Which is entity is greater and by how much ?

Calculus Level 4

Let A = 0 1 x x d x \displaystyle A = \int_0^1 x^x \, dx and B = 0 1 x x d x \displaystyle B = \int_0^1 x^{-x} \, dx .

Evaluate the value of 10000 A B \big \lceil \, 10000 |A - B| \, \big \rceil .

Notations:


The answer is 5079.

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.

1 solution

Guilherme Niedu
Sep 11, 2017

A = 0 1 x x d x \large \displaystyle A = \int_0^1 x^x dx

A = 0 1 e x ln ( x ) d x \large \displaystyle A = \int_0^1 e^{x \ln(x) } dx

A = 0 1 k = 0 [ x ln ( x ) ] k k ! d x \large \displaystyle A = \int_0^1 \sum_{k=0}^{\infty}\frac{ [x \ln(x)]^k}{k!} dx

A = k = 0 0 1 [ x ln ( x ) ] k k ! d x \large \displaystyle A = \sum_{k=0}^{\infty} \int_0^1 \frac{ [x \ln(x)]^k}{k!} dx

Make u = ln ( x ) u = -\ln(x) , d u = 1 x d x d x = e u d u du = -\frac1x dx \rightarrow dx = -e^{-u} du

A = k = 0 ( 1 ) k k ! 0 e u ( k + 1 ) u k d u \large \displaystyle A = \sum_{k=0}^{\infty} \frac{(-1)^k}{k!} \int_0^{\infty} e^{-u(k+1)} u^k du

A = k = 0 ( 1 ) k k ! ( k + 1 ) k 0 e u ( k + 1 ) [ u ( k + 1 ) ] k d u \large \displaystyle A = \sum_{k=0}^{\infty} \frac{(-1)^k}{k! \cdot (k+1)^k} \int_0^{\infty} e^{-u(k+1)} [u(k+1)]^k du

Make t = u ( k + 1 ) t = u(k+1) , d t = d u ( k + 1 ) dt = du(k+1)

A = k = 0 ( 1 ) k k ! ( k + 1 ) ( k + 1 ) k 0 e t t k d t \large \displaystyle A = \sum_{k=0}^{\infty} \frac{(-1)^k}{k! \cdot (k+1) \cdot (k+1)^k} \int_0^{\infty} e^{-t} t^k dt

A = k = 0 ( 1 ) k k ! ( k + 1 ) k + 1 Γ ( k + 1 ) \large \displaystyle A = \sum_{k=0}^{\infty} \frac{(-1)^k}{k! \cdot (k+1)^{k+1}} \Gamma(k+1)

A = k = 0 ( 1 ) k k ! ( k + 1 ) k + 1 k ! \large \displaystyle A = \sum_{k=0}^{\infty} \frac{(-1)^k}{k! \cdot (k+1)^{k+1}} k!

Make n = k + 1 n = k+1

A = n = 1 ( 1 ) n + 1 n n \color{#20A900} \boxed{ \large \displaystyle A = \sum_{n=1}^{\infty} \frac{(-1)^{n+1}}{n^n} }

Likewise:

B = 0 1 x x d x \large \displaystyle B = \int_0^1 x^{-x} dx

B = 0 1 e x ln ( x ) d x \large \displaystyle B = \int_0^1 e^{-x \ln(x) } dx

B = 0 1 k = 0 [ x ln ( x ) ] k k ! d x \large \displaystyle B = \int_0^1 \sum_{k=0}^{\infty} \frac{[-x \ln(x)]^k}{k!} dx

B = k = 0 0 1 [ x ln ( x ) ] k k ! d x \large \displaystyle B = \sum_{k=0}^{\infty} \int_0^1 \frac{ [-x \ln(x)]^k}{k!} dx

B = k = 0 ( 1 ) k 0 1 [ x ln ( x ) ] k k ! d x \large \displaystyle B = \sum_{k=0}^{\infty} (-1)^k \int_0^1 \frac{ [x \ln(x)]^k}{k!} dx

We already know this integral from A A :

B = k = 0 ( 1 ) k ( 1 ) k ( k + 1 ) k + 1 \large \displaystyle B = \sum_{k=0}^{\infty} (-1)^k \frac{(-1)^k}{(k+1)^{k+1}}

B = k = 0 1 ( k + 1 ) k + 1 \large \displaystyle B = \sum_{k=0}^{\infty} \frac{1}{(k+1)^{k+1}}

Make n = k + 1 n = k+1

B = n = 1 1 n n \color{#20A900} \boxed{ \large \displaystyle B = \sum_{n=1}^{\infty} \frac{1}{n^n} }

So, A B A-B will cancel out each term of odd n n , remaining only the terms of even n n :

A B = n = 1 2 ( 2 n ) 2 n \large \displaystyle |A-B| = \left | \sum_{n=1}^{\infty} \frac{-2}{(2n)^{2n}} \right |

A B = 2 n = 1 1 ( 2 n ) 2 n \large \displaystyle |A-B| = 2 \sum_{n=1}^{\infty} \frac{1}{(2n)^{2n}}

This sum quickly converges to:

A B = 0.507855486 \color{#20A900} \boxed{ \large \displaystyle |A-B| = 0.507855486 }

Correct to 9 9 decimal places. Thus:

10000 A B = 5079 \color{#3D99F6} \boxed{ \large \displaystyle \left \lceil 10000 |A-B| \right \rceil = 5079 }


Proof of convergence of the series above:

( 2 n ) 2 n 2 2 n \large \displaystyle (2n)^{2n} \geq 2^{2n}

1 ( 2 n ) 2 n 1 2 2 n \large \displaystyle \frac{1}{(2n)^{2n}} \leq \frac{1}{2^{2n}}

n = 1 1 ( 2 n ) 2 n < n = 1 ( 1 4 ) n \large \displaystyle \sum_{n=1}^{\infty} \frac{1}{(2n)^{2n}} < \sum_{n=1}^{\infty} \left ( \frac14 \right) ^n

n = 1 1 ( 2 n ) 2 n < 1 3 q . e . d . \color{#20A900} \boxed{\large \displaystyle \sum_{n=1}^{\infty} \frac{1}{(2n)^{2n}} < \frac13 } \color{#333333} \ \ q.e.d.

We don't need Stirling's approximation to prove convergence. Since ( 2 n ) ! (2n)! is the product of the integers 1 , 2 , 3 , . . . , 2 n 1,2,3,...,2n , all less than or equal to 2 n 2n , it is clear that ( 2 n ) ! < ( 2 n ) 2 n (2n)! < (2n)^{2n} . For that matter, convergence is shown even more simply by noting that ( 2 n ) 2 n 2 2 n (2n)^{2n} \ge 2^{2n} , and comparing the series with a GP.

Mark Hennings - 3 years, 9 months ago

Log in to reply

Indeed, much easier!

Guilherme Niedu - 3 years, 9 months ago

Can I edit my solution to incoporate your proof of convergence?

Guilherme Niedu - 3 years, 9 months ago

Log in to reply

No worries. Of course, the fact that 0 n = 0 N ( 1 ) n ( ln x ) n x n n ! x x 0 x 1 , N N 0 \; \le \; \sum_{n=0}^N \frac{(-1)^n (\ln x)^n x^n}{n!} \; \le \; x^{-x} \hspace{2cm} 0 \le x \; \le 1\;,\; N \in \mathbb{N} and the fact that x x x^{-x} is continuous on [ 0 , 1 ] [0,1] means that 0 n = 0 N 1 ( n + 1 ) n + 1 0 1 x x d x 0 \; \le \; \sum_{n=0}^N \frac{1}{(n+1)^{n+1}} \; \le \; \int_0^1 x^{-x}\,dx for all N N , which means that the series is convergent without further comment (and the Monotone Convergence Theorem makes the infinite sum equal to the integral),.

Mark Hennings - 3 years, 9 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...