Centering on the unit cube #
centerCell h = h - ⨍_{(0,1)²} h is the mean-zero part of an L² function on the unit cube. This
generic API is used by the grouped mean-zero assembly of BigBoundInputs (the grouped mean-zero
decomposition (12)–(18)):
three terms (twistie4, twistie5, normie3) need not be individually mean zero, only their sum
is; their centered versions are mean zero, are bounded in HDot⁻¹ through the L² estimate,
and the
centered family has the same pointwise sum as the original one.
centerCell_meanZero,l2NormSq_centerCell_le: mean zero andL²non-expansiveness;sum_centerCell_eq_of_meanZero,centerCell_add_add: centering a family whose sum has mean zero does not change the sum;continuousOn_centerCell: centering preserves joint continuity on(0,1) × ℝ².
The unit cube has volume one.
h minus its mean over the unit cube.
Equations
- AVenhance.Infra.Section5.Integration.centerCell h x = h x - ∫ (y : Homogenization.Vec 2) in AVenhance.unitCube, h y
Instances For
An L² function on the unit cube is integrable there.
The mean of a function over the unit cube is the integral (volume one).
The centered function has mean zero on the unit cube.
The centered function is L² on the unit cube.
Centering is L²-nonexpansive: ‖h - ⨍ h‖² = ‖h‖² - (⨍ h)² ≤ ‖h‖².
Centering a finite family whose sum has mean zero does not change the sum.
Three-term version of sum_centerCell_eq_of_meanZero.
Centering preserves joint continuity on (0,1) × ℝ².