Why is Minkowski's inequality for L^p spaces considered a foundational result, beyond being a useful estimate?
AIt provides a computable formula for calculating the L^p norm of a sum of two functions
BIt establishes the triangle inequality, thereby certifying that L^p with the p-norm is a normed vector space
CIt shows that L^p functions are bounded almost everywhere
DIt generalizes the Cauchy-Schwarz inequality to arbitrary exponents p
The triangle inequality — ‖f + g‖_p ≤ ‖f‖_p + ‖g‖_p — is the hardest of the three normed space axioms to verify for L^p (positivity and scaling are straightforward). Without it, L^p would have a notion of size but no guarantee that adding functions behaves consistently with that size. Minkowski's inequality supplies this missing axiom, completing the verification that L^p is a normed vector space. Combined with completeness (Riesz-Fischer), it makes L^p a Banach space. Option D conflates Minkowski with Hölder, which actually generalizes Cauchy-Schwarz.
Question 2 Multiple Choice
A student tries to prove ‖f + g‖_p ≤ ‖f‖_p + ‖g‖_p for 1 < p < ∞ by using the pointwise inequality |f(x) + g(x)| ≤ |f(x)| + |g(x)|, raising both sides to the p-th power, and integrating. Why does this approach fail?
AThe pointwise triangle inequality |f + g| ≤ |f| + |g| is false for general L^p functions
BRaising both sides to the p-th power and integrating does not give the right bound, because (a + b)^p ≠ a^p + b^p for p > 1
CThe approach fails because L^p does not contain pointwise-defined functions
DThe pointwise inequality gives a lower bound for |f + g|, not an upper bound
The pointwise triangle inequality is valid, but (a + b)^p ≥ a^p + b^p when a, b ≥ 0 and p > 1 — so integrating after taking the p-th power gives ∫|f + g|^p ≤ ∫(|f| + |g|)^p, which does NOT imply ‖f + g‖_p ≤ ‖f‖_p + ‖g‖_p. The correct proof must use Hölder's inequality: factor |f + g|^p = |f + g| · |f + g|^(p-1), apply the pointwise bound to the first factor, then apply Hölder to each integral term. The nonlinearity of the p-th power is precisely what makes this case nontrivial.
Question 3 True / False
Minkowski's inequality for L^p relies on Hölder's inequality in its proof for 1 < p < ∞.
TTrue
FFalse
Answer: True
The proof for 1 < p < ∞ works by factoring ‖f + g‖_p^p = ∫|f + g| · |f + g|^(p-1) dμ, bounding the first factor pointwise, and applying Hölder's inequality with conjugate exponents (p, q) where 1/p + 1/q = 1 to each resulting integral. The Hölder conjugate relationship makes the algebra close: ‖|f+g|^(p-1)‖_q = ‖f+g‖_p^(p-1), and dividing both sides by this factor yields the inequality. Hölder is the engine of the proof; Minkowski is a consequence.
Question 4 True / False
Minkowski's inequality ‖f + g‖_p ≤ ‖f‖_p + ‖g‖_p holds for most p > 0.
TTrue
FFalse
Answer: False
Minkowski's inequality requires p ≥ 1. For 0 < p < 1, the triangle inequality *fails* — the quantity (∫|f|^p dμ)^(1/p) does not define a norm because it is not subadditive. This is why L^p spaces as normed (and Banach) spaces require p ≥ 1. The boundary cases p = 1 and p = ∞ have direct proofs; the Hölder-based argument handles 1 < p < ∞. Knowing where the inequality breaks is as important as knowing where it holds.
Question 5 Short Answer
Explain why the p = 1 case of Minkowski's inequality is immediate, and why the case 1 < p < ∞ requires a more indirect argument.
Think about your answer, then reveal below.
Model answer: For p = 1: ‖f + g‖₁ = ∫|f + g| dμ ≤ ∫(|f| + |g|) dμ = ‖f‖₁ + ‖g‖₁. This works directly because integration is linear and the pointwise triangle inequality integrates as-is. For 1 < p < ∞: the p-th power is nonlinear, so integrating (|f| + |g|)^p produces mixed cross-terms rather than a clean separation. The proof must factor |f + g|^p = |f + g| · |f + g|^(p-1), apply the pointwise bound to the first factor, and then use Hölder's inequality to control each integral. The Hölder conjugate relationship makes the algebra close.
The p = 1 case is elementary because addition and integration interact simply when the exponent is 1. The nonlinearity introduced by p-th powers for p > 1 requires the indirect approach via Hölder. This is a recurring pattern in analysis: linear estimates come directly, nonlinear ones require a conjugate tool. Hölder pairs two functions via conjugate exponents and provides the leverage needed to control the nonlinear interaction — which is why Minkowski's inequality is, in this sense, a corollary of Hölder's.