NodeContract #
Ported from scottnarmstrong/Superdiffusion at 7f20476385d4af418f110303085ad0740c9eba46. Module imports, support namespace and local option directives adapted for Lean Pool.
The closure node's contract, transcribed, and what it feeds #
The closure of l.approximate.recurrence.formula asks, at
e.what.do.we.have, for the two displayed upper estimates for
sigmaBar_(n+h)/sigmaBar_n and its reciprocal, with C depending only on d.
TwoSidedClosure below is that contract, transcribed with the manuscript's own
binders: the standing hypotheses of the enclosing lemma
S(m0-1, E) valid , E >= C cstar^{-1} , cgamma <= E^{-5} ,
and the quantifier range
n in (-infty, m0] cap Z , h in N , h <= 6 cstar cgamma^{-1} ,
mstarstar <= n <= m0 - h .
The manuscript's standing dimension hypothesis d >= 2 is carried as the
antecedent 2 <= d of the contract. It is a standing assumption of the whole
paper, not a proof step, and it is what the proved principal comparison
exists_const_descendantsAverage_integral_principalEnergy_le_annealedLimit
binds as hd : 2 <= d; a discharger of the contract obtains it by intro hd.
Nothing here proves the contract. TwoSidedClosure is a Prop-valued
definition: the exact target, written down so that it can be verified against
the source before the two remaining endpoints are proved, and so that the
assembly skeletons of Closure.Junctions have a stated destination.
The second half of the module discharges the other side of the interface: the
contract's two displays deliver, verbatim, the hupper and hlower binders of
RecurrenceIntegration.integrate_approx_recurrence at the sequence
s = fun m => (Annealed.sigmaBar M m : R). That is proved unconditionally, and
it is the reason the displays are stated at these carriers.
The scale gate on m0 #
The gate carried on m0 below is mStarStar M < m0, not the printed m0 in (mstar, infty) cap Z.
References #
- ABK26,
l.approximate.recurrence.formula. - ABK26,
e.mstarstar;l.integrate.approx.recurrence.
The contract #
The closure node's exact contract.
In dimension d >= 2 there exists C(d) < infinity such that, if E >= C cstar^{-1}
and cgamma <= E^{-5}, then for every n in (-infty, m0] cap Z and h in N with
h <= 6 cstar cgamma^{-1} and mstarstar <= n <= m0 - h, both lines of
e.what.do.we.have hold.
The standing dimension hypothesis 2 <= d is the leading antecedent; it is the
binder the proved principal comparison carries as hd : 2 <= d. The constant
is quantified outside the model, so "depends only on d" is rendered as
"depends only on d", not "depends on M". The manuscript's h in N is
transcribed with no positivity: the degenerate shell h = 0 is covered by
Closure.recurrenceUpperDisplay_zero and
Closure.recurrenceLowerDisplay_zero, so a proof of this contract may assume
0 < h at no cost.
Equations
- One or more equations did not get rendered due to their size.
Instances For
What the contract feeds #
The two displays deliver e.assump.upper and e.assump.lower.modified.
At one admissible pair (n, n+h) the two printed displays give exactly the two
binders of RecurrenceIntegration.integrate_approx_recurrence, at
s = fun m => (Annealed.sigmaBar M m : R) ,
E_integration = (2 + shellIncrementCap) C E^2 |log cgamma|^2 ,
F = C .
hcap is produced by Closure.shellDrift_le_cap from the induction state and
the shell budget; herr is the smallness of the printed error, which
e.cgamma.constraints supplies. Unconditional otherwise.