3
[2026-09-12] · src PandaX Collaboration; arXiv:2609.10048 (2026-09-09)
Neutrinoless quadruple beta decay (0nu4beta) is a hypothetical process in which a nucleus converts four neutrons to protons while emitting no neutrinos at all - a more exotic cousin of neutrinoless double beta decay (0nu2beta), and one whose observation without any accompanying 0nu2beta signal has been argued to point specifically to neutrinos being Dirac rather than Majorana particles. The PandaX-4T collaboration searched 148.4 kg.yr of xenon-136 exposure, collected during commissioning and the experiment's first science runs, for 0nu4beta decay and saw no excess over background, setting a lower limit on the half-life of 6.01e24 yr at 90% confidence - the most stringent bound on this process yet reported in any xenon target, and roughly a thousand times tighter than NEMO-3's earlier (1.1-3.2)e21 yr limit on the analogous decay of neodymium-150, though the two isotopes' different phase-space factors mean the numbers are not directly interchangeable.
This is a null result on a process nobody has ever seen in any isotope, so it does not touch the Dirac-versus-Majorana question directly - only an observed 0nu4beta without 0nu2beta would do that. What it does establish is that a multi-tonne xenon TPC built for dark-matter detection is now the most sensitive instrument available for this particular exotic-decay channel, well past what dedicated double-beta experiments in other isotopes have managed, and with far more exposure still to come as PandaX-4T continues to run. The number to watch is not this limit itself but how fast it moves as exposure grows, since a rapid tightening is what would eventually make a Dirac-favoring observation - or a decisive exclusion - possible.
[2026-09-12] · src Fu, Shen, Hu, et al.; arXiv:2609.11922 (2026-09-10)
Every Gaussian boson sampling (GBS) demonstration to beat classical simulation so far - including the Jiuzhang line's 3,050-photon result last year - has run on room-sized free-space or fiber optical tables, because reaching the needed low loss, high phase precision and high-speed modulation together has demanded bulk components. Fu et al. instead built the whole interferometric network - high-speed electro-optic modulators, on-chip delay lines, and a time-space multiplexed mesh - monolithically on a single thin-film lithium niobate chip, running it at a 4 GHz clock rate to record detection events of up to 11,059 photons within one millisecond, more than three times the photon count of the previous largest (tabletop) GBS experiment. Few-photon calibration fidelities (0.9978 single-photon, 0.9962 two-photon bunching, 0.9911 two-photon non-bunching) and a Bayesian analysis favoring genuine GBS statistics over classical spoofing models validate the sampling; the authors estimate that classically reproducing a 2.5-microsecond sampling run would take a tensor-network simulation on an NVIDIA A100 GPU at least 9.7e6 years.
The photon count is not really the news - Jiuzhang 4.0 was already three orders of magnitude past classical reach at a third the photon number. What is new is doing it on a chip: the barrier to useful photonic quantum hardware has never been crossing the classical-simulation line once on an optical table, it has been building a device that can be manufactured, aligned and run outside a national lab. The authors' own framing is the honest one - a photon-count record is "a necessary milestone, not the final destination" - and they spend more of the paper on reconfiguring the same chip as a physical "world model" that outpredicts a classical echo-state-network baseline with fewer trainable parameters than on the sampling record itself, which is the direction (useful computation, not just a hard-to-simulate distribution) this line of work still has to prove out.
[2026-09-12] · src Wang, Huang, Jing, Yao, Wang, Liu; arXiv:2609.11730 (2026-09-10)
Neutral-atom architectures want to combine two species in one array: a divalent atom like ytterbium-171, whose nuclear spin gives a long-lived data qubit, alongside an alkali like rubidium-87, whose bright, fast, species-selective readout makes it a good control and ancilla qubit that can be measured and reset without disturbing the data qubits sitting nearby. The obstacle has been that entangling the two species by the usual route - Rydberg dipole-dipole interaction tuned with a DC Stark field - induces severe Stark mixing in the divalent atom's Rydberg levels. Wang et al. instead find an optically accessible S+S<->P+P Forster resonance between the two species that sits at zero electric field, with a sub-megahertz Forster defect, so the interspecies exchange interaction is strong at ordinary array spacing without any Stark tuning at all. Driving this exchange with a shaped optical pulse, they demonstrate an interspecies controlled-Z gate in 0.36 microseconds with 99.91% intrinsic fidelity, staying above 99.85% under realistic position and laser-amplitude perturbations, and they separately identify an auxiliary van der Waals channel for the same atom pair.
This closes a specific, previously open gap rather than pushing an existing number further: a fast, high-fidelity gate directly between two different atomic species, with no destructive field-mixing tax, is what a rubidium-ytterbium processor needs in order to use rubidium as an ancilla that can be read out and reset mid-circuit while the ytterbium data qubits' long coherence survives untouched. That capability - measurement without disturbing the rest of the register - is exactly the piece fault-tolerant neutral-atom quantum computing needs and single-species arrays cannot cleanly provide. It is a component demonstration, not a processor; whether it holds up when scaled into an array with many simultaneous interspecies gates is the next thing to watch.
2
[2026-09-12] · src Arban, Complete Conservatory Method — 14 Characteristic Studies
Trumpet takes today's slot: three days since its last session is the longest gap of the three instruments right now, and piano — which went yesterday — stays off regardless, since it's the one that rewards a longer, less frequent block rather than today's short turnaround. Six sessions into the 14 Characteristic Studies, each one has isolated a demand the others didn't touch — endurance, dynamic contrast, legato across a register break, slur-to-tongue switching, a soft level held without swelling, pulse held steady across changing note density. No. 7, marked Allegro — the fastest of the seven so far — asks something none of those tested: it is staccato throughout rather than a mix of articulations, and the standard trap with it is treating "detached" as "disconnected," so that a technically accurate performance still sounds like a pile of separate notes instead of one fast phrase.
5 minutes: the C major scale, one octave, quarter = 88. First play it fully slurred, listening for its overall shape — where it rises, where it settles. Then play the identical scale lightly tongued, same tempo, same dynamic, with the image that only the tongue interrupts the air between notes; the air itself never stops moving. Alternate slurred and tongued three times, keeping the shape from the slurred version audible in the tongued one — the same scale, only now with commas where there were none.
Characteristic Study No. 7 straight through at the marked Allegro tempo. Because the printed articulation is staccato through most of the study, success is not "each note short and separate" but "each note short and the phrase still whole": three consecutive clean passes where a listener could hum the line back afterward, not just describe how detached it was. Where a passage keeps breaking into disconnected notes, go back to the warmup's method — play that passage fully slurred once to relearn its shape, then add the staccato back without losing it.
Art of Phrasing, Melody No. 20. Play it once for notes, then once carrying over today's discipline: wherever the melody calls for a detached or lightly-tongued note, keep the underlying phrase audible through it exactly as in the study, rather than letting articulation reset the ear each time.
Stop after the three clean technical passes, or sooner if the tongue's separation starts turning into an accent on every note rather than a light interruption of the air — that is staccato curdling into something percussive, and it is the sound of the chops giving out before the ear notices anything else has.
[2026-09-12] · src —
hearWynton Marsalis — Variations Sur "Le Carnaval de Venise" · Tidal · Spotify
Arban wrote this theme specifically to be a showcase for staccato technique, and it is the same demand today's technical target makes in miniature, run at full virtuoso length: chorus after chorus of fast detached notes that still have to read as one melody being decorated rather than as a display of separation for its own sake. Marsalis, with the Eastman Wind Ensemble under Donald Hunsberger, keeps every variation singable even at its most rapid-fire — listen for how the line never stops sounding like the theme, no matter how many notes get inserted between its landmarks.