An inequality that bounds the integral (or sum) of the product of two measurable functions by the product of their L^p and L^q norms when the exponents p and q are conjugate (1/p + 1/q = 1).

Principle

For 1 ≤ p,q ≤ ∞ with 1/p + 1/q = 1, if f∈L^p and g∈L^q then fg is integrable and ∫ |f g| ≤ ||f||_p ||g||_q. The general form extends to finite families via Hölder's inequality for multiple factors with exponents summing reciprocals to 1.

Demonstration

Example: with p=q=2 on a probability space, Hölder reduces to Cauchy–Schwarz: |∫ f g| ≤ ||f||_2 ||g||_2. Concretely, for square-integrable signals f and g the inner product is bounded by the product of their L^2 norms.

Misapplication

Attempting to apply Hölder with exponents that are not conjugate or with functions not in the stated L^p spaces is invalid; similarly, using Hölder for p<1 fails because L^p is not a normed space there and the inequality does not hold.

Consequence

Hölder provides a fundamental tool for estimating integrals and sums, proving duality between L^p and L^q spaces, and deriving other inequalities (Minkowski, Young) and boundedness of linear operators.

Reversal

There is no general reverse inequality: one cannot bound ||f||_p ||g||_q from above by ∫ |f g|. The equality case characterizes proportionality conditions between |f|^p and |g|^q (up to null sets).

Boundary

Valid for 1 ≤ p,q ≤ ∞ with conjugacy 1/p+1/q=1; excludes quasi-norm regimes p<1 and requires membership of f and g in the respective L^p and L^q spaces.

Semantic Tension

Tensions appear with Minkowski (triangle inequality in L^p) and Young's convolution inequality; conceptually Hölder trades integrability between factors whereas other inequalities trade summation or convolution structure.

Synthesis

Hölder's inequality asserts that the integral of a product is controlled by the product of L^p norms for conjugate exponents, underpinning norm duality and norm estimates across analysis.