4
[2026-08-15] · src Double Chooz Collaboration; Phys. Rev. Lett. (2026), doi:10.1103/dr26-j19g, arXiv:2510.04869
A running reactor's fission products emit a well-modeled antineutrino flux; less well tested is what happens after shutdown, when long-lived fission isotopes left in the partially burnt core fuel and in nearby spent-fuel cooling pools keep decaying and emitting a much fainter residual flux. Double Chooz's near detector, 400 m from the Chooz B reactor cores in France, took 17.2 days of data while both units were completely shut down and recorded 106 +/- 18 antineutrino candidate events in the energy window where this residual signal is strongest - a 5.9-sigma excess over background, matching simulations of the remaining fuel inventory.
This turns a quantity that was previously only modeled into a calibrated measurement, and it means antineutrino detectors - already used to verify that a reactor is operating and at what power - could extend the same remote, non-intrusive monitoring to spent fuel and shut-down cores, relevant to safeguards during decommissioning or extended outages when direct access isn't available. It's one site and one detector design; whether the roughly 18% statistical uncertainty tightens enough to distinguish real fuel-handling scenarios (fuel moved out versus left in place) is the question that decides whether this becomes a usable monitoring tool rather than a proof of principle.
[2026-08-15] · src Leindecker, Gallagher, Brucke, Zehnder, Milanovic, Marinelli, Gerritsma, Spreeuw, Home, Hempel; Phys. Rev. Lett. 137, 063202 (2026), doi:10.1103/kj5p-qqs5, arXiv:2601.22981
The optical Magnus effect is the light analogue of the mechanical one: the intrinsic spin-orbit coupling of a tightly focused, circularly polarized beam displaces its effective interaction profile transverse to propagation, by an amount and direction set by the light's handedness. It has been predicted for tightly focused beams for years but not directly observed in an atom-light interaction. Probing a single trapped 40Ca+ ion on a quadrupole transition and spatially mapping the interaction profile for different beam configurations, the team measured transverse shifts of several hundred nanometers between opposite-handedness beams, tracing the effect to non-paraxial electric-field components that standard tight-focus models for ion addressing omit.
Tightly focused addressing beams are exactly what individual-ion and individual-atom qubit control relies on, and an unmodeled few-hundred-nm transverse shift is large enough to contribute to gate error and crosstalk in dense qubit arrays. This is a characterization, not a fix - the open question is how much of today's laser-driven single-qubit gate error this one effect actually accounts for once it's folded into control-pulse models.
[2026-08-15] · src XENON Collaboration; Phys. Rev. Lett. (2026), doi:10.1103/2lrq-f6bk
XENONnT analyzed 7.83 tonne-years of ionization-only ("S2-only") data - 579 days across three science runs, its largest low-threshold exposure to date - built on a newly complete background model for this channel. A blind analysis found no significant excess over background, and the collaboration set new 90%-CL upper limits on spin-independent dark-matter-nucleon and spin-dependent dark-matter-neutron scattering, dark-matter-electron scattering, and axion-like particles and dark photons, reaching dark-matter masses as low as roughly 3 GeV - the tightest limits yet in that range from any liquid-xenon experiment, and close to the "neutrino fog," the point where coherent solar-neutrino scattering becomes an irreducible background indistinguishable from a genuine light-WIMP signal.
A null result at record sensitivity still rules out real parameter space, and closing in on the neutrino fog in this channel marks roughly where conventional liquid-xenon detectors run out of room to improve without directional or spectral tricks to separate solar neutrinos from a real signal. Nothing here is a detection; the number worth watching is how much further this specific channel can go before it hits that floor.
[2026-08-15] · src Liu et al.; Phys. Rev. Lett. 137, 066502 (2026), doi:10.1103/pxpf-6bsv
1T-TaS2 hosts three charge-density-wave (CDW) phases - incommensurate, nearly-commensurate, commensurate - that in bulk crystals give way to a correlated, Mott-like insulating state. Thinning usually weakens a CDW, as screening and structural relaxation degrade the order. Using temperature-dependent Raman spectroscopy and transport down to the true monolayer limit, this team found the opposite: transition temperatures rise as thickness drops, sheet resistance climbs by orders of magnitude, carrier localization length falls sharply, and the bulk's first-order, hysteretic commensurate-to-nearly-commensurate transition disappears entirely in the monolayer. Calculations attribute the enhancement to strengthened long-range Coulomb interactions once out-of-plane dielectric screening is lost.
This runs against the usual thinning-degrades-order pattern seen in most 2D CDW materials, and suggests monolayer 1T-TaS2's correlated insulating state is a genuinely different regime rather than a weaker echo of the bulk one - reduced dimensionality here strengthens, rather than suppresses, the interactions that drive the order. It's one material system; whether the same reduced-screening enhancement shows up in other layered CDW hosts, or is specific to 1T-TaS2's particular Coulomb-to-bandwidth ratio, is untested.
1
[2026-08-15] · src build/orrery-test.el
build/orrery.el splits the build into two functions on purpose.
orrery-build-ir (orrery.el:72-83) takes model-dir, feed-dir,
inbox-dir, provenance and generated-at as plain arguments and returns
an IR plist — it never reads a file it wasn't handed, never calls git,
never checks the clock. orrery-build (orrery.el:85-110) is the one
impure wrapper around it: it computes provenance from git and passes
generated-at through to write dist/ir.json, digest.org and
index.html. Only orrery-build-batch (orrery.el:112-121), the actual
make build entry point, calls orrery-now (orrery.el:32-37) to get a
real timestamp. Everything below that layer is a pure function of its
inputs.
build/orrery-test.el is where that split pays for itself.
orrery-test-golden-stamp and orrery-test-golden-provenance
(orrery-test.el:30-39) are two fixed constants — a timestamp frozen at
2026-01-01T00:00:00Z and a provenance plist that says :version "golden"
instead of a real git describe. orrery-test--golden-actual
(orrery-test.el:69-77) calls orrery-build-ir directly — not
orrery-build — with the test/golden/ fixtures and those two constants
in place of the real clock and the real git lookup. orrery-verify-golden
(orrery-test.el:112-121) then diffs that output byte-for-byte against
test/golden/ir.json and archive.json.
The one Lisp idea: push effects to the edges. A function that calls
current-time or shells out to git produces a different answer every time
you call it, so you cannot write down "the answer" to compare against —
there is no fixed target for equal to check against tomorrow. Once the
clock and git are arguments instead of calls, orrery-build-ir becomes
referentially transparent: same five inputs, same IR, forever. That is what
makes a golden test possible at all — not a testing trick bolted on
afterward, but a consequence of where the impurity was allowed to live in
the first place. If orrery-build-ir called orrery-now internally, every
make golden run would need to diff against a different expected file
each time it ran, which isn't a fixture, it's just re-implementing the
function next to itself.
This is the same shape as last lesson's nil discipline in
orrery-ir--strings — a boundary designed so a whole category of "worked
on my machine, broke in CI" bug can't occur, rather than a rule you have to
remember to follow.
Exercise (verifiable): open build/orrery-test.el and change
orrery-test-golden-stamp from \"2026-01-01T00:00:00Z\" to any other
ISO-8601 string — say tomorrow's date. Run make golden. It fails, and the
failure message names exactly one field's worth of byte difference. Revert
the constant and run make golden again to confirm it's green. In one
sentence: why does editing that constant fail the gate, when editing your
own system clock while running make build would not touch make golden
at all?
2
[2026-08-15] · src twelve-keys/src/exercises/stride/exercise.py
Key: B major — five sharps, and the next stop on the fourths cycle after 2026-08-12's Gb/F#. First comping-family exercise in the rotation; the last two piano sessions were both harmony-shape drills (upper-structure triads, the Coltrane cycle), so today trades chord vocabulary for LH independence — a different muscle, and one that rewards the longer block piano gets today.
5 minutes, hands separately, on the ii–V–I in B (C♯m7 – F♯7 – Bmaj7): first just the bass alone — root → 5th → root → 5th, quarter notes, one bar per chord, looped until the leaps stop needing a look down. Then just the chord stab alone — rest-chord-rest-chord — to get the 2-and-4 rhythm in the hand before asking it to also find the bass.
Hands together through the full C♯m7 – F♯7 – Bmaj7 cycle at quarter=80 (the exercise's own tempo). The leap from the low bass note up to the mid-register shell is the hardest physical move in the pattern, and B major's five sharps put it in less-familiar territory under the hand than the naturals-heavy keys usually get. Success criterion: three clean loops with no hesitation on the leap and no fumbled landing on the shell — the tell that the geography is still being found by eye rather than felt.
Once the straight root-5th bass is solid, swap in the exercise's stepwise variant — root → 3rd → 5th → 6th-down-to-next-root — through the same B cycle. That's the move the exercise notes describe as "pre-bebop stride starting to sound like swing": play the turnaround as a phrase that walks somewhere, not a pattern that repeats.
Stop the moment the low-bass leap starts being aimed — a visible glance down, or a landing that's close but adjusted after the fact — rather than felt. That's the hand losing the geography, and reps past that point train hunting for the note instead of the leap itself.
[2026-08-15] · src —
hearFats Waller — Handful of Keys (Ain't Misbehavin') · Tidal · Spotify
An uptempo solo stride showcase built almost entirely out of exactly today's move — low bass leaping to a mid-register chord stab, over and over, at a tempo that leaves no room to look for either landing. Listen for how level the time stays through the leaps, which is the whole point of today's technical target at a much slower tempo.