Hölder’s inequality
Let
where
Proof
Let
and 𝐴 = ‖ 𝑓 ‖ 𝑝 . If 𝐵 = ‖ 𝑔 ‖ 𝑞 or 𝐴 = 0 the inequality holds trivially, and similarly if 𝐵 = 0 or 𝐴 = ∞ . 𝐵 = ∞ Now assume
, 𝐴 , 𝐵 ∈ ( 0 , ∞ ) and take 𝑝 , 𝑞 ∈ ( 1 , ∞ ) . Taking Young’s inequality with 𝑥 ∈ 𝑋 and 𝛼 = | 𝑓 ( 𝑥 ) | 𝐴 gives 𝛽 = | 𝑔 ( 𝑥 ) | 𝐵 𝛼 𝛽 = | 𝑓 ( 𝑥 ) 𝑔 ( 𝑥 ) | 𝐴 𝐵 ≤ | 𝑓 ( 𝑥 ) | 𝑝 𝑝 𝐴 𝑝 + | 𝑔 ( 𝑥 ) | 𝑞 𝑞 𝐵 𝑞 = 𝛼 𝑝 𝑝 + 𝛽 𝑞 𝑞 whence
‖ 𝑓 𝑔 ‖ 1 ‖ 𝑓 ‖ 𝑝 ‖ 𝑔 ‖ 𝑞 = 1 𝐴 𝐵 ∫ 𝑋 | 𝑓 ( 𝑥 ) 𝑔 ( 𝑥 ) | 𝑑 𝜇 ( 𝑥 ) ≤ 1 𝑝 𝐴 𝑝 ∫ 𝑋 | 𝑓 ( 𝑥 ) | 𝑝 𝑑 𝜇 ( 𝑥 ) + 1 𝑞 𝐵 𝑞 ∫ 𝑋 | 𝑔 ( 𝑥 ) | 𝑞 𝑑 𝑥 = 𝐴 𝑝 𝑝 𝐴 𝑝 + 𝐵 𝑞 𝑞 𝐵 𝑞 = 1 𝑝 + 1 𝑞 = 1 wherefore
‖ 𝑓 𝑔 ‖ 1 ≤ ‖ 𝑓 ‖ 𝑝 ‖ 𝑔 ‖ 𝑞 The only case yet to be handled is
. It follows immediately that 𝑝 = ∞ , 𝑞 = 1 almost everywhere and thus | 𝑓 𝑔 | ≤ ‖ 𝑓 ‖ ∞ | 𝑔 | , as required. ‖ 𝑓 𝑔 ‖ 1 ≤ ‖ 𝑓 ‖ ∞ ‖ 𝑔 ‖ 1
The Cauchy-Schwarz inequality for the [[Lebesgue space|
Footnotes
-
If in addition, if
and𝑝 are finite,𝑞 , and𝑓 ∈ 𝐿 𝑝 ( 𝜇 ) , then𝑔 ∈ 𝐿 𝑞 ( 𝜇 ) iff‖ 𝑓 𝑔 ‖ 1 = ‖ 𝑓 ‖ 𝑝 ‖ 𝑔 ‖ 𝑞 are linearly dependent. ↩| 𝑓 | 𝑝 , | 𝑔 | 𝑞 ∈ 𝐿 1 ( 𝜇 )