Scribd
Upload
44 views1 page

Problem Sheet 02: Ax BX C

This document contains 3 problems related to computational techniques: 1) Calculating the roots of an equation using 5-digit arithmetic results in round-off errors. 2) Using Taylor series expansions to approximate functions, the relative error decreases as more terms are included for finite series but not necessarily for infinite series like logarithms. 3) Taylor series expansions of exponentials and cosines are infinite series. The number of terms needed to achieve a relative error of 10^-5 depends on the function. Truncating after the first term, the error decreases linearly with step size on a log-log plot.

Uploaded by

Daulot Sarma
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
44 views1 page

Problem Sheet 02: Ax BX C

This document contains 3 problems related to computational techniques: 1) Calculating the roots of an equation using 5-digit arithmetic results in round-off errors. 2) Using Taylor series expansions to approximate functions, the relative error decreases as more terms are included for finite series but not necessarily for infinite series like logarithms. 3) Taylor series expansions of exponentials and cosines are infinite series. The number of terms needed to achieve a relative error of 10^-5 depends on the function. Truncating after the first term, the error decreases linearly with step size on a log-log plot.

Uploaded by

Daulot Sarma
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Computational Techniques

NPTEL Video Course

Problem Sheet 02

Problem 1: Round-off Errors

The roots of the equation ax 2  bx  c  0 can be computed in the following two ways:

b  b 2  4ac 2c
x1 , x2  or x1 , x2  .
2a b  b 2  4ac

Use five-digit arithmetic with chopping (i.e., the sixth digit is discarded) to compute the roots of
the equation x 2  5000.002 x  10  0 from the two formulae given above.

Problem 2: Taylor’s Series for Finite and Infinite Series

Consider the following functions. It is known that f(1) = 0 for both these cases. Use zero to
fourth order Taylor’s series expansion to obtain f(3). The two functions are:

f  x   18 x 4  25 x3  6 x 2  7 x  44 and f  x   ln  x  .

Comment on the relative error of the two cases, as more terms are included from the Taylor’s
series expansion.

Problem 3: Taylor’s Series and Effect of Step Size

 Using the Taylor’s series expansion:


h2 hn n
f  a  h   f  a   hf   a   f   a     f a 
2! 2!
Obtain the infinite series expansion of ex and cos(x). [Assume f(0) is known and h = x]

 For each of these functions, calculate the function values at x = /4, each time increasing
the number of terms used by 1 until the relative error reduces to 10-5 (i.e., report the relative
error when 1, 2, 3, … terms are included from the infinite series). Compute the true relative
error at each stage. Observe that the truncation error reduces as the number of terms
increase. How many terms are required to reduce the relative error to 10-5?

 Alternatively, we can truncate the infinite series at the first order term hf   x  . Compute the

value of cos(/4) using this first-order approximation, using h = /4, /8, /16, …, /64.
Plot error vs. step size (h) on a log-log plot and comment about its slope.

Computational Techniques

You might also like