Finding K for integral approximation errors help

In summary, the student is having difficulty finding the errors in the approximations and is not getting close to what the book says the answers should be. The approximations T8 and M8 are 0.902333 and 0.905620. The errors in the approximations are estimated to be ET=<or=0.0078 and EM=<or=0.0039. The error formula for trapezoidal is <or= [K(b-a)3] / 12n2. If the error is strictly increasing, then the error bound is at the least value of f''(x) at x=1.
  • #1
LBK
11
0
Hello, I am having a hard time getting my errors to come out to what the book says the answers should be. My approximations are correct, so I think I'm just misunderstanding how to find K.
Q.a) Find the approximations T8 and M8 for ∫(0 to 1) cos(x2)dx
I found these to be T8=0.902333 and M8=0.905620

b) Estimate the errors in the approximations. And here's where my numbers don't match. What I did:

2nd deriv f''(x)=-2x*sin(x2)-4x2cos(x2)

since the graph is bounded by |1| I thought the max. would be at f''(0) but that would make f''(0)=0
So I'm really confused on where to go from here.
FYI--the book answer gives a value of ET=<or=0.0078 and EM=<or=0.0039

I have the formula for error for trapezoidal as <or= [K(b-a)3] / 12n2
In this case then, I should have (b-a)3=1 and 12n3=768
Working backward from the book's correct answer that would make 0.0078=K/768 and K=5.99
So I'm not seeing where that would come from. Any help, please? I'm really struggling in this class and very confused.
 
Last edited:
Physics news on Phys.org
  • #2
LBK said:
Hello, I am having a hard time getting my errors to come out to what the book says the answers should be. My approximations are correct, so I think I'm just misunderstanding how to find K.
Q.a) Find the approximations T8 and M8 for ∫(0 to 1) cos(x2)dx
I found these to be T8=0.902333 and M8=0.905620

b) Estimate the errors in the approximations. And here's where my numbers don't match. What I did:

2nd deriv f''(x)=-2x*sin(x2)-4x2cos(x2)

since the graph is bounded by |1| I thought the max. would be at f''(0) but that would make f''(0)=0
So I'm really confused on where to go from here.
FYI--the book answer gives a value of ET=<or=0.0078 and EM=<or=0.0039

I have the formula for error for trapezoidal as <or= [K(b-a)3] / 12n3
In this case then, I should have (b-a)3=1 and 12n3=768
Working backward from the book's correct answer that would make 0.0078=K/768 and K=5.99
So I'm not seeing where that would come from. Any help, please? I'm really struggling in this class and very confused.

What do you mean by ##T_8## and ##M_8##?
 
  • #3
Oh, I'm sorry that would be integral approximation by trapezoidal rule and by midpoint rule, respectively with n=8

*also original post corrected for error formula, denominator should be 12n^2 not cubed
 
  • #4
LBK said:
Hello, I am having a hard time getting my errors to come out to what the book says the answers should be. My approximations are correct, so I think I'm just misunderstanding how to find K.
Q.a) Find the approximations T8 and M8 for ∫(0 to 1) cos(x2)dx
I found these to be T8=0.902333 and M8=0.905620

b) Estimate the errors in the approximations. And here's where my numbers don't match. What I did:

2nd deriv f''(x)=-2x*sin(x2)-4x2cos(x2)

since the graph is bounded by |1| I thought the max. would be at f''(0) but that would make f''(0)=0
So I'm really confused on where to go from here.
FYI--the book answer gives a value of ET=<or=0.0078 and EM=<or=0.0039

I have the formula for error for trapezoidal as <or= [K(b-a)3] / 12n2
In this case then, I should have (b-a)3=1 and 12n3=768
Working backward from the book's correct answer that would make 0.0078=K/768 and K=5.99
So I'm not seeing where that would come from. Any help, please? I'm really struggling in this class and very confused.

For your function ##f(x) = \cos(x^2)## the second derivative should be
[tex] f''(x) = -2 \sin(x^2)-4x^2 \cos(x^2),[/tex]
which is a bit different from what you wrote. If you plot ##f''(x)## you will see that it is negative and strictly decreasing on the interval ##0 \leq x \leq 1##. Thus ##|f''(x)|## is strictly increasing.

Anyway, for Trapezoidal, the error bound is
[tex] |E| \leq \frac{K(b-a)^3}{12 n^2} = \frac{K}{12\times 8^2}= \frac{K}{768}, \: K \equiv \max_{0 \leq x \leq 1} |f''(x)|[/tex]
In your case, ##K = |f''(1)| \doteq 3.84415##, so ##|E| \leq 0.0050054 \doteq 0.005.## I cannot get the book's answer.

I think I can see where you went wrong: the actual error formula is
[tex] E = -\frac{f''(\xi) (b-a)^3}{12 n^2} = - \frac{f''(\xi)}{768}[/tex]
In this case, ##f''(\xi) < 0## so the error is ##>0##. To bound the error we want the least value of ##f''##, that is the most negative value. That is at ##\xi = 1##.
 
  • #5
I worked on it just now and I got .005 too. A classmate said he couldn't get the book answer either, so I feel much better now. And for the record, your explanation is clearer than my teacher and the book, so thank you again!
 

FAQ: Finding K for integral approximation errors help

What is "Finding K" in relation to integral approximation errors?

"Finding K" refers to the process of determining the maximum value of the second derivative of a function in order to minimize the error in approximating the integral of that function.

Why is "Finding K" important in integral approximation?

Knowing the maximum value of the second derivative of a function allows us to choose the appropriate value for K, which minimizes the error in our approximation. This ensures that our integral approximation is as accurate as possible.

How is "Finding K" calculated?

To find K, we first take the second derivative of the function and then find the maximum value of that derivative over the desired interval. This value is then used in the formula for calculating the error in the approximation.

What is the significance of the error in integral approximation?

The error in integral approximation tells us how close our calculated value is to the actual value of the integral. By minimizing this error, we can improve the accuracy of our approximation.

Are there any limitations to "Finding K" for integral approximation errors?

While "Finding K" is a useful tool for minimizing the error in integral approximation, it is not always possible to find the exact maximum value of the second derivative. In these cases, we can use estimation techniques to find a value for K that still improves the accuracy of our approximation.

Similar threads

Back
Top