Expected Difference Surds

Calculus Level 5

Let m , n m,n be two randomly chosen real numbers satisfying that the equation x m + m = x n + n \sqrt{x-m}+m=\sqrt{x-n}+n has a single real solution x = x 1 x=x_1 .

If x 1 m + m = x 1 n + n = P \sqrt{x_1-m}+m=\sqrt{x_1-n}+n=P , then the expected value of P x 1 P-x_1 can be expressed as p q \dfrac{p}{q} for positive coprime integers p , q p,q . Find p + q p+q .


The answer is 7.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

2 solutions

Daniel Liu
Jul 21, 2014

Let y = x k + k y=\sqrt{x-k}+k or sine real k k . We rearrange to get y k = x k y-k=\sqrt{x-k} . This is the equation of the y = x y=\sqrt{x} graph translated k k units right and up.

This means we can assume WLOG that m = 0 m=0 . We thus have to solve the equation x = x n + n \sqrt{x}=\sqrt{x-n}+n

First off, note that as long as n [ 1 , 1 ] n\in [-1,1] then there will be a solution x 1 x_1 . This can be easily seen by sliding the graph x \sqrt{x} along the line y = x y=x .

Then, we can proceed to solve the equation:

Squaring both sides: x = x n + n 2 + 2 n x n x=x-n+n^2+2n\sqrt{x-n} 1 n = 2 x n 1-n=2\sqrt{x-n}

Squaring both sides again: n 2 2 n + 1 = 4 ( x n ) n^2-2n+1=4(x-n) x = n 2 + 2 n + 1 4 = ( n + 1 ) 2 4 x=\dfrac{n^2+2n+1}{4}=\dfrac{(n+1)^2}{4}

Thus, x 1 = ( n + 1 ) 2 4 \boxed{x_1=\dfrac{(n+1)^2}{4}} . This means P = x 1 = n + 1 2 P=\sqrt{x_1}=\dfrac{n+1}{2} .

We want to find the expected value of P x 1 = n + 1 2 ( n + 1 ) 2 4 P-x_1=\dfrac{n+1}{2}-\dfrac{(n+1)^2}{4} in the range n [ 1 , 1 ] n\in [-1,1] . Thus we want to find the value of the constant c c such that 1 1 n + 1 2 ( n + 1 ) 2 4 c d n = 0 \int_{-1}^1\dfrac{n+1}{2}-\dfrac{(n+1)^2}{4}-c\text{ d}n=0

It is easy to compute that 1 1 n + 1 2 ( n + 1 ) 2 4 d n = 1 3 \int_{-1}^1\dfrac{n+1}{2}-\dfrac{(n+1)^2}{4}\text{ d}n=\dfrac{1}{3} and 1 1 c d n = 2 c \int_{-1}^1c\text{ d}n=2c so 1 3 2 c = 0 c = 1 6 \dfrac{1}{3}-2c=0\implies c=\dfrac{1}{6}

Thus, the expected value of P x 1 = 1 6 P-x_1=\dfrac{1}{6} and our answer is 1 + 6 = 7 1+6=\boxed{7} .

David Vaccaro
Jul 23, 2014

y = x m x n n m = 1 x m + x n y=\frac{\sqrt{x-m}-\sqrt{x-n}}{n-m}=\frac{1}{\sqrt{x-m}+\sqrt{x-n}}

The right-hand-side is continuous and decreasing on x m a x ( n , m ) x\geq max(n,m) , and tends to 0 as x x goes to infinity. So by Intermediate Value Theorem we have a unique solution for y = 1 y=1 at x = x 1 x=x_{1} provided n m < 1 |n-m|<1

At x 1 x_{1} we have x m + x n = 1 \sqrt{x-m}+\sqrt{x-n}=1 so adding the two expressions for P P gives 2 P = 2 + m + n 2P=2+m+n . Combining this with x 1 = ( P m ) 2 + m x_{1}=(P-m)^{2}+m gives with very little pain that: P x 1 = 1 ( n m ) 2 4 P-x_{1}=\frac{1-(n-m)^{2}}{4}

Taking Expectations gives that p q = 1 V a r ( n m ) 4 \frac{p}{q}=\frac{1-Var(n-m)}{4}

It is not completely clear that everything does not depend on the distribution of n n and m m and technically there is no "Uniform" distribution on the unbounded region n m < 1 |n-m|<1 but assuming ( n m ) (n-m) is uniformly distributed on [ 1 , 1 ] [-1,1] gives the variance as 1 3 \frac{1}{3} and the required value as 1 6 \boxed{\frac{1}{6}} .

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...