n → ∞ lim ∫ 0 1 ( x n + ( 1 − x ) n ) n 1 d x = ?
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.
Another possible solution uses Lebesgue's Dominated Convergence Theorem . We define a sequence of functions f n ( x ) and rewrite it to: f n : D : = [ 0 ; 1 ] → R , f n ( x ) : = ( x n + ( 1 − x ) n ) n 1 = max { x , 1 − x } ( 1 + max { x , 1 − x } n min { x , 1 − x } n ) n 1 We use the rewritten form of f n ( x ) to find an upper and a lower estimate. We obtain its point-wise limit for n → ∞ via "Squeeze Theorem": x ∈ D : 0 ≤ max { x , 1 − x } ≤ f n ( x ) ≤ max { x , 1 − x } ⋅ 2 n 1 ⇒ n → ∞ lim f n ( x ) = max { x , 1 − x } = : f ( x ) A slightly sloppier estimation yields a dominating function g ( x ) of the sequence f n ( x ) ≥ 0 : n ≥ 1 : ∣ f n ( x ) ∣ = f n ( x ) ≤ f ( x ) ⋅ 2 n 1 ≤ 2 f ( x ) = : g ( x ) integrable over D The sequence f n ( x ) satisfies all prerequisites of the Dominated Convergence Theorem , so we may interchange limit and integration: n → ∞ lim ∫ D f n ( x ) d x = ∫ D f ( x ) d x symmetry = 2 ∫ 0 2 1 f ( x ) d x = 2 ∫ 0 2 1 1 − x d x = 2 [ x − 2 x 2 ] 0 2 1 = 2 ⋅ 8 3 = 4 3
Rem.: Another possible approach is to show uniform convergence of f n ( x ) → f ( x ) over the entire interval D , see @Aman Rajput 's solution . Point-wise convergence of f n ( x ) is not enough to interchange limit and integration! A nice counter-example is the "moving spike" sequence
f n : D : = [ 0 ; 1 ] → R , f n ( x ) : = n Λ ( n x ) , Λ ( x ) : = ⎩ ⎪ ⎨ ⎪ ⎧ 2 x : 2 ( 1 − x ) : 0 : 0 ≤ x < 2 1 2 1 ≤ x < 1 else We have point-wise convergence f n ( x ) → 0 = : f ( x ) for all x ∈ D , but that does not carry over to the integral: n ∈ N : ∫ D f ( x ) d x = 0 = 1 = ∫ D f n ( x ) d x
I = ∫ 0 1 ( x n + ( 1 − x ) n ) 1 / n d x
Consider the expression as x varies from 0 to 1 .
p ( x ) = 1 − x x
As x varies from 0 to 0 . 5 , p ≤ 1 and beyond x = 0 . 5 , p > 1 . Keeping this in mind, the integral can be split as follows:
I = ∫ 0 0 . 5 ( x n + ( 1 − x ) n ) 1 / n d x + ∫ 0 . 5 1 ( x n + ( 1 − x ) n ) 1 / n d x I = ∫ 0 0 . 5 ( 1 − x ) ( 1 + ( 1 − x x ) n ) 1 / n d x + ∫ 0 . 5 1 x ( 1 + ( x 1 − x ) n ) 1 / n d x
Now, since the range of p ( x ) is established the two integrands can be approximated using a binomial expansion as such:
I = ∫ 0 0 . 5 ( 1 − x ) ( 1 + n 1 ( 1 − x x ) n ) d x + ∫ 0 . 5 1 x ( 1 + n 1 ( x 1 − x ) n ) d x I = ∫ 0 0 . 5 ( 1 − x ) ( 1 + n ( p ( x ) ) n ) d x + ∫ 0 . 5 1 x ( 1 + n 1 ( p ( x ) 1 ) n ) d x
Now, when 0 ≤ x < 0 . 5 then:
p = 1 − x x < 1 ⟹ n → ∞ lim n ( p ( x ) ) n = 0 ∀ 0 ≤ x < 0 . 5
And when 0 . 5 < x ≤ 1 , then:
p 1 = x 1 − x < 1 ⟹ n → ∞ lim n ( 1 / p ( x ) ) n = 0 ∀ 0 . 5 < x ≤ 1
Applying these results to the integrands leads to (for n → ∞ ):
I = ∫ 0 0 . 5 ( 1 − x ) d x + ∫ 0 . 5 1 x d x I = 8 3 + 8 3 = 0 . 7 5
Method not correct, you didn't prove the uniform convergence! I guess you took the limit firstly and then applied integration, you can only do it if you can prove uniform convergence.
Log in to reply
I do not see the problem yet. Essentially n is a very large positive real number which is treated as a constant within the integrand. I have exploited this understanding to approximate the given integrand. I show separately why:
n ( 1 − x x ) n
converges to zero in the event where n is a very large positive real number when x lies between 0 and 0.5.
The premise of my solution appears to be correct to me because I have approximated the integrand, which I have explicitly stated. I have not yet edited my solution.
If you could point me to a useful resource regarding uniform convergence, that would be some additional reading for me as I am unfamiliar with this. Thanks in advance.
Log in to reply
sure look at this link , and read the theorems highlighted in this link https://math.stackexchange.com/questions/253696/can-a-limit-of-an-integral-be-moved-inside-the-integra
@Karan Chatrath I've posted a counter-example under my solution where we cannot interchange limit and integration even though f n ( x ) converges (point-wise) to an integrable function f ( x ) . Make a sketch of Λ ( x ) and f n ( x ) to find their integrals and see why it's called the "moving spike" ^^
The "moving spike" is one of my favorite counter-examples. its sketch really helped me understand the different types of convergence that otherwise might be quite difficult to grasp. Plus it does not depend on Lebesgue-Integration in any way!
sure look at this link , and read the theorems highlighted in this link https://math.stackexchange.com/questions/253696/can-a-limit-of-an-integral-be-moved-inside-the-integral
Problem Loading...
Note Loading...
Set Loading...
Let y = ( x n + ( 1 − x ) n ) n 1 = ( 1 − x ) ( 1 + ( 1 − x x ) n ) n 1 Now, in 0 ≤ x ≤ 2 1 0 ≤ 1 − x x ≤ 1 0 ≤ ( 1 − x x ) n ≤ 1 1 ≤ ( 1 + ( 1 − x x ) n ) n 1 ≤ 2 n 1 ( 1 − x ) ≤ y ≤ 2 n 1 ( 1 − x ) ∫ 0 2 1 ( 1 − x ) d x ≤ ∫ 0 2 1 y d x ≤ 2 n 1 ∫ 0 2 1 ( 1 − x ) d x 8 3 ≤ ∫ 0 2 1 y d x ≤ 2 n 1 8 3
for x tending to infinity 2 n 1 tends to 1 . So, the limit is : ∫ 0 2 1 y d x = 8 3
n → ∞ lim ∫ 0 1 ( x n + ( 1 − x ) n ) n 1 d x = 2 n → ∞ lim ∫ 0 2 1 ( x n + ( 1 − x ) n ) n 1 d x = 2 ⋅ 8 3 = 4 3