Task:Unnamed inequality

From Maths
Jump to: navigation, search

Statement

Let λ(0,1)R be given. Then for all tR0 we claim:

  • tλ1λ+λt

Proof

Define:
Behaviour: <1 =1 >1
f(t)
<0
0
>0
f(t)
>0
=0
>0
So f(1) is a minimum

Broughall table for f(1)
Here d is some arbitrarily small positive value.
Reasoning: tλ1 for λ(0,1)R means λ1(1,0)R. Thus (1±d)λ1=1(1±d)1λ and now 1λ(0,1).

  1. Thus (1d)1λ<1
    • so 1(1d)1λ>1
    • so 11(1d)1λ<0
    • and finally λ(11(1d)1λ)<0
      • As λ(0,1) (i.e. "positive")
    • Thus λ(1(1d)λ1)<0
  2. Thus (1+d)1λ>1
    • so 1(1+d)1λ<1
    • so 11(1+d)1λ>0
    • and finally λ(11(1+d)1λ)>0
      • As λ(0,1)
    • Thus λ(1(1+d)λ1)>0
  • f:R0R by f:t1λ+λttλ, so f(t):=1λ+λttλ. Then:
    • f:tλλtλ1 by differentiation (for f:R>0R - we'd have to use a limit-from-the-right to find f(0) if it exists.
      • We wish to find a turning point of f(t)=λλtλ1=λ(1tλ1), that is f(a)=0 for some a
        • Suppose λλtλ1=0, then λλtλ1=λ(1tλ1)=0 and clearly:
          • If t=1 then 1tλ1=11=0 and we're done (there is no other root)
      • Thus 1 is a turning point of f (that is f(1)=0)
      • Next we must find out what f is doing. We can use a Broughall table (shown right)
        • Or we can use the second derivative:
          • f
            • The quadratic curve \lambda(\lambda-1) has roots 0 and 1 and tends towards +\infty as tends to \pm\infty, thus is standard \cup shaped.
            • The quadratic curve -\lambda(\lambda-1) is the same but flipped about the x-axis. Thus a \cap shape.
            • As \lambda\in(0,1) and this is between the roots, we see -\lambda(\lambda-1)<0 over the relevant domain.
            • t > 0 so t^\text{anything real} > 0 (remember we have to do t=0 separately, with a directed limit)
          • So -\lambda(\lambda-1)t^{\lambda-2} is a \cap-shape with respect to \lambda and >0 always on the relevant \lambda-domain.
          • We conclude f' '(1) > 0 (we can use this in the Broughall table, knowing f'(1)=0 and is increasing at 1 means f'(\text{slightly after }1)>0 and as it must increase to zero f'(\text{slightly before }1)<0, as shown
          • Thus f(1) is a minimum
      • We have shown f is at a global minimum for all relevant \lambda and all t except t=0
        • We evaluate f(1) to get f(1)=0. Thus f(t)>0 for t\in\mathbb{R}_{>0} (but not t=0 yet!)
      • We evaluate f(0) to get 1-\lambda (as t^\lambda=0^\lambda=0 as \lambda\in(0,1) and 0^x is only undefined for x=0)
        • We observe that as \lambda\in(0,1) that 1-\lambda\in(0,1) also, thus f(0)=1-\lambda>0 as required.
    • We have seen that f(t)\ge 0 for all t\in\mathbb{R}_{\ge 0}
      • Thus 1-\lambda+\lambda t-t^\lambda\ge 0 (we will no longer explicitly mention the domain of t or \lambda)
      • So 1-\lambda+\lambda t \ge t^\lambda
        • As required
  • This completes the proof.