MomentBoostedLowPowerBernstein #
Ported from scottnarmstrong/Superdiffusion at 7f20476385d4af418f110303085ad0740c9eba46. Module imports, support namespace and local option directives adapted for Lean Pool.
Internal two-regime Bernstein spine for the low-power large-cube route #
This module records the finite-family estimate used before the concave p / 2
power in the 1 <= p <= 2 branch. It starts from centered unit Gamma_1
summands, invokes only CoarseGraining's small-tilt moment/MGF lemma, and keeps
the quadratic and linear Chernoff regimes separate. The two terms are needed
by the later concave-power transport; collapsing them first to a Gamma_1 tail
would lose the required square-root spatial gain.
The fixed first-moment MGF coefficient in the low-power Bernstein route.
Equations
Instances For
The quadratic denominator in the raw finite-family Bernstein tail.
Equations
Instances For
The linear denominator in the raw finite-family Bernstein tail.
Equations
Instances For
Raw one-sided Bernstein concentration. A centered independent finite
family with unit two-sided Gamma_1 control obeys the displayed quadratic plus
linear tail. The statement is internal infrastructure: all its hypotheses are
discharged by the colored descendant application below the source-facing
surface.