Distance of the boundary contour from ρ + 1 #
The geometry of the shifted contour t ↦ fdBoundary H t - (ρ + 1) about the corner
ρ + 1 at parameter t = 1: the purely imaginary linear form along the right vertical,
the chord distance along the arc, the norm lower bounds on the far pieces, the
closed-upper-half-plane confinement — and the exact endpoint logarithms beside the corner.
The confinement holds only once the ceiling clears the corner row, and the confinement
results accordingly carry √3/2 ≤ H. Under that hypothesis the shifted contour touches the
branch cut at t = 3, where its value is -1, but never crosses it; at the degenerate
height H = √3/2 the whole left vertical degenerates to that point and lies on the cut
throughout. Below the corner row the claim fails outright: on the left vertical the shifted
contour is -1 + (t - 3)(H - √3/2)i, so for H < √3/2 the imaginary part is negative
immediately after t = 3 and the contour does cross the cut.
The corner joins the vertical to the arc at the interior
angle π/3, which is exactly the gap between the one-sided argument limits π/2 and
5π/6 — the source of the winding value -1/6 at ρ + 1.
Main declarations #
TauCeti.ModularForm.rho_add_one_im(the corner shares the rowIm = √3/2withρ).TauCeti.ModularForm.fdBoundary_sub_rho_add_one_of_mem_Icc_zero_one(the linear form).TauCeti.ModularForm.norm_fdBoundary_sub_rho_add_one_arc(the chord distance) andTauCeti.ModularForm.norm_fdBoundary_sub_rho_add_one_arc_le(its monotonicity in the angular gap).TauCeti.ModularForm.log_fdBoundary_one_sub_sub_rho_add_one,TauCeti.ModularForm.log_fdBoundary_one_add_sub_rho_add_one(the endpoint logarithms).
References #
- AINTLIB
LeanModularForms— the valence-formula development (ForMathlib/ValenceFormula/WindingWeights/RhoPlusOne.lean) this file ports onto the current Mathlib pin.
The corner ρ + 1 sits on the same row Im = √3/2 as ρ: the shift is by a real
number, which does not move the imaginary part.
Not @[simp], for the same reason as rho_im: simp reaches ρ.im through
Complex.add_im, UpperHalfPlane.coe_im and Complex.one_im before this could fire.
On the right vertical the shifted contour is the purely imaginary linear form
(1 - t)·(H - √3/2)·i.
On the arc the distance from ρ + 1 is the chord distance: 2·sin(|t - 1|·π/12)
up to the absolute value inside the sine.
Along the arc the chord to ρ + 1 grows with the parameter. A point of the arc at
parameter t is no further from the corner than the point at parameter 1 + δ is, whose
chord is 2·sin(δ·π/12).
This is the generic monotonicity of a chord in the angular gap,
TauCeti.dist_circleMap_le_dist_circleMap_of_abs_sub_le, read along the arc, whose
parameter runs at π/6 per unit; δ ≤ 6 keeps the comparison point within half a turn of
the corner, where the chord is still increasing.
On the left vertical the shifted contour is -1 plus the imaginary linear form of the
ρ-shift, so its real part is constantly -1 and it runs alongside the branch cut. Where the
ceiling misses the corner row the imaginary part vanishes only at t = 3, so the cut is met
there alone — above it when the ceiling clears the row, below it otherwise. At the degenerate
height H = √3/2 the imaginary part vanishes identically and the whole segment is the point
-1, lying on the cut throughout.
On the left vertical the contour keeps distance at least 1 from ρ + 1: the real parts
differ by exactly 1.
On the ceiling the contour keeps distance at least |H - √3/2| from ρ + 1. The
absolute value is what the height difference gives, and it is the only form with content
below the corner row, where H - √3/2 < 0 makes the signed bound vacuous.
On the arc the shifted contour stays in the closed upper half-plane.
Strictly inside the arc the shifted contour has positive height.
The principal logarithm of the shifted contour just after the corner.
The contour passes through ρ + 1 only at the corner t = 1.