , and subsequently by experimental results2 and updates of the reference standard-model predictions using lattice results for a μ LO-HVP (ref. 3). Here we present a new calculation for this crucial quantity, obtaining a μ LO-HVP = 715.1 ( 2.5 ) ( 2.3 ) [ 3.4 ] × 1 0 - 10 . This reduces the uncertainty by a factor of 1.6 compared with our earlier computation1. We use a hybrid approach that includes a small, long-distance contribution from experiments in a low-energy regime in which they all agree. Our approach combines the strengths of experimental and lattice data in different energy ranges, achieving better precision than with either alone. Our lattice quantum chromodynamics (QCD) simulations are performed on finer lattices than in ref. 1, allowing for an even more accurate continuum extrapolation. Combined with the calculations of the other standard-model contributions summarized in ref. 3, our result leads to a prediction that differs from the recent measurement of aμ (ref. 4) by only 0.5 standard deviations. This provides a notable validation of the standard model to 11 digits.
The mean pairwise velocity of massive halos reflects the gravitational force law on cosmic scales. We combine cosmic microwave background intensity maps from the Atacama Cosmology Telescope and a galaxy catalog from the Sloan Digital Sky Survey to estimate the mean pairwise velocity using the kinematic Sunyaev-Zeldovich (kSZ) effect. On scales from 30 to 230 megaparsecs, we constrain the gravitational acceleration between pairs of halos at separation r to be g∝1/r^{n} with n=2.1±0.3, which is consistent with Newtonian gravity in an expanding spacetime (i.e., the standard ΛCDM model). This constraint shows agreement with an inverse quadratic radial dependence over the large distances separating galaxy halos, as expected in standard cosmology. Upcoming surveys have the potential to rule out n=1 at 10σ significance. Our results establish the kSZ effect as a powerful tool for testing gravity on cosmological scales.
The Future Circular Collider (FCC) integrated programme begins with the FCC-ee, an electron-positron collider, followed by the FCC-hh, a proton–proton collider installed in the same 91 km circumference tunnel near CERN. Spanning 15 years from the mid-to-late 2040s through the early 2060s, the FCC-ee will operate at centre-of-mass energies between approximately 90 and 365 GeV, consistently delivering the highest possible luminosities to four experiments in a sustainable and energy-efficient manner. A key element of its design is top-up injection from a full-energy booster housed in the same 91 km tunnel, along with the world’s most intense positron source and 20 GeV injector linacs. The FCC-ee injector complex, comprising a high intensity positron source, a damping ring, and a linac accelerating electrons and positrons up to 20 GeV, is expected to start operation several years earlier than the booster and the collider. The primary objective of the FCC-ee is its rich High Energy Physics programme based on electron-positron collisions at various centre-of-mass energies (Benedikt et al. in Eur Phys J C 85:1468 https://doi.org/10.1140/epjc/s10052-025-15077-x , 2025). In addition, thanks to its large circumference, high beam energy, abundant positron production, and low-emittance beams, the FCC-ee also offers unique opportunities for various fields of physics and science. These include the potential production of true muonium, the creation of a Bose-Einstein condensate of positronium, Compton imaging with high-energy photons, the generation of spatially coherent photon beams, possibly down to 0.1 Åwavelengths—achieving several orders of magnitude higher average and peak brightness than any existing or planned light source—radioactive isotope production, and an electron- or photon-beam-driven neutron source. We present these and other science exploitations of the FCC-ee accelerator complex.
The Circular Electron-Positron Collider (CEPC), a next-generation high-luminosity collider, employs a crab waist scheme to achieve ultrahigh 5 & times; 1034 cm-2s-1 luminosity at Higgs mode. Owing to the extremely small beam size, the luminosity is highly sensitive to the stability of final focusing elements, where mechanical vibrations (e.g. ground motion) may induce beam offsets and luminosity degradation. To address this, a luminosity-driven dithering system is implemented for horizontal beam stabilization. In this work, we develop an optimized 4H-SiC fast luminosity detector scheme using an array of radiation detectors with picosecond time resolution positioned at critical locations. By using self-development software RAdiation SEmiconductoR (RASER), we optimize the active area of the detector to achieve 2% relative precision at 1 kHz. Furthermore, the RAW SUM exhibits a near-linear correlation with luminosity attenuation, enabling real-time luminosity monitoring.
Detailed numerical simulation remains the workhorse of accelerator design and operation, but it is slow, and most machine-learning surrogates built to replace it only return a handful of beam observables, useful for monitoring but of little use as an input to the next beamline element or the next simulator, since both need a full beam. LinacNet closes that gap: it predicts the complete macro- particle cloud one segment at a time, so a trained module’s output can be handed directly to the next module or to a standard tracking code. The architecture adapts PointNet, a per-particle multilayer perceptron combined with a permutation-invariant pooling step, to eight-dimensional macro-particles conditioned on each segment’s control settings. Trained and tested on 30 000 ASTRA simulations of the ThomX linac, a single segment module reaches R 2 above 0.99 on both the beam observables and the particle distribution, with residuals smaller than what the installed diagnostics can resolve; chaining all 25 segments preserves a faithful beam description end to end, although balancing per- segment and end-to-end accuracy remains the main open problem. We close by asking what this composable design implies beyond a 50 MeV linac: as accelerator facilities turn to shared, cloud-based machine-learning infrastructure, a library of interoperable, segment-level surrogates of this kind is one plausible route toward simulating machines an order of magnitude larger and more complex than ThomX, such as the Future Circular Collider.
