Young's Inequality alternative proof

  • Thread starter Thread starter grabthat123
  • Start date Start date
  • Tags Tags
    Inequality Proof
Click For Summary

Homework Help Overview

The discussion revolves around proving Young's Inequality, specifically the form s^(x)t^(1-x) ≤ xs + (1-x)t for s, t ≥ 0 and 0 < x < 1. Participants are exploring various approaches to establish this inequality.

Discussion Character

  • Exploratory, Mathematical reasoning, Assumption checking

Approaches and Questions Raised

  • The original poster attempts substitutions involving exponential functions and partial differentiation but struggles to reach the desired form. Other participants suggest alternative formulations and consider properties of specific functions related to the inequality.

Discussion Status

Multiple lines of reasoning are being explored, including the behavior of functions derived from the inequality. Some participants have proposed specific conditions under which the inequality may hold, while others are questioning the assumptions made in their approaches.

Contextual Notes

Participants are working under the constraints of proving the inequality without assuming certain properties, such as the convexity of the exponential function. There is also an emphasis on ensuring that the reasoning applies for different ranges of the variable a.

grabthat123
Messages
1
Reaction score
0
Young's Inequality can be restated as:

s^(x)t^(1-x)<=xs + (1-x)t where s,t>=0 and 0<x<1.

Basically I've been asked to prove this. I've been fiddling about with it for a couple of hours
to no avail.

I've tried to substitute t=e^u and s=e^v and then use partial differentiation w.r.t to v on st, but I'm not getting the required form. (I can't assume that exp is a convex function - otherwise it follows trivially)

Thanks in advance.
 
Physics news on Phys.org
I don't know how to prove it yet, but my 2 cents would be:
[tex](\frac{s}{t})^x t \leq sx + (1-x)t[/tex]
[tex](\frac{s}{t})^x \leq \frac{s}{t} x + 1-x[/tex]
[tex]a^x \leq a x + 1 - x[/tex]
for [itex]a > 0[/itex]
This way, you have one variable less to worry about
 
Actually, think about it, consider the function
[tex]f(x)=ax + 1 - x - a^x[/tex]
[tex]f(0)=1-1=0[/tex]
[tex]f(1) = a + 1 - 1 - a = 0[/tex]
[tex]f'(x_0)= a - 1 - \ln a a^{x_0} = 0[/tex]
[tex]a^{x_0} = \frac{a - 1}{\ln a}[/tex]
Now assume [itex]a>1[/itex]. Then if you can show that the rhs is smaller than a, you have shown that [itex]0 < x_0 < 1[/itex]. so
[tex]a-1 < \ln a a[/tex]
[tex]1 - \frac{1}{a} < \ln a[/tex]
I think you can find this property somewhere.
[tex]f''(x)=- (\ln a) ^2 a^{x}[/tex] which is smaller than 0 everywhere, particularly at [itex]x_0[/itex].
So for a > 1 you have already proven everything. For a < 1, you have to show that the rhs is larger than a. Let [itex]a = 1/b[/itex] with [itex]b > 1[/itex].
[tex]\frac{\frac{1}{b}-1}{\ln \frac{1}{b}} > \frac{1}{b}[/tex]
[tex]\frac{\frac{1}{b}-1}{-\ln b} > \frac{1}{b}[/tex]
[tex]1 - b < \ln b[/tex]
since b > 1, this should hold trivially.
Anyhow, it all comes down to [itex]1 - \frac{1}{a} < \ln a[/itex]. You should look it up (and make sure I haven't made any mistakes) lol
 
[tex]f(a)=\ln a a +1 - a \geq 0[/tex]
For [itex]a = 1[/itex] you have f(1)=0, so it's correct.
[tex]f'(a)= \frac{a}{a} + \ln a - 1 = \ln a[/tex]
which is always greater than 0, so you have a monotonically growing function.
 

Similar threads

Replies
6
Views
3K
  • · Replies 24 ·
Replies
24
Views
4K
  • · Replies 2 ·
Replies
2
Views
1K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 11 ·
Replies
11
Views
4K
Replies
2
Views
1K
Replies
9
Views
3K
  • · Replies 13 ·
Replies
13
Views
2K