You are on page 1of 4

INTERPOLATION

The following table gives the population of a city in U.S.A from 1950 to 2000.

Year 1950 1960 1970 1980 1990 2000


Population 151326 179323 203302 226542 249633 281422

Figure 1. The population of a city from 1950 to 2000

In reviewing the population data, we can ask whether we can estimate the population in 1985
or even in 2010. Predictions can be obtained by utilizing a function, which fits the shown data.
This technique is known as interpolation to estimate intermediate values between the given data
points.
Interpolation is the process of estimating intermediate values between the given data points.
Polynomial interpolation includes calculation of the unique nth order polynomial fitting n+1
data points. The polynomial provides a formula for calculating intermediate values.

A1) Linear Interpolation: The simplest interpolation is to connect two data points with a
straight line.
Using the similat triangles
(f1(x) – f(x0)) / (x – x0) = (f(x1) – f(x0)) / (x1 – x0)

which can be arranged to yield


f(x) = f(x0) + (f(x1) – f(x0)) / (x1 – x0) x (x – x0)
The term (f(x1) – f(x0) / (x1 – x0) given the equation is a finite-divided difference
approximation of the first order derivative
The smaller the interval between the given data points used to perform the linear interpolation,
the better the approximation.

Example: Estimate the common logarithm of 10 using the linear interpolation technique.
a) Interpolate between log 6 = 0.77815 and log 14 = 1.1461
b) Interpolate between log 8 = 0.90309 and log 12 = 1.0792
c) Interpolate between log 9 = 0.95424 and log 11 = 1.0414
Compute the percent relative error depending on the true value of log 10.

Solution: a) f1(10) = f(6) + (f(14) – f(6))/(14 – 6) x (10 – 6)


= 0.77815 + (1.1461 – 0.77815)/(14 – 6) x (10 – 6) = 0.96213
ϵt = (True value – Approximation) / True value x 100
= (1 – 0.96213) / 1 x 100 = 3.787%

b) f1(10) = f(8) + (f(12) – f(8))/(12 – 8) x (10 – 8)


= 0.90309 + (1.0792 – 0.9039)/(12 – 8) x (10 – 8) = 0.99115
ϵt = (True value – Approximation) / True value x 100
= (1 – 0.99115) / 1 x 100 = 0.885%

c) f1(10) = f(9) + (f(11) – f(9))/(11 – 9) x (10 – 9)


= 0.95424 + (1.0414 – 0.95424)/(11 – 9) x (10 – 9) = 0.99115
ϵt = (True value – Approximation) / True value x 100
= (1 – 0.99782) / 1 x 100 = 0.218%
According to the results, it is clear to say that using the shorter interval reduces the true
percent relative error.

A2) Quadratic Interpolation: One strategy for improving prediction is to introduce some
curvature in the line connecting the points; this can be accomplished with a quadratic
polynomial in the presence of three data points.
Linear interpolation formula is given below:

Quadratic interpolation formula is given below:

Example: Estimate the common logarithm of 10 using the second order Newton’s interpolating
polynomial. log 8 = 0.90309, log 9 = 0.95424 and log 11 = 1.04

Solution: f2(x) = 0.90309 + 0.05115 (x – 8) – 0.0025233 (x – 8)(x – 9)


f2(10) = 0.90309 + 0. 05115 (10 – 8) – 0.0025233 (10 – 8)(10 – 9) = 1.0003

A3) General Form of Newton’s Interpolating Polynomials: The linear interpolation formula
and quadratic interpolation formula can be generalized to fit an n-order polynomial to the n+1
data points.
The nth-order polynomial is
fn(x) = b0 + b1(x − x0)+ ………………. +bn(x − x0)(x − x1) · · · (x − xn−1)
n+1 data points can be used to evaluate the coefficients b0, b1, . . . , bn.

B) Lagrange Interpolating Polynomials: The Lagrange interpolating polynomial is the


reformulation of the Newton’s polynomial to prevent the computation of divided differences.
fn(x) = ΣLi(x)f(xi)
RERENCES:
S.C. Chapra and R.P. Canale, “Numerical Methods for Engineers”, 6th ed., McGraw-Hill,, NY,
2010
Richard L. Burden and J. Douglas Faires, “Numerical Analysis”, 9th ed., Brooks/Cole, Cengage
Learning, Canada.

You might also like