The Weyl operator's symbol class #
Category: 1-Mathlib (CSD-free; staged as a Mathlib-upstream candidate). BACKLOG #92(a), brick (a1).
#92's measurement of its own part (a) was that Op(a) as a continuous operator on ๐ข(โ, โ) is not
a gap in the proofs but a statement needing a different symbol class: with weylOp's datum โ a
family b : โ โ ๐ข(โ, โ) with no regularity in its first slot โ weylOp b ฯ need not even be
continuous. WignerWeyl.lean works around this by assuming what it needs: integrable_weylPair
takes joint continuity (hbc) and a uniform bound integrable in the first slot (hM, hbM) as
hypotheses.
This file fixes the class and discharges those hypotheses as theorems.
- โ
exists_bound_fst, โexists_bound_sndโ a Schwartz function on the plane decays like(1 + tยฒ)โปยนin each slot separately, uniformly in the other. Two orders ofSchwartzMap.decay(k = 2andk = 0) andโpโยฒ โฅ pแตขยฒ; weylOpKโ the Weyl operator of a jointly Schwartz kernel, and โweylOpK_eq_weylOp: it agrees withweylOpof any slice family, so everything already proved aboutweylOptransfers whenever a slice family is available;- โ
continuous_weylKerneland โ โexists_integrable_boundโWignerWeyl.lean's three assumed hypotheses, now theorems on this class; - โ
integrable_weylOpK_integrandand โ โcontinuous_weylOpKโ the defect #92 recorded, fixed: on the Schwartz kernel class the Weyl operator's output is continuous, by dominated continuity against4Cยท(1 + (y โ xโ)ยฒ)โปยน, the majorant the second-slot decay supplies uniformly on a unit ball of parameters.
Honest scope #
โ ๏ธ This is the symbol class, not the calculus. Op(a) : ๐ข(โ, โ) โL ๐ข(โ, โ) is not proved โ
continuity of the output function is the first of the Schwartz seminorm estimates, not the last. The
remaining obstruction is the one #92 measured and it is unchanged: smoothness of the output to all
orders needs differentiation under the integral sign to all orders with bounds, which the pin has
in no form for integrals over โ (the corpus's ContDiffParametricIntervalIntegral.lean, from #60,
is interval integrals on a compact interval, where the bounds come free from continuity).
โ ๏ธ No slicing here. weylOpK_eq_weylOp takes a slice family as a hypothesis. That a Schwartz
function on the plane slices into a Schwartz-valued family is the two-variable Schwartz API #92 named
as absent when this file was written; it is now SchwartzSlice.lean (#121(i),
SchwartzMap.slice), and the symbol of a sliced kernel is a Schwartz function on the plane by
PartialFourierCLM.lean (#121(ii), partialFourierCLM_apply_eq_weylSymbol).
The properties above are still proved for weylOpK directly, as written.
References: WignerWeyl.lean (weylOp, integrable_weylPair,
integral_conj_mul_weylOp), WignerCalculus.lean (#92(b)(c)),
Mathlib/Analysis/Calculus/ContDiffParametricIntervalIntegral.lean (#60, the interval case);
specs/BACKLOG.md #92, #63, #60.
A Schwartz function on the plane decays in each slot separately #
The Weyl operator of a jointly Schwartz kernel #
The Weyl operator of a jointly Schwartz kernel. Same integral as weylOp, with the kernel
given as one Schwartz function on the plane instead of a family with no regularity in its first
slot.
Instances For
โ
The bridge. Whenever a slice family is available, this is weylOp of it โ so every
statement already proved about weylOp transfers. The hypothesis is what the absent two-variable
Schwartz API would supply.
โ
WignerWeyl.lean's joint-continuity hypothesis, now a theorem.
โ
โ
WignerWeyl.lean's two bound hypotheses, now a theorem: on the Schwartz kernel class there
is a bound integrable in the first slot and uniform in the second.
The majorant the second-slot decay supplies, uniformly for parameters in a unit ball.
โ The defining integral converges, with no hypotheses.
โ
โ
The defect #92 recorded, fixed: on the Schwartz kernel class the Weyl operator's output is
continuous. Dominated continuity against the second-slot majorant, which is uniform for parameters
within 1 of the point.