3
$\begingroup$

I am trying to improve my discrete Math skills by doing some proof exercises. But I am struggling to understand how to start proving the following hypothesis:

$$ S = {x}_1{y}_1 + {x}_2{y}_2 + \ldots + {x}_n{y}_n $$ Where ${x}_1,{x}_2\ldots{x}_n$ and ${y}_1,{y}_2\ldots{y}_n $ are ordering of two different sequences of positive real numbers, each containing n elements. S takes its maximum value when ${x}_1,{x}_2\ldots{x}_n$ and ${y}_1,{y}_2\ldots{y}_n $ are sorted in non decreasing order.

I was thinking that I should maybe use a proof by contradiction. Where the negation of the hypothesis is: S takes it minimum value over all ordering of the two sequences when both sequences are sorted into non increasing order.

However I am not sure my negation of the hypothesis is correct or how proceed from there. I do not want the complete proof because I am trying to learn how to do it myself but if some one can tell me how to make a start on this I would be grateful.

$\endgroup$
4
  • $\begingroup$ Proof by contradiction works like this. You intend to prove $P \Rightarrow Q$. You assume $P$ and the negation of $Q$, and argue towards a contradiction. $\endgroup$ Commented Apr 8, 2013 at 21:15
  • $\begingroup$ In particular, you don't want to negate the hypothesis. You negate the conclusion. $\endgroup$ Commented Apr 8, 2013 at 21:16
  • $\begingroup$ I assume you actually mean $x_1 y_1 + x_2 y_2 + \ldots + x_n y_n$? And your contradictory assumption needs modification. 1) Just because you assume $S$ is not a maximum does not mean that $S$ is now a minimum. 2) You don't want to change the given, which is that $x_1, x_2, \ldots, x_n$ and $y_1, y_2, \ldots y_n$ are sorted in non decreasing order. $\endgroup$
    – august
    Commented Apr 8, 2013 at 21:18
  • $\begingroup$ This is called the Rearrangement theorem. $\endgroup$
    – Aryabhata
    Commented Apr 8, 2013 at 23:29

4 Answers 4

2
$\begingroup$

Hint: consider a pair of indices $i < j$, so that $x_i \le x_j$ and $y_i \le y_j$. Now, study the difference between the expression $S$, when they are sorted, and the analogous expression $S'$, when one pair is out of order: $$ S - S' = (x_i y_i + x_j y_j) - (x_i y_j + x_j y_i) $$ Try to factor this expression to show that the difference $S - S'>0$.

$\endgroup$
3
  • $\begingroup$ Thank you. This has given me the starting point I need. $\endgroup$
    – codefi
    Commented Apr 8, 2013 at 21:36
  • $\begingroup$ I did not think to consider a subset of the series and was trying to find a more generalized solution involving both the complete series. $\endgroup$
    – codefi
    Commented Apr 8, 2013 at 21:42
  • $\begingroup$ Every permutation $(1, 2, \ldots, n) \mapsto (s_1, s_2, \ldots, s_n)$ can be written as a composition of transpositions (in which two elements are exchanged and everything else is held fixed). $\endgroup$ Commented Apr 8, 2013 at 22:15
1
$\begingroup$

The negation of the statement is: the value of $x_1 y_1 + \cdots x_n y_n$ is NOT maximum when $x$ and $y$ are sorted. This is the same as: there exist a different ordering of $x$ and $y$ such that $x_1 y_1 + \cdots x_n y_n$ is greater that the same expression with $x$ and $y$ ordered.

$\endgroup$
1
$\begingroup$

First, you need to determine what the hypothesis is and what the conclusion is.

To proceed with a proof by contradiction, you ASSUME the hypothesis and NEGATE the conclusion. Then argue to reach a contradiction.

If the implication you are trying to prove is as follows:

"if ("when") $x_1,x_2, …, xn$ and $y_1, y_2, …y_n\;$ are sorted in non decreasing order, then $S$ takes a maximal value"

then the conclusion is "$S$ takes a maximal value."

The negation of the conclusion is "$S$ does not take a maximal value". There exists some other, greater $S' > S$ such that S is not maximal.

Then you affirm the hypothesis, state the negation of the conclusion, and show that a contradiction result.

$\endgroup$
2
  • $\begingroup$ I may have been mixing up proof by contraposition and proof by contradiction but even then I would have been wrong. I have chosen Samuel Blacks answer because he has given my the hint I need to start on this proof. Thank you for clear explanation. $\endgroup$
    – codefi
    Commented Apr 8, 2013 at 21:47
  • $\begingroup$ Glad to help! I certainly would have given you a prompt or suggestion for approaching this if you had followed up with a comment below my answer. I just thought it was crucial, first, to make sure you understood what you want to contradict, and had the proper negation, to start you off on sound footing, if you went that route. If you want to check back on your progress, feel free to comment. $\endgroup$
    – amWhy
    Commented Apr 8, 2013 at 22:55
0
$\begingroup$

Using the hint by Sammy Black.

Proof: Assume that $i<j, x_i<x_j$ and $y_i<y_j, S$ as the sum of which both x and y sequence are sorted, and without loss of generality, $S'$ is the sum in which x is sorted and y is not in nondecreasing order. Thus subtracting $S'$ from $S$, we get:

$$ S - S' = x_iy_i +x_iy_i - x_iy_j - x_jy_i$$

Factoring $y_i-y_j$ give us: $$ S - S' = (y_i - y_j)(x_i - x_j)$$

Because $y_i<y_j \implies y_i-y_j<0$ and $x_i<x_j \implies x_i-x_j<0$ we can conclude that: $$(y_i - y_j)(x_i - x_j)=S-S'>0$$ $$S>S'$$ $\therefore$ The sum in which x and y are sorted is the maximum value. $\blacksquare$

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .