Can Sum to Product Inequalities Hold for Non-Negative Reals?

Join the discussion
Registration is free. Start your own thread to ask a follow-up.
5 replies · 2K views
lfdahl
Gold Member
MHB
Messages
747
Reaction score
0
Given non-negative reals, $\alpha_i$, where $i = 1,2,...,n.$

Prove, that

$\alpha_1+\alpha_2+...+\alpha_n \leq \frac{1}{2}$ $\Rightarrow$ $(1-\alpha_1)(1-\alpha_2)...(1-\alpha_n) \geq \frac{1}{2}.$
 
Physics news on Phys.org
Immediate consequence of Weiertrass inequality:

$$\prod_{1 \le k \le n}(1-a_k) \ge 1-\sum_{1 \le k \le n}a_k \ge 1-\frac{1}{2} = \frac{1}{2}.$$

There's an elegant proof of Weierstrass inequality.
 
June29 said:
Immediate consequence of Weiertrass inequality:

$$\prod_{1 \le k \le n}(1-a_k) \ge 1-\sum_{1 \le k \le n}a_k \ge 1-\frac{1}{2} = \frac{1}{2}.$$

There's an elegant proof of Weierstrass inequality.
Thankyou for a clever solution, June29, and for your participation!(Cool)
 
Can anyone prove the above statement by induction? (Wave)
 
lfdahl said:
Can anyone prove the above statement by induction? (Wave)

It's obviously true for $n=1$ since we have $ {\alpha}_1 \leqslant \frac{1}{2} = 1-\frac{1}{2} \implies 1-\alpha_1 \geqslant \frac{1}{2}. $

Now, suppose it's true for $n = k \in \mathbb{N}$. We shall prove that it's true for $n=k+1$. $\displaystyle \begin{aligned} \frac{1}{2} & \leqslant 1- \sum_{1 \leqslant j \leqslant k+1}\alpha_j = 1-\alpha_{k+1}-\sum_{1 \leqslant j \leqslant k}\alpha_j \leqslant 1-\alpha_{k+1}-\sum_{1 \leqslant j \leqslant k}\alpha_j+a_{k+1}\sum_{1 \leqslant j \leqslant k} \alpha_j \\& =\left(1-\alpha_{k+1}\right)\left(1-\sum_{1 \leqslant j \leqslant k} \alpha_j\right)
\leqslant \left(1-\alpha_{k+1}\right) \prod_{1 \leqslant j \leqslant k} \left(1-\alpha_j \right) = \prod_{1 \leqslant j \leqslant k+1} \left(1-\alpha_j \right) \end{aligned} $

So it's true for $n=k+1$. Since it's true for $n=1, k+1$, it's true for all $n\in\mathbb{N}$.
 
Last edited by a moderator:
June29 said:
It's obviously true for $n=1$ since we have $ {\alpha}_1 \leqslant \frac{1}{2} = 1-\frac{1}{2} \implies 1-\alpha_1 \geqslant \frac{1}{2}. $

Now, suppose it's true for $n = k \in \mathbb{N}$. We shall prove that it's true for $n=k+1$. $\displaystyle \begin{aligned} \frac{1}{2} & \leqslant 1- \sum_{1 \leqslant j \leqslant k+1}\alpha_j = 1-\alpha_{k+1}-\sum_{1 \leqslant j \leqslant k}\alpha_j \leqslant 1-\alpha_{k+1}-\sum_{1 \leqslant j \leqslant k}\alpha_j+a_{k+1}\sum_{1 \leqslant j \leqslant k} \alpha_j \\& =\left(1-\alpha_{k+1}\right)\left(1-\sum_{1 \leqslant j \leqslant k} \alpha_j\right)
\leqslant \left(1-\alpha_{k+1}\right) \prod_{1 \leqslant j \leqslant k} \left(1-\alpha_j \right) = \prod_{1 \leqslant j \leqslant k+1} \left(1-\alpha_j \right) \end{aligned} $

So it's true for $n=k+1$. Since it's true for $n=1, k+1$, it's true for all $n\in\mathbb{N}$.

A nice solution, June29! Thankyou for your participation!

Please remember to hide your solution in SP tags. Other forum users might try to solve the challenge preferably without knowing your solution. Thankyou in advance!