Skip to content
nonconvexcard

Moments from tail bounds via the layer-cake formula

ProofIntermediate⏱ 30 minProbabilitytail formulamomentssub-gaussian

Problem

Suppose X≥0X\ge0 satisfies P(X≥t)≤2e−t2/K2\P(X\ge t)\le 2e^{-t^2/K^2} for all t≥0t\ge0. Using the tail formula E[Xp]=p∫0∞tp−1P(X>t) dt\E[X^p] = p\int_0^\infty t^{p-1}\P(X>t)\,dt, prove that

E[Xp]1/p≤CKp\E[X^p]^{1/p} \le CK\sqrt p

for an absolute constant CC and all p≥1p \ge 1. Compute EX\E X and EX2\E X^2 explicitly in terms of KK.

Write a complete argument. Every inequality you use should be one you could state precisely.

Hints

Hints — 0 of 2 revealed

Solution

Opening this before you have a real attempt on paper is the fastest way to feel like you understand something you cannot reproduce.

Related problems

All problems
SolvedProblemFieldTypeDifficultyTime
A variable with all moments finite that is not sub-GaussianHigh-Dim ProbabilityProofIntermediate30 min
Conditional expectation minimises mean squared errorProbabilityProofIntermediate30 min
Law of total variance and its useProbabilityProofIntermediate30 min
The 1/n variance rate requires uncorrelated samplesProbabilityNumericalIntermediate30 min
Catalan numbers are the moments of the semicircle lawRandom MatricesProofAdvanced1 hour