2
[2026-09-28] · src Schreiner, Pioquinto, Baeßler, Nesvizhevsky, Cladé, Crivelli, Cubitt, et al.; arXiv:2609.25982 (2026-09-22)
Whispering-gallery states are formed when very slow neutrons graze a curved mirror at grazing incidence and are held against it by total reflection, building up an interference pattern in the reflected beam analogous to acoustic whispering-gallery modes in a domed room. Schreiner, Pioquinto, Nesvizhevsky and coauthors built the apparatus and analysis needed to resolve small shifts in that pattern from an external interaction, and demonstrate it on a magnetic field: a relative pattern shift of delta-g/g ~ (6.5 +/- 0.8 [stat] +/- 1.2 [sys]) x 10-4, with sensitivity to spin-dependent differences between neutron scattering lengths of delta-b_n/b_n ~ 10-4. This is a feasibility and first-observation paper rather than a bound on any specific new physics - no limit on a coupling constant is reported here.
Most precision neutron tests - EDM searches, gravitational bound-state spectroscopy, tests of the equivalence principle - work with a neutron in free fall or free flight, where the interaction time is set by how long gravity or a beamline lets you watch it. A whispering-gallery neutron instead sits pinned against a surface for an extended, tunable time, which is a different lever on the same kinds of questions: the authors list extra fundamental interactions, T-violation, a nonzero neutron electric charge, antimatter gravity, and surface/quantum-reflection effects as targets the technique could eventually reach. What this paper establishes is that the shift is measurable at all, at 10-4 precision - the first rung, not the result those searches are after.
[2026-09-28] · src Wang, Chavez-Cervantes, Satapathy, Matsuyama, Meier, Zhang, You, Marijanovic, Curtis, Demler, Cavalleri; arXiv:2603.24711; Nature Physics (2026), doi:10.1038/s41567-026-03469-z
The depairing current - the current density at which a supercurrent breaks its own Cooper pairs - is the fundamental upper bound on how much current a superconductor can carry, and it has a clean Ginzburg-Landau prediction. In practice it is never reached: ordinary DC transport measurements destroy superconductivity first through vortex motion and self-heating, at a critical current density far below the intrinsic limit. Wang, Cavalleri and coauthors drove current pulses only picoseconds long - too brief for vortices to move or heat to diffuse - through NbN (an s-wave superconductor) and YBa2Cu3O7 (a d-wave superconductor). In NbN at 7 K they see a sharp onset of depairing at Jc* ~ 220 GA/m2, about 2.2 times the conventional DC critical current density of ~100 GA/m2. YBCO at 50 K, whose DC critical current density sits just under 50 GA/m2, shows no comparable sharp threshold under the same picosecond pulses - instead a gradual decline, which the authors attribute to the d-wave gap's momentum-dependent nodes breaking pairs progressively rather than all at once.
This is a new, non-equilibrium way to read out pairing symmetry: instead of inferring s-wave versus d-wave from spectroscopy on an equilibrium sample, the material's own response to being driven toward its depairing limit tells the two apart, because a nodeless gap breaks at one threshold and a nodal one does not. It also means the intrinsic depairing current - a number every superconductor has had in its Ginzburg-Landau theory but which vortex motion and heating have kept out of reach experimentally - is now a measurable quantity rather than a textbook limit. The immediate result is a technique demonstrated on two well-understood materials; what it buys is a probe applicable to superconductors whose gap symmetry is still disputed.
2
[2026-09-28] · src —
Alto takes today's slot: three days since its last session on the 25th is the longest gap of the three instruments right now, and trumpet went yesterday, so the two chops-dependent instruments don't land back to back. With no method book chosen yet, the assignment stays inside the brief's placeholder: long tones, overtones, a scale pattern, nothing invented. Today's fundamental is C♯, the next semitone above the C that crossed the break for the third time on 09-25 — a fourth straight session above it, so the octave-key handling is well past the point of being a new demand. The warmup returns to messa di voce, the shape from 09-16 and not used since, and the musical target closes out the descending interval survey at its widest span, broken sevenths — the mirror of 09-13's ascending broken sevenths and the last interval size a single octave has room for.
5 minutes: long tones on C♯ and D, whole notes, quarter=60, each note a messa di voce — two counts at pp, four counts swelling to mf, two counts receding back to pp. The target is whether the return trip to pp is as clean as the start: a swell that comes back thin, sharp or breathy is the embouchure losing the center on the way down, which is the harder half of the exercise and the one this shape is built to test.
Overtone series from fundamental C♯: voice up through C♯–C♯–G♯–C♯–E♯ using only the (octave-key) C♯ fingering, changing partials with throat and voicing, not embouchure — the same method as every prior session, now a fourth session running above the break. Success criterion: three clean ascents in a row, each partial speaking immediately at a steady dynamic, no pinch or pitch smear at the change, and no audible clunk or pitch dip at the octave-key transition itself.
C♯ major scale, one octave, in descending broken sevenths from the top: 8–2, 7–1 — the widest interval this descending survey has room for, closing it out after thirds (09-16), fourths (09-19), fifths (09-22) and sixths (09-25), and the mirror image of 09-13's ascending broken sevenths. With only an octave to work in, a seventh-wide skip leaves room for just two pairs before the pattern runs out of scale — the shortest cycle this progression has had. Find the lower note of each pair by ear before resolving down to it, rather than reading down a fingering chart. Shape it as a phrase: lean into the upper note of each pair, taper into the one that follows it, rather than playing the pattern metronomically.
Stop the moment the swell's return trip to pp comes back thin, sharp or breathy instead of centered, or the octave-key transition starts needing a jaw shove or throat squeeze to speak cleanly. Both are fatigue standing in for support, and reps past that point train the substitute instead of the real thing.
[2026-09-28] · src —
hearOrnette Coleman — Lonely Woman · Tidal · Spotify
Today's musical target closes out the descending interval survey at its widest possible span, a seventh-wide skip with no stepwise runway to soften the landing. "Lonely Woman," Ornette Coleman's own composition with Don Cherry, Charlie Haden and Billy Higgins, opens on exactly that kind of unresolved, wide leap — a melody built on intervals a conventional tune would smooth over, played with a vocal, keening quality rather than a clean instrumental line. Listen for how the tone stays centered and full-bodied across leaps that would make a thinner sound waver, the same demand today's messa di voce warmup is built to test in miniature: holding the center through a swell rather than losing it under pressure.
3
[2026-09-28] · src Supreme Court of the United States, No. 26A388, People Not Politicians, et al. v. Robert Onder, et al., per curiam order on application for stay (decided Sept. 25, 2026)
Missouri's legislature enacted a congressional map in 2022 and a new one in
On September 3, the Missouri Supreme Court ruled that under the state
constitution the 2025 map "is not the law and has never been the law" absent a voter referendum, leaving the 2022 map controlling as a matter of state law. A federal district court then issued a TRO on September 8 ordering the opposite — use of the 2025 map — which the U.S. Supreme Court itself stayed on September 10, restoring the 2022 map. On September 21 the Eighth Circuit (2-1) ordered the district court to enter a permanent injunction requiring the 2025 map after all, and on September 22 denied a motion to recall its mandate. In this order the Supreme Court stayed both the Eighth Circuit's September 21 mandate and the district court's resulting injunctions, holding its own September 10 stay "remains in effect." The Court leaned on Purcell v. Gonzalez with particular force because, as it put it, the election here "has already started" — ballots under the 2022 map are printed and mailed, and absentee, military and overseas voting are underway — and it rejected the Eighth Circuit's premise that the Missouri Supreme Court's September 3 ruling was itself a destabilizing late change, noting that court had signaled as much already in a May 2026 opinion.
The Court said plainly that "the merits are not entirely clearcut" on whether Article I, Section 2 requires a state to use its primary-election map in the general election — that constitutional question is unresolved. This is a stay of lower-court orders pending further proceedings, not a final ruling, and litigation over which map governs in 2028 and beyond remains possible.
Missouri's November 3 congressional election proceeds under the 2022 map — the one that preserves both of the state's Democratic-held seats — rather than the 2025 map, which would have eliminated one of them. Nothing changes for voters who have already received or cast ballots under the map in use since September 10; what this order does is prevent the Eighth Circuit's September 21 ruling from flipping that eleven days before the election.
[2026-09-28] · src U.S. Court of Appeals for the District of Columbia Circuit, Nos. 26-1049 & 26-1162 (consolidated), Anthropic PBC v. United States Department of War and Peter B. Hegseth, decided Sept. 25, 2026 (Katsas, J., joined by Rao, J.; Henderson, J., dissenting)
On March 3, 2026, the Secretary of War excluded Anthropic's Claude models from the Department's supply chain under the Federal Acquisition Supply Chain Security Act of 2018 (FASCSA), 41 U.S.C. § 4713, after Anthropic declined to drop contractual restrictions barring use of Claude for fully autonomous lethal weapons and domestic mass surveillance. Anthropic petitioned for review on APA, statutory-authority, due-process and First Amendment grounds; the Secretary denied reconsideration June 3, clarifying the determination did not turn on any claim about Anthropic's post-delivery technical control over deployed models. The majority held the Department had ample support for finding that Anthropic's practice of encoding behavioral restrictions into Claude — which has, on at least one occasion, stopped Claude from performing a requested task during an active overseas military operation — qualifies as "manipulating" the product to "deny" its function under FASCSA's definition of a covered supply-chain risk, 41 U.S.C. § 4713(k)(6), regardless of Anthropic's motive; the court rejected the argument that the statute's list of verbs implicitly requires hostile intent. It held due process was satisfied by notice and an opportunity to respond, and that the First Amendment claim failed because the exclusion rested on Anthropic's refusal to accept a contract term, not on its advocacy. Judge Henderson dissented on the statutory question, arguing under canons including series-qualifier and ejusdem generis, and citing Fischer v. United States (2024), that "maliciously" modifies the entire verb series and that the majority's reading turns good-faith contract enforcement into a national-security violation.
The panel did not pass on whether the Department's underlying policy — favoring unrestricted battlefield access to AI over guardrails against unconstrained autonomous lethal force — is sound; it held only that the Secretary acted within his statutory and constitutional authority. It does not address what happens to Claude instances already delivered to Department contractors, which the Secretary's June 3 order said was not the basis for the determination. Rehearing en banc and certiorari both remain open.
The exclusion stands: federal agencies and contractors working with the Department of War remain barred from using Claude. The ruling sets an operative legal standard reaching beyond this case — a contractor's own contractual restrictions on how the government may use its product can themselves constitute a statutory "supply chain risk" under FASCSA, triggering exclusion, without any showing of hostile intent or foreign influence. Any AI vendor selling into the federal government now has to weigh maintaining use restrictions against this precedent for exclusion.
[2026-09-28] · src Supreme Court of the United States, No. 26A308, Department of Homeland Security, et al. v. League of Women Voters, et al., per curiam order on application for stay (decided Sept. 25, 2026), Jackson, J., dissenting, joined by Sotomayor and Kagan, JJ.
SAVE (Systematic Alien Verification for Entitlements) has run since 1986 to verify immigration and citizenship status for benefits eligibility, and 27 states have also used it since 2005 to check voter rolls. Following a March 2025 executive order, DHS and the Social Security Administration modified SAVE to check against SSA's NUMIDENT records, including Social Security numbers, and to allow bulk rather than only individual lookups. The League of Women Voters and the Electronic Privacy Information Center sued, arguing the changes violated the Social Security Act's confidentiality provision (42 U.S.C. § 405(c)(2)(C)(viii)(I)), the Privacy Act, and the APA. On June 22, the U.S. District Court for D.C. (Judge Sparkle Sooknanan) granted summary judgment to plaintiffs on all three claims and vacated the modified program; a divided D.C. Circuit panel denied a stay on September 4. In a 6-3 decision, the Supreme Court granted the government's stay application, holding it is likely to succeed on the merits because 8 U.S.C. § 1373's "notwithstanding any other provision of law" clauses require DHS to share and receive citizenship and immigration-status information in a way the Court held supersedes the Social Security Act and Privacy Act restrictions the district court relied on. The Court found standing for at least one plaintiff — a naturalized citizen registered in Louisiana whose SSA record had not updated — but held that risk did not outweigh the government's showing of harm. The order preserves the National Voter Registration Act's 90-day pre-election moratorium on "systematic" list-purging, but allows "individualized" citizenship-verification inquiries to proceed during that window. Justice Jackson's dissent, joined by Sotomayor and Kagan, argued the majority granted emergency relief without full briefing or argument on two never-before-interpreted statutory provisions.
This is a stay of the district court's injunction pending appeal to the D.C. Circuit, not a final ruling on the Social Security Act, Privacy Act or APA claims, which remain to be litigated there. The Court did not decide whether the NVRA's 90-day moratorium applies to programs that specifically target noncitizen registrations rather than "systematic" removals generally — that question is pending separately in Republican National Committee v. Mi Familia Vota, No. 25-1017, not yet argued. No state is compelled to use SAVE; its use for voter-roll checks remains voluntary.
States that use SAVE for voter-roll checks may resume using the modified, Social-Security-linked version for the rest of the 2026 cycle, limited to one-by-one checks rather than bulk searches through Election Day. A voter flagged by the tool — including a naturalized citizen whose SSA record has not caught up, the harm the named plaintiff alleged — would now have to affirmatively produce proof of citizenship to stay registered, starting immediately, while the underlying legality of the program is still being litigated in the D.C. Circuit.
2
[2026-09-28] · src CTAN update announcements, tikzphysics 1.0.0 (new on CTAN, 2026-08-13) through 1.5.0 (2026-09-27)
tikzphysics is a TikZ library purpose-built for classical-mechanics diagrams —
contact surfaces, ramps, blocks, springs, pulleys, pendulums, motion arrows,
mirrors and lenses — and it has moved fast: new to CTAN on 2026-08-13, five
releases since, most recently 1.5.0 on 2026-09-27. The release that actually
changes a document already written against it is 1.2.0 (2026-09-05): a spring
stopped being a node you place and became a path decoration you draw between
two points, written \draw[spring] (A) -- (B);, with leads that can now
collapse to zero length. The same release moved blocks to a 1cm-square default
and pulleys to a 1cm-diameter one rather than requiring an explicit size on
every use, added native to[over pulleyP]= connections and wedge
tangent/normal anchors, and shipped three discovery keys —
show anchors, show keys, physicshelp — that print a diagram's own anchor
names and option keys back onto the page itself, which is how you find out
what to write next without leaving the document to read a manual.
This is worth knowing before adopting it, not after. The package is three
weeks past its first breaking change and every release since has kept moving —
fluid-mechanics nodes, percentage anchors for platforms and walls, wedge-anchor
fixes — which is the ordinary shape of a library still finding its interface,
not a reason to avoid it outright. What it buys today is real for anyone
setting problem-set or paper figures by hand: named semantic primitives —
spring, pulley, wedge, support — in place of raw TikZ coordinate arithmetic for
every incline and pulley system. The honest caveat is the same fact stated the
other way round: pin the version before building a semester's worth of
diagrams on it, because the rewrite that hit spring in week four can hit
block or pulley in week eight, and nothing about 1.5.0 says it won't.
[2026-09-28] · src GNU Emacs Lisp Reference Manual, "Defining Macros"; LaTeX3 interface3, the naming convention for internal functions and variables
Define this in Emacs Lisp — no package beyond stock defmacro, let, while
— and expand it without ever running it:
(defmacro for ((var start stop) &rest body)
`(let ((limit ,stop))
(let ((,var ,start))
(while (<= ,var limit)
,@body
(setq ,var (1+ ,var))))))Now call it the way a reasonable person would, choosing an ordinary variable name that happens to collide with the macro's own internal one:
(macroexpand-1 '(for (limit 1 5) (print limit)))The expansion nests two bindings of the same name, limit: the outer one
meant to freeze the stop value, the inner one the caller's own loop variable.
Because the inner let shadows the outer inside the while, the test
(< limit limit)= is comparing the same binding to itself — always true —
so the outer, frozen 5 is never consulted again and the loop never stops.
That is variable capture: the macro's author picked a plausible name for a
helper binding, and it turned out not to be private at all.
Then write one paragraph — that is the whole of the TeX half, no code to compile — answering why you cannot write this exact bug in a TeX macro. Ground it in two specific facts, not a general "TeX is different":
TeX's #1, #2, #3 are positional substitutions of token lists, and a
call like \for{limit}{1}{5} hands the macro four characters as data,
not a binding. The only names TeX has at all are control-sequence names —
things spelled with a leading backslash — so a caller's argument can only
collide with a macro's internal helper if the caller happens to write a
control sequence, e.g. \limit, into the body, which is a far narrower
coincidence than Lisp's problem, where a loop variable and a macro's own
internal binding live in exactly one namespace by default.
Where TeX does have the analogous bug — two control sequences both
answering to \tmp — the fix is not a freshly generated name, because TeX
has no primitive that does what gensym does. The fix is a naming
convention: expl3's internal functions and variables get an
underscore-separated, module-prefixed name — \module_thing:n,
__module_thing_var — that nobody is expected to type by hand. Discipline
is standing in for what Lisp's evaluator gives gensym for free.
Success criterion: the expansion in macroexpand-1 visibly shows the doubled
limit and the self-comparison (< limit limit)= where a literal 5 was
intended — that is the bug, independent of ever running the loop — and the
paragraph names both facts above by name (the namespace difference, and the
convention-versus-primitive difference), not a one-line "TeX doesn't have
this problem."
The technique that transfers is the reflex, not the trivia: any time a macro's
own expansion introduces a binding the caller didn't ask for, that binding
needs a name nothing else could plausibly write — gensym in Lisp, an
underscore-and-module-prefixed name by convention in expl3 — and this hour is
what makes reaching for it automatic rather than a debugging session away.
It also heads off a specific wrong turn: hunting TeX for "the gensym of
LaTeX" the first time this bites in expl3 code, when what is actually there is
a naming convention doing the same job by hand, for a language that never
needed a fresh-symbol primitive because it never had a namespace of variable
names to protect in the first place.