28

Finite Differences and Interpolation

1.Finite diflerences. 2 Differences of a polynomial. 3. Pactorial notation. 4. Relations botween
theoperators. 5. Newton's interpolation formulae. 6. Central difference interpolation formulae
Gauss's interpolation formulae; Stirling's formula; Bessel's formula; Everett'sformula 2
Choice ofan interpolation formula. 8. Interpolation with unequal interv lsLagrange'sformula
Newton's divided differenceformula. 9. Inverse interpolation. 10. Numerical differentiation
Maxima &minima ofa tabulated function. 12. Numerical integration Newton-Cote's
rule.
fcrmula
ezoidal rule Simpson's 1/3rdrule, Simpson's 8/8th rule; Weddle's
28-1. FINITE DIFFERENCES
Suppose we are given the following values ofy =f(x) for a set of values ofx:
*1 2 n
1 Y2 yYn
Then the process of finding the values of y corresponding to any value of x = x; between xo

the value of a
and x, is called interpolation. Thus interpolation is the technique of estimating
variable while the process of computing
function for any intermediate value of the independent
called extrapolation. The study of the
the value of the function outside the given range is
differences of a function which we proceed to discuss. For
interpolation is based on the concept of
a detailed study, the reader
should refer to author's book Numerical Methods in Engineering
and Science'.
that the function y =f«) is tabulated for the equally spaced values x x0, Xo + h, =

Suppose or f"(x) for some

Yn. To determine the values off ()
X0+ 2h,.. X0 + nh
givingy yo
=
1 . Y2 . ,

found useful:
intermediate values of z, the following three types of differences are

(1) Forward differences. The

differences y1-0.J2 -1, n-Yn -1 when denoted by
where A is the forward
are called the first forward differences
Ayo, Ay1,., Ayn-1 respectively
forward differences are Ay, =yy+1-y.
difference operator. Thus the first
Similarly the second forward
differences are defined by
A y, = Ay,r+1-Ayr

In general, P'y, =P'y.1-P-y,

defines the pth forward differences.
These differences are systematically set out as follows in what is called a Forward Difference Table.
In a difference table, x is called the argument andy the function or the entry. yo», the first

entry is called the leading term and Ayo, ayo, ayo ete. are called the leading differences.
875
876 HIGHER ENGINEERING MATHEMATICS

Obs. Any higher order forward difference can be expressed in terms of the entries.
We have A'Yo = Ay1- Avg =(2-'1)-01-o) =ya -21+Yo

'Yoa-a'o= 0a-3ys+3ya-yi)-(s-3y+3y1 -o) =y-43+6-47 *o

The coefficients occurring on the right hand side being the binomial coefficient, we have in general,
yo= -"1 n -1+ "cayn-2..+(- 1'%
Forward Difference Table
Value of x Value of y 1st. diff 2nd diff. 3rd diff 4th diff. 5th diff.
yo
Ayo
y1
A1 yo
X0+ 2h 2

A2
X0+3h 3
y
o+4h
y4
y2
Ay4
o+ 5h
Y5
(2) Backward differences.
The
Vy1. Vy2,. differences y1-X0 J2-Y1, ..., yn-Yn -1 when denoted
Vy respectively, are called the first backward by
difference operator. Similarly we define differences where V is the backward
higher order backward differences. Thus
we have
Vy, =y,-yr-1.Vy,= Vy,- Vy, -1
These differences
vy,=- V¥-1 etc.
are exhibited in the following
Backward Difference Table
Value of x Value of y 1st. diff. 2nd diff. 3rd diff.
Yo 4th diff. 5th diff.

x0 + h
Vy1
y2
y2
xo + 2h
y2

x0 + 3h Y3

x0 + 4h 4 ys
VY5
0+5h Y5
FINITE DIFFERENCES AND INTERPOLATION
877
(3) Central differences. Sometimes it is convenient to employ another system of differences
central differences. In this system, the central difference operator 8 is defined by the
relations 1 - y o = oy1/2, 2 - y 1 = öy3/2 YnYn - 1 Yn - 1/2

Similarly, higher order central differences are defined as

oy3/2-dy1/2=8y1, ôy5/2 -Ôy3/2=8y2,..

8y2-8y1 =8'y3/2 and so on.
These differences are shown in the following
Central Difference Table

Value of x Value of y 1st. diff 2nd diff. 3rd diff. 4th diff. 5th diff

öy1/2
ro +h 1
8 Y3/2 &'y3/2
xo +2h
y5/2 &'ys/2 y5/2
x0+ 3h y3
y1/2 y1/2
xo + 4h y4
o y9/2

0 +5h y5

differences on the same horizontal line have the same

We see from this table that the central
known only for half values of the suffix and those of
suffix. Also the differences of odd order are
even order for only integral
values of the suffix.
mean of adjacent values in
the same column of differences.
It is often required to find the
by H. Thus
We denote this mean

etc.
poy1 =~ôy12+ys/),uö'Ys/2 =~6y1+8y2)
notation which changes and not the differences.
Obs. The reader should
note that it is only the
1-Yo= Ay% Vy1 =dy1/2
=

those involving central differences are most useful in practice as

Of all the interpolation formulae,

decrease much more rapidly.
formulae
the coefficients in such

2x) (ii) A(*/cos 2x) (iv) A cos 2x

Example 28-1. Evaluate (i)Atan"x (ii) (e° log
A
(Madras, 1996 S)

Sol. i) A tan"x = tan" (x + h) tan'x

-

tan +h-x_ tan-1

=tan+a+h)xtan 1+hx+2
878 HIGHER ENGINEERING MATHEMATICS
() A(e log 2x) = e**k log 2 (r + h) - * log 2
=* log 2(x + h) *h log 2x+er*h log 2x
-

log 2x
=eh log +( +h -x) log 2x

log|1.4|-1) log2
) a r+h (r +h cos 2x-x cos 2(x +h)
cos 2x cos 2(r + h) cos 2x cos 2(r + h) cos 2a

_Lx+h-x] cos 2: +x (cos 2-cos 2 (x+h]

cos 2 (x +h) cos 2x

(2hx +h cos 2x+2x sin (h) sin (2 +h)

cos 2 (x +h) cos 2x
(iu) a cos
2x =A (cos 2 (x +h) -

cos 2x)
=
A cos 2
(x +h) -A cos 2x
=
[cos 2 (x +2h) cos 2 (x + h)]
[cos
-

2 (x +h) -

cos 2]
=-2 sin (2x +
3h) sin h +2 sin (2x+h) sin h
= -

2 sin h
[sin (2x +3h) sin (2x -

+h)]
=-2 sin h (2 cos
(2x + 2h) sin h] =
4 sin h (2r +2h).
-

cos
Example 28 2. Evaluate (i) A ¥+12_
6x+16 16) () 4 (ab")
interval of differencing
being unity.
(Rohtak, 2003)
Sol.)2
Sol. ) 2-+12=?x+12
x+6x+66 +2 +3)=a+2x+3
1
-2 +2) ( +3) -34 1
*+3) (x 4) +

-2
3)&+4+2)(*+2) (r +3) 4
+4) (x + 5) +3)
(*+3) & +454)
(x+
x
+2) (x +3) (x +4) (x+ 6
3) (* + 4) (x +5) 2 5x +16)
(iü) A (ab) =a A(B) =a (6*** b) (x +2) (* + 3) (x + 4) (x +5)
=ab* (6 1) -

a (ab) = a (A ab'] a (b 1) A(B*) = -

=a (b 1) (63* - b5) =a (b 1)*

b&
-

(tt) Ae = ë= (e 1) e"
*
-
-

Ae=A(A)= Ale -1) e)

(e 1) Ae =(e-1)(e 1) e (e 1
=
= -

Similarly A'¥ (e 1)', s'e =(¢-1)",.. and a

= -

=
(e -

1ye.
FINITE DIFFERENCES ANDINTERPOLATION 879
282. DIFFERENCES OF APOLYNOMIAL
The nth differences of a polynomial of the nth degree are constant and all higher order
differences are zero.

Let the polynomial of the nth degree in x, be

f(x) =ar" +bx"-+ cx"-2 . . + k(x +h) +
Af x) = f ( +h) - f(*)

= alx +h" -x"] + bl« +h"--x"-)+ ... +kh

= anhx"l+b'"-2 +c'"-3+... +k'* + .1)
where 6, c,'are new constantco-efficients.
Thus the first differences of a polynomial of the nth degree is a polynomial of degree
(n 1).
Similarly a f ) = Alf( + h)-fw)] = 4f («+ h) - 4f()
anhlx + hy"-1-n-)+b lt +h -2-x-)+..+kh
an(n - 1)h'-2+bx-3 + c ' = ' t ... +k", by (1))
T h e second differences represent a polynomial of degree (n - 2).

Continuing this process, for the nth differences we get a polynomial of degree zero i.e.

A fc)=ann -

1)0n -2)...1 h =an !h" ..(2)

which is a constant. Hence the (n + 1)th and higher differences of a polynomial of nth degree will
be zero.
Obs. The converse of this theorem is also true i.e. if the nth differences ofa function tabulated at
equally spaced intervals are constant, the function is a polynomial of degree n. This fact is important in
numerical analysis as it enables us to approximate a function by a polynomial of nth degree, if its nth
order differences become nearly constant.

(Madras, 1996 S)
Example 28-3. Evaluate A"[1 -ax)(1 bx*1 ex*(1 dx*))
- -
-

Sol. I1-ax)(1 -bx1 -cxx1-d«)=a1° [abcd 10+()+ ()z*+.. +1

abed al a19 I a ) = 0 for n < 10)
= abed (10 ) by (2) above.

28-3. (1) FACTORIAL NOTATION

and is called factorial.
A product oftheform x(*
-

1) (x -2).. (x-r+1) is denoted by el a

In particular x = x, k l = x(* -1)

1= x ( -1)(x-2), etc.
" = x(x - 1)(r - 2 ) . . . x - n + 1)
In general
In case, the interval of differencing is h, then
lal" = z{r -h)x - 2h)...x-n - 1h)
The factorial notation is ofspecial utility in the theory offinite differences. It helps in finding
the successive differences ofa polynomial directly by simple rule of diferentiation.
The result of differencing bwl is analogous to that of diferentiating x'.
(2) To express a polynomial in the factorial notation
(i) arrange the coefficients of the powers of x in descending order, replacing missing powers by zeros
(i) using detached coefficients divide by x, x - 1, x-2,etc. successively.
880 HIGHER ENGINEERING MATHEMATICS
Obs. Every polynomial ofdegreen can be expressed as a factorial polynomial ol the same degree and

vice versa.
Example 28-4. Express y =2 - 3 x + 3x- 10 in a factorial notation and hence show that

ay=12
Sol. Let y =Aul+Bul+ Cxl+D.
Then

1 2 3 - 10 D
-3
2 -1
2 2 -1 2C

3 2 3 = B

2A
Hence y =2«1+3t«l+2] - 10
Ay=2x 3]+3 x 2x]+2
Ay = 6x2x] +6
Ay 12, which shows that the third differences ofy are constant,
=

Example 28-5. Obtain the function whose first difference is

as they should be.
9+11x+5. (Madras, 1996)
Sol. Let f (x) be the
required function so that
Afx) =
9+ 1lr +5 =
9tel+Ale]+B
9x(x - 1) + Ar + B (sayl
Putting = 0, we get B = 5.
Putting x = 1, we get A =20.

Af(x) = 9xl +20[]+5

Integrating, we get f(r) =9+3

where c is the constant of
20+5+c =33++x+ce
integration.
Example 286. Find the missing values in
45 50
the following table:
55 60
y: 3.0 65
2.0
Sol. The difference -2.4
table is as follows: (V.T.U, 2001)
y
45 0 3 ay
50
1-3
1
2-1
5-21
55 22 3y1 +3-9
1+y3-4
Ya-2
3.6-1-3y3
60 3 -2.4-3
65 Y-24
-0-4-2y3
FINITE DIFFERENCES ANDINTERPOLATION 881
As only three entries yo. y2, Y4 are given, the function y can be represented by a second degree
polynomial

ayo=0 and
ie. 3y1 +y3=9; 1+3y3=3-6.
Solving these, we get y= 2925, y3 = 0-225.
Otherwise: As only three entries yo 3, y2 =2, y4 =- 24 are given, the function y can be
represented by a second degree polynomial.

a yo=0 and a=0

1.e. (E- 1 o 0 and (E-1 0
Le.
i.e.
(ES-3E+3E-1)yo =0 and (E 3E+3E-1)y1 =0
y3-3y2+3y1 -Yo=0
Y4-3y3+3y2-J1 =0
Le
y3+3y19;3y3 +y1=3.6
Solving these, we get y1=2925, y3 = 0-225.
Example 287. Assuming that the following values ofy belong to a polynomial of degree 4,
compute the next three values:
1 2 3 4 5 6 7
1 1
(Assam, 1998, Tiruchirapalli, 1996 S)
Sol. We construct the following difference table from the given data:

Ay
o1
2
1 Y1-1

2 8
A 16
2 2=1

-2 8
16
3 Y3-1

41
16

Ay4
5 16
y5
Ays
Y6

Ay6
7
Since the values of y belong to a polynomial of degree 4, the fourth differences must be
constant. But Ay = 16.
882 HIGHER ENGINEERING MATHEMAIICS

The other fourth order differences must also be 16. Thus

aty- 16 =aya - a*y

.e.. ay= ay+Ay= 8+ 16-24

Ay= t+a'y2- 4 +24 28
Av yst Ays=2+ 28 30
and Ys4+ Ay4= 1 +30 = 31
Similarly starting with A'y2= 16, we get
Ay= 40, Ay4 = 68, Ays = 98, ys = 129.
Starting with Ay3 = 16, we obtain
a'ys = 56, Ays = 124, Ays = 222, y7 = 351.

Problems 281
1. Write forward difference table if
10 20
y 1-1 30 40
2. 2-0 4.4
Construct the table of differences for the data below: 79
0
2 3
fx) 10 1-5 2.2
4
Evaluate af (2). 3.1 4-6
3.
If u =
3, u =
12,
ug 81, ug =2000, u4 100,
=

4.
calculate Auo. =

Show that ay,

5. Form the table
= +2t 3i yi+3-3yi +1-yi
of backward
differences of the function (Madras, 1996)
fx) = -3x 5x-7 for x = -1,0,
6. Form a table of differences for
the function
1,2,3,4,5.
fx)=+ 5x 7
for x= -

1,0, 1, 2, 3, 4, 5.
Continue the table to obtain
7. Extend the f(6).
following table to two more terms on either side
by constructing the difference table
23 0-0
3-0
0-2
3-4
0-4 06
8. Show that 428 08 10
7-08 14-2 29-0

9. Evaluate
fx) f r+ 1) Gi)
alog f) log{1+ fCx))
=

() A (x + cos x)

10. Evaluate:
natan (i) a+5x+6)
(i) a" (a)
(Burdwan, 2003)
11. Prove that
() ys =y2 +41 +Ayo+ A'yo (ii)
12. Evaluate: ys=ys- 2y7+y6&'ys =
() A I(1-x)(1-2x) (1-3x)] ys-2y5 +4
i) a (1 -x)(1-2x) (1 3r)(1-4x*), -

if the interval of differencing is 2.

FINITE DIFFERENCES AND INTERPOLATION
883
13. Express r-2 +*- 1 into factorial
polynomial. A.M.I.E., 1997)
14. Express u =x* 12" + 24x- 30r+ 9 and its successive differences in factorial notation.
-

that a'u =0. Hence show

15. Express 3r -

4+ 6x2
+ 2x+ l as a factorial
polynomial and find differences of all orders.
16. Find the first and second differences
difference is constant.
of - 6+ 11x -

5x+8 with h =1. Show that the fourth

17. Obtain the function whose

first difference is 2x+ 3x-6x + 4.
18. Estimate the missing term in the following table
2 3 4
f(x): 3 81
(Bhopal, 2002 S; Coimbatore, 1997
19. Find the missing values in the following table
1 2 3 4 5
5 11 22 40 140
(V.T.U. 2003 S)
20. Find the first term of the series whose second and subsequent terms are 8, 3, 0,-1, 0.
Nagpur, 1997)
21. If u(x) and v(x) be two functions ofx, prove that
()A lu(x) v(x)) = u(r) A u(x) + v(* + 1) A u(x). (Madras, 1996)
(i) A)vE)Au(r)-ulx) Aut)
vr) v(x) v(x + 1)

28-4. (1) OTHER DIFFERENCE OPERATORS

We have already introduced the operators A, V and 8. Besides these, there are the operators
E and , which we define below
(i) Shift operator E is the operation of increasing the argument x by h so that

E ft)=fa +h), E*fx) =f(*

The inverse operator E '
+ 2h), E' f) =f(* +3h) ete.
is defined by E * ' f « ) = f ( x - h)

Ify, is the function fo), then Ey,Ys +h. ">»=x-h»B =s + nh»

where n may be any real number.
(i) Averaging operator p is defined by the equation 4Hy = 0 +h/2 +yr - h/2)

regarded as the fundamental operators and V, 8, u can be

Obs. In the difference calculus, A and E are

expressed in terms ofthese.

(2) Relations between the operators. We

shall now establish the following identities:
(i) V =1 -E
(i)A =E-1
(iii) 8 =El/2-e-1/2 (iu) =(E+E-1/2
(v) A EV VE =8E/2
= = = (vi') E = ehD.
Proofs. (i) Ay, = ys +h -yx = By-yx = (E- 1)y

This shows that the operators A and E are connected by the symbolic relation

A E-1 or E = 1 +A.

(iü) Vy -s-h =) -E y =
(1-E )y
V=1-E' or E =(1 - V)*
 886 HIGHER ENGINEERING MATHEMATICs

2
4 Au u+244 =R.H.S.
Problems 282
1. Explain thedifference between (2 4, and (Madras, 2003)
E
2. Evaluate taking h as the interval of differencing
)sinx ",(h = 1) (Madras, 1996 S)

hl 4 sin (r+h)
s i n (x + h)+ i v ) (a+Vj'«*+x), th =1).
g sin (x +h)
3. With the usual notations, show that
) A =1-er hD (ii) =2 sinh
ii) (1+A1 - V) = 1. (iv) AV =VA =8* (Madras, 1996)
4. Prove that

) 8=A(1+A)"/2=V(1 V)-1/2 -

i)p=1+
(i) 8(El/2+E-1/2) = AE +A
5. Show that (i)
uð={(A+ V) (ii) 1+8/2 = V(1 +8°4
(Madras, 1998)
i~i) v=h*D?- h^D"+ h*D-. (iv) a+V=- (Madurai, 1996 S)
6. Prove that

G)f=fh-r Ci) A= +fh+ 1) Af

ii) E=
(iv) El/2=(1+8/4)1/2+8/2.
7. Prove that V yn+1 =

8. Using the method of separation of symbols,

prove that
(i) uo +U1+Ug t..+ 4, ="* "C140+ *"CaAug+
(ii) y, =yn- n* CaA^uo +..+R*C,+1A"ug
9.
"*C]4yn -1+ C2A°yn -2 -..+(-1" =* A" =*y
"**

Using the method of finite differences, sum the Yn-(n -*)

i) 2.5+ 58+8-11+ 11.14+.. to n terms. following series:
(ii) 12-3 +2-3-4+3-45 +... to n terms. (Madurai, 1996 S)
10. Prove that uo +ux + ux +. (Coimbatore, 1997)

Hence sum the series 1-(1-(1-

12+23x +34x+.
3 . 0o

28-5. NEWTON'S INTERPOLATION FORMULAE

We now derive two important interpolation formulae by means of
the forward and
differences of a function. These formulae are often employed in backward
Newton's forward interpolation formula.
(1)
engineering and scientific problems.
Let the
Yo... corresponding to the values Xo, *0 + h, x0
Ya. Y1, +
functiony=f(x)take the values
evaluate fx) for x = *o +ph, where p is any real number.
2h,.. of x.
Suppose it is required to

See foot note p. 431.

FINITE DIFFERENCES AND INTERPOLATION
887
For any real number p, we have defined E such that
EP fx) =
fx + ph)
p f *o +ph) = B°f (to) = (1 + A)Pyo
E =1 +A
=1+pa+P)2,Pp-Dp-2)
3!
a.Yo using Binomial theorem
L.e.
p=y0+payo+P yo +PPDe-2
3! 4yot. . ...1)
It is called Newton's forward interpolation formula as (1) contains yo and the forward

differences of yo
Obs. This formula is used for interpolating the values of y near the beginning of a set of tabulated
values and extrapolating values
of y a little backward (i.e. to the left) of Yo
(2) Newton's backward interpolation formula. Let the function y =f(x) take the values
Yo-1 Y2. corresponding to the values xo, zo +h, xo +2h,. of x. Suppose it is required to
evaluatefx) for x = xo +ph, where p is any real number. Then we have

p fn+ph) =B° fn)= (1- V)P yn E=1-V]

=1+pV+P 2!
2,PP +1p
3!
+2) y3+ yn lusing Binomial theorem)
i.e. .(2)
3!
It is called Newton's backward interpolation formula as (2) contains y', and backward
differences of yn
Obs. This formula is used for interpolating the values of y near the end of a set of tabulated values
and also for extrapolating values of y a little ahead (to the right) ofyn
Example 2811. The table gives the distances in nautical miles of the visible horizon for the
given heights in feet above the earth's surface:
x = height 100 150 200 250 300 350 400
y =distance: 1063 13-03 15-04 16-81 18-42 19.90 21-27
Find the values of y when x = 218 ft and 410 ft.
Sol. The difference table is as under:
A 4 A

100 1063
2.40
- 0-39
150 13-03
2-01 0 15
- 024 - 0-07
200 15-04
1.77 0-08
- 0-16 0-05
250 1681
1-61 0-03
- 0-13 - 001
300 1842
1.48 0-02
350 19.90 - 0.11

137
400 21-27
888 HIGHER ENGINEERING MATHEMATICS
()Ifwe take xo = 200, then yo = 16-04, Ayo = 1-77, Ayo = - 016, A'yo =003 etc

18
Since x 218 and h =
50, pa 036
h 50
Using Newton's forward interpolation formula, we get

Ayo+PP-p-2) avo+.
218 o+pAyo+ 12 3
f(218) 1504+ 036(177) + 0-36(-0-64) (-016) + 0-36(-0-64)(- 164) (0 03)+
2 6 (0-03)+.
1504 0637+0-018+0-001.. = 15-696 i.e. 157 nautical miles
Since =410 is near the end of the table, we use Newton's backward interpolation
formula.

Taking , =
400, p= 02
Using the line of backward differences
yn21-27, Vyn = 137, vy = - 0-11, vy, = 0-02 etc.
Newton's backward formula gives

V410 Y400 pVy400 +

2 a00+ 2P+ 1p +2) y3.400+
1 2.3
21-27+0-2(137) + (-0-11) + . =21-53 nautical miles.
Example 2812. From the following table, estimate the number of students who
marks between 40 and 45: obtained
Marks 30-40 40-50 50-60
No. of Students 31 42
60-70
70-80
51 35
31
Sol. First we prepare the cumulative
frequency table, as follows (P.T.U., 2003)
Marks less than (x)
40 50
No. of Students 60 70
(,) 31 73 124
80
Now the difference table is 159 190

Ayx
40 31

4
50 73
51
60 124 -25
-16
35 37
70 12
159
31
80 190
FINITE DIFFERENCES AND INTERPOLATION
We shall find ya5 i.e. 889
number of students with marks less than 45.
Taking xo =40, =45, we havep= x 5
1005 h =10
Using Newton's forward interpolation formula, we
get
y45=Y40+ pdyso+ 2 sy40 +PP 1 2 3 0t.
=31 +0-5x 42 +5-0:5), a 05(0-5(-1-5),oEi 0-5-0-5-15(-25)27
2 6 24
47-87,on simplification.
The number of students with
.
marks less than 45 is 47-87 i.e., 48.
But the number of students with
marks less than 40 is 31.
Hence the number
of students getting marks between 40 and 45 =48-31 17. =

Example 28 In the table below, the values of y are consecutive terms a series which
13.
23-6 is the 6th term. Find the of of
first and tenth terms of the series:
3 4 5 6 8 9
y: 4-8 8-4 145 23-6 36-2 52-8 739
Sol. The difference table is

Ay
4-8
3.6
4 8.-4 2-5
6-1 0-5
5 145 3-0
9.1 0-5
6 23-6 3-5
12-6 05
36-2 4-0 0
16-6 0-5
8 52 8 4-5
211
9 73.9

To find the first term, use Newton's forward interpolation formula with xo = 3,x = 1, h =1

andp =-2. We have

(1) = 48 + x36+2)x 25+ (3)-4),0.5 =31
y 12 123
To obtain the tenth term, use Newton's backward interpolation formula with x, = 9, x = 10,

h 1 andp = 1. This gives

y(10)=73-9+x21-1+x
12
45 123
x05= 100.
Example 28-14. Find the cubic polynomial which takes the following values:
2 3

1 2 10
f(x)
Hence or otherwise evaluate f (4). (Madras, 1997 S)
 890 HIGHER ENGINEERING MATHEMATICS
Sol. The difference table is

f(a) Afx) af)

1 2 -2
-1 12
10

3 10

We take xo = 0 and p =*=«

h h =1]
Using Newton's forward interpolation formula, we get
f=f0)+4f0) 2 ()+*1*-2)
12 3
f
=
1 +x(1) + -2) +*-1t-2) (12)
6
2x3 7 +6ax +1, which is
the required polynomial.
To
compute f(4), we take z, =
3, x =4 so
that p= =1
Using Newton's backward interpolation
h=1
formula, we get
f4)=f(3) +pVf (3) +Pp pf(3)+ PD +1 Dp+2)
2 3 f (3)
=
10 +9 + 10 + 12 + 41.
which is the same value
that obtained by as
substituting * 4 in the cubic polynomial above. =

Obs. The above example shows that

extrapolation give the same values. if a tabulated function is a polynomial, then
interpolation and
Problems 28:3
1. Using Newton's forward formula, find the value off (16), if

1 14 18 22
flx) 3.49 4-82 5-96 6.5

2. State Newton's interpolation formula and use it to0 calculate (Kottayam, 1999; Tirupati, 1998)
the value
table : of exp (1-85), given the following
17 18 19 20 21 2:2 2.3
f): 6474 6-050 6-686 7-389 8-166 9025 9974
3. Iff(1-15) =1-0723,f(120)= 1-0954,f(125) 1-1180 andf(130) 11401,
= =

find f (128).
4. Given sin 45°=0-7071, sin 50° 0 7660, sin 55° 08192, sin 60°=
0-8660, find sin 52, using
=
=

Newton's forward formula.

FINITE DIFFERENCES AND INTERPOLATION 891
5. From the following table:
0-1 0.2 03 0.4 05 06

fx) 2.68 3-04 3.38 3 68 3 96 421

find f(0-7) approximately.

6. The area A of a circle
of diameter d is given for the following values
d 80 85 90 95 100
A: 5026 5674 6362 7088 7854

Calculate the area of a circle of diameter 105.

(V.T.U., 2004)
7. Estimate the values of f (22) and f (42) from the following available data:

20 25 30 35 40 45

fx): 354 332 291 260 231 204

8. Find the number of men getting wages between Rs. 10 and 15 from the following data

Wages in Rs. 0-10 10-20 20-30 30-40

9 30 35 42
Frequency
9. From the following data estimate the rumber of persons having incomes between 2000 and 2500

Below 500 500-10000 1000-2000 2000-3000 3000-4000

Income
6000 4250 3600 1500 650
No. of persons
10. Construct Newton's forward interpolation polynomial for the following data:

6 10

3 8 16
1
(Madras, 2000 S)
Hence evaluate y for x = 5.

Newton's backward difference formula to the data below, to obtain a polynomial of degree 4 in
11. Apply

1 2 3 4
1 -1 1 -1
of a series of which 125 is the 5th term.
values of y are consecutive terms
12. In the following table, the
of the series
Find the first and tenth terms
5 6 7 8
3 4 729
12-5 216 34-3 512
2.7 6.4
(Panjab, 1997)

13. Given uj =
40, ug =
45, us =
54, find ug and u4.
14. If u =
10, u1 =
8, u 10, u4 50, find ug and u3.
= =

without forming the difference table, find

=3, y1 - 12, y2 81, y3 200, y4 100, Ys 8,
= = = =
15. Given yo

Ayo
 892 HIGHER ENGINEERING MATHEMATICS

28 6. CENTRAL DIFFERENCE INTERPOLATION FORMULAE

In the preceding section, we derived Newton's forward and backward interpolation formulae
which are applicable for interpolation near the beginning and end of tabulated values. Now we
shall develop central difference formulae which are best suited for interpolation near the middle
of the table.
If x takes the values xo - 2h, xo - h, xo, *o + h, xo + 2h and the corresponding values of
table in the two notations as
y=f(x) are y-2. y- Y2, then we can write the difference
1 Yo» Y1,
follows:

Ist diff. 2nd diff. 3rd diff. 4th diff.

o -2h y-2

Ay-2 (= oy-3/2)
Xo-h y-1
Ay-1(= ôy- 1/2) ay-a(8y- 1/
yo
Ayo=oy1/2) ay-81
+h Y1

Ay1=8y3/2)
o+2h Y2

(1) Gauss's forward interpolation formula. The Newton's forward interpolation formula is
YpYo tp Ayo + (pAyo +PP)p-2) 3
1 2 3 1)
We have
Ayo-Ay-1=4'y-1
i.e. ayo =Ay-1+A°y- 1 2)
Similarly Ayo=Ay-1 +Ay-1 .(3)
Ayo=A^y1+A'y-1 etc. ..4)
Also Ay-1-4-2 =4^y- 2
i.e. Ay =Ay2+Ay-2
Similarly A-1=A'y-2 +A°y- 2 etc. .5
Substituting for Ayo, Ayo, A0 .from (2), (3), (4).. in (1), we get
p yo tp Ayo +* 1-2 y1+ay-)+P2 y-1+ay-
P(p-1) 12.3.4
(p-2) (p-3) (ay+ A`Y- 1) *.
Hence p =yo +p4yo + y 1 4 P*)p
3!
(p-1) ,
p+ 1)pp-1p-2)
4! y-2tusing (5)
interpolation formuda.
which is called Gauss's forward
FINITE DIFFERENCES AND INTERPOLATION

Cor.In the central differences 893

notation, this formula will be
p Yo+phy1/2 2 ya+ P+Dp (p -) '1/2* P+ 1)p p- 1)p-2) o+
3! 4!
Obs. 1. It employs odd
.

shownbelow: differences just below the central line and even

difference on the central line as
Xo
2 A°y 3 ..Centralline
Ayo
Obs. 2. This formula is used
to
a'y.3
the origin. interpolate the values of y for p (0<p < 1) measured forwardly from

(2) Gauss's backward

formula is interpolation formula. The Newton's forward interpolation

p =Yo+pAyo + )yo+2P=)p-2)
12-3
yn+ ...(1)
We have Ayo 4y1 4^y-1
L.e.
Ayo Ay1+Ay-1 ..(2)
Similarly o= ay-1 +Ay-1 ...(3)
Ayo=A*y-1+Ay.1 ete. ...4)
Also y1-ay-2 =ay-2
L.e. ay-1 =ay-2 +Ay- 2 ..(5)
Similarly 4y-1 = Ay-2 +A'y- 2 ete. .6)

Substituting for Ayo, Ayo, A°y0, . from (2), (3), (4) in (1), we get

=yo*p (y- 1+4y-1) + 12

- 1+ay-1)+2PPa'y-1+4'y.
123
)
P(p-1) 12.3.4
(p-2) p-3) (A'y-1+a'y-1)+..
=yo +pAy1+ 2 y1+P*9pp=) 3,,+(p+ 1)p (p-1) p-2)
1-2 1-2-3 12-3.4

P p-1)(p-2)p-3) y+...
12.3.4

+pay-+ ay, p+1)p

+ (p-
1-2.3
+ ay-2)
12
(p+ 1)p (p-1p-2) (ay-2+45y_2)t... (using (5) and (6)
12-3.4

p+o+py-1+ y+P+)p(p-1) -2
Hence 3!

+2) (p+ 1)pp-1) y-2t .

4!
which is called Gauss's backward interpolation formula.
894 HIGHER ENGINEERING MATHEMATICS

Cor. In the central differences notation, this formula will be

=o+po y. 1/2+ P 8y+P*p p-) s, +2)(p+4!)p (p-1) s ,

3!
Obs. 1. This formula contains odd differences above the central line and even differences
central line as shown below;
on the
Ay 3

Yo. ** y- aty. . y_3 Central line

Obs. 2. It is used to
interpolate the values of y for a negative value ofp lying between 1 and 0. -

Obs. 3. Gauss's forward and backward

formulae are not of much practical use. However, these
as intermediate steps for obtaining the serve
important formulae of the following sections.
(3) Stirling's formula.* Gauss's
forward interpolation formula is
p =0+payo+ y-1+Ptp3!P=)
+1p (p-1)p-2 y-2*.
Gauss's backward 4! .
.1)
interpolation formula is
p =yo+pAy-1+ 2! y +P*1p
3! p-) -2
P+2) (p+ 1)p(p-1)
Taking the mean of (1) and (2), we 4! (2)
obtain
Yp Yo+ P Ayo+y-1
3!
2

which is called Stirling's formula.

Cor. In the central
differences notation,
ay-at. 4! .3)

(3) takes the

form
for
(4yo+ y)=1/2+ôy- 1/2)= 3!
&yo
differences
y+y.)-(®2+8y-
Obs. This
formula this line
involves means of the odd
1/)=u &y ete.
on
as shown below: differences just above and below
the central line and even
Ay Ay-1 y
y Ay- 2 Ay-3 ... Central line
Bessel's formula.** Gauss's
(4)
forward
interpolation formula is
p =yo +pAyo+ 2 !
We have
Pp2y-1+P+p 3! y-1 ,P+)p(p-
(p-Dp-2),
ie.
ayo-ay1ay.1 4!
y-2 .1)
ay =yo-ay1
Named after the Scotish
**
See foot note p. 506. mathematician James Stirling (1692-1770). (2)
FINITE DIFFERENCES AND INTERPOLATION
895
Similarly A'y2 Ay.1- A'y_ 2etc. ..(3)
Now (1) can be written as

p y o +payo + 1)(1
P P-
3'

+1pp-1)
4
p-2) sy_+1Pp"- p-2) x(ay-1 -4y- z) +... using (2), (3) ete.
+pAya +2P-1),ay-1+AY% (p- 1
=Yo
2
p
2 -a
p (p-1)(p-2) 4y- 2 +4y-1
4! 2

Hence yp =yo+payo +p .41+4% (p-1/2)p P-)y-

2 3!

(P +1)p(p-1) (p - 2) Ay_2+A'y-1 ..(4)

4! 2
which is known as the Bessel's formula.

Cor. In the central differences notation, (4) becomes

, =Yo+pôy1/2+ oy/2+ ppDsy+P+ 1)p(p-1)p-2)

for ( y a y o ) =yö'y1/n(ay2+a'y- = uöy1/2 ete.

Obs. This is a very useful formula for practicalpurposes. It involves odd diferences below the centro!
line and means ofeven differences of and below his line as shown below:

Yo y Ay2 y-3. Central line

Ayo ay2
| ay-1}
5) Everett's formula. Gauss's forward interpolation formula is

y1 + Pt)p p*) +P+ 1)p (p4!-1)p-2) ,4

p =yo +p4yo+ 2! 3!

(p +2) (p + 1)p (p-l) P=a ay_2t....(.1)

5!

by using the relations

We eliminate the odd difference in (1)

Ayo 1-yo, A'y-1 =A*yo-4y- 1, A°y- 2 =a'y1 4y-2

etc. -

Then (1) becomes

p =o tpo1 -o)+ 2! y1+ pP=

3!
(4ya -A?y)
p+ 1p p-1 p-2 ay 2+ P+2)p + Dp (p-Dp-2) x (a'y- 1- ay. 2)+... .

4! 5!
896 HIGHER ENGINEERING MATHEMAICS

( P) Yo Py pp 1)G 2) y P D p (p- 1) *yo

3 31
2) 3) +2) (p +)p (p-1)(p- 2)
51
To change the torma with negative aign, puttinggp 1 - q, we obtain

92o y q -19-2 ay2t. +py1+p p1) yo

3 6 3
p p2-1) p°-2)
This
5! s'y1.
i known as verett'» formula.
Obs. This for mula is extensively used and involves only cven differences on and below the central line
s shown below

Y Ay I..Ay..Ay .Central line

Ay Ay Ay
28-7. CHOICE OF AN INTERPOLATION FORMULA
The coeflicients in the contral
difference formulae are smaller and
in Newton's formulae. After a few converge faster than those
terms, the coefficients in the Stirling's formula decrease
pidly than those of the Bessel's formula and the more
rapidly than those of Newton's formula. As coefficients of Bessel's formula decrease more
hould be used much, whenever possible, central difference
in
preference to
formulae. Newton's formulae
The right choice of
interpolation
mterpolated value in the given data.
an
formula however, depends on the
position of the
The following rules will
a
be
1.To find tabulated valuefound useful
near the
2. To finda value near the end of the beginning of the table, use Newton's
table, use Newton's backward forward formula.
3. To find an
or Everett's formula.interpolated value near the centre of the table, use formula.
either Stirling's or Bessel's
If
interpolation is required for p lying between 1/4 and
If
interpolation is
desired for p lying between 1/4 1/4, prefer Stirling's formula.
and 3/4, use Bessel's
Example 28 15. Use Gauss's forward or Everett's
formula.
2317 8144, y29= 171070, y33 16-3432 formula to evaluate y30, given that y21 184708,
=
and y7 =15-5154. =

Sol. Taking *g 29, h

(Rohtak, 2003)
=

=4, we require the value of y forx =

30 i.e. for p=*
The difference table is given below: h
_
J0 Z2 - 0.25

P p
21 2 18 4708
Ayp
06564
25 17 8144
-0-0510
- 0 7074
- 0-0074
29 171070 -00564
-07638 -0-0022
33 163432 - 00640
- 00076
- 0 8278

37 2 155154
FINITE DIFFERENCES AND INTERPOLATION
897
Gauss's forward formula is

12.3 1234
y30 17.1070 + (25)(-0.7638) + 26) (07b) (-0.0564)+25) (C25) (= 06) -0.00761
2 6
1 25) 25) (-075) (-173)
24
g (-00022)
=
17 1070 -019095 +000529 +0-0003 0-00004 169216 approx.
Example 28-16. Interpolate by means of Gauss's backward formula, the population of a town
for the year 1974, given that:
Year 1939 1949 1959 1969 1979 1989
Population (in thousands): 12 15 20 27 39 52
(Madras 2003)
Sol. Taking xo = 1969, h = 10, the population of the town is to be found for
1974 -1969
10 0.5. -

The central difference tableis

p Ayp
1939 -3 12
3

1949 -2 15 2

1959 - 1 20 2 3
- 10
3
1969 0 2Z
12

1979 39 1

13

1989 2 52

Gauss's backward formula is

y1+P*Dpp=)y,+ p +1)(p+ 1)p (p - 1) ,4

p =yo +psy-1+ 2!P 3! 4!

(p+2) (p+ 1)p (p - 1p - 2) 5 +

y3t
5!

(5)
ie. yg=27 +(0-5)(7)+ ) (5) + L5)(5)-5) a1+(2:5) (15)
24

+
(2.5)(15) (5) (-5) (-15)10
120

27+ 3-5 +1875 -0-1875 +0:2743 01172 = 32345 thousands approx.

FINITE DIFFERENCES ANDINTERPOLATION 911
Find the rate of change of
) pressure with respect to volume when v =2, (Madras, 1999; Osmania, 1999)
(i) volume with respect to pressure whenp = 105
11. A rod is rotating in a plane. The following table gives the angle 9 (radians) through which the rod ha
turned for various values of the time t second
02 04 06 08 10 12
8: 0 12 049 112 202 3 467

Calculate the angular velocity and the angular acceleration of the rod, whent = 06 second
(V.T.U., 2004; Burduwan, 2003; Madras, 2003)

12. Ify =f x) and y, denotes f (to + nh), prove that, if powers of h above h° be neglected.
d 3
d 4h O-1-5v2-y.2+45 0-Y.3
13. Using the following data, find x for which y is minimum and find this value ofy.

0-60 0-65 070 075

06221 0-6155 06138 06170
4. From the following table, find the value ofx for which y is maximum and find this value ofy.

12 13 1-4 15 16
09320 0-9636 0-9855 0-9975 0-9996
(Madras, 1998)

2812. NUMERICALINTEGRATION
The process of evaluating a definite integral from a set oftabulated values of the integrand
fa) is called numerical integration. This process when applied to a function of a single variable,
is known as quadrature.

y=fx)

o 2

0 Xth*o+2h
Fig. 281
(1) Newton-Cote's quadrature formula

Let I=f dx
where f(x) takes the values yo, 1, Y2--n for x
=xo, X1, X2,
Let us divide the interval (a, b) into n sub-intervals of width h so that xo = a, x1 = *0t h,

2 0+2h, .., =Xo + nh = b. Then

912 HIGHER ENGINEERING MATHEMATICs
nh
f ) dr (Putx =
x0 + rh, dx =
hdr|
2) 3.
-fg hdr =h, or Ayo + 2! 'yo+ r-1r-
3!
by Newton's forward interpolation formulal
Integrating term by term, we obtain

f d=nh o Avo+ n(2n-3) y+24 y+..

nn-2 yt 12 ...1)
which is known as the Newton-Cote's
from this the following
quadrature formula. Being a general formul>, we deduce
quadrature
rules by taking n =1, 2, 3,.
2 Trapezoidal rule. Putting n =1 in (1) and taking the curve through («o, and
(1.1) as a straight line i.e. Yo)
first become zero, we
a
polynomial of first order so that differences of order higher than
get
+h

rmdepo*~aovo
2h
Similarly fd=a|*v 1+)
*** ****. *°°*°°°°.

o+nh

n-1hf) dr =5 On-1+yn)
Adding these n integrals, we obtain
nh
fx) dx =
+yn) +2(y1 +V2t.
This is known
+Yn -1 .2)
as the trapezoidal rule.
Obs. The areaof each strip
ordinates at x and zo+ nh is (trapezium) is found
separately. Then the area under the curve and
approximately equal to the sum of the
areas of the n the
3) Simpson's one-third trapeziums.
X0 Yo), (x1, y1) and (2, Y2) as a rule". Puttingn
parabola i.e. a
=2 in (1) above and taking the
order higher than second
vanish, we get polynomial of second order so thatcurve through
(o+ 2h
differences o
fdx 2hivo +Ayo =

o)=o+41+y
fo+4h

Similarly..f)dx =2+ 4y3 +y)

1A, f(a) dx =q On-2+ 4n - 1 tyn), nbeingeven.

After the self-taught English mathematician Thomas Stmps0n (1710-1761).

and Newton in 1676 This rule was used by
Gregory in 1668
FINITE DIFFERENCES AND INTERPOLATION 913
Adding all these integrals, we have when n is even

fxdxY% + Yn) + 4(y1 +y3t... +yn - 1) +2{a+ y4t +Yn- ..3)

This is known as the Simpson's one-third rule or simply Simpson's rule and is most commonly
used.

Obs. While applying (3), the given interval must be divided into even number of equal sub-intervals.
since wefind thearea of twostrips at a time.
(4) Simpson's three-eighth rule. Putting n =3 in (1) above and taking the curve through
x yd:i =0, 1, 2,3 as a polynomial of third order so that differences above the third order
vanish, we get
f*o+ 3h

Similarly,
o64
Bh
3h f) da =0s+3y4 +3ys +J6) and so on.

Adding all such expressions from xo to zo + nh, where n is a multiple of 3, we obtain

nh
fx) dx=oo+»)+ 301 +2 +4 +5.. +n-1)+ 203 +Y6 +3n-3) ..4)

which is known as Simpson's three-eighth rule.

Obs. While applying (4), the number of sub-intervals should be taken as a multiple of 3.

all differences above the sixth,

5) Weddle's rule. Putting n =6 in (1) above and neglecting
we obtain
123
fd=6h |yo+34yo +yo+4 yo+ yo+20 Ao 14ayo

Ifwe replacea o by 1 o , the error made will be negligible.

(o+6h
f(x) dx = o+5y1+2 +6y3+y4 +5y5 +Y6)

Similarly
o + 12h
3h
+6y9+ 1 0 + 6y11 t12)
Af(x) dx 16 + 6y7 +8 =

and so on.
Adding all these integrals from xo to zo nh,
+ where n is a multiple of 6, we get
(otnh
3h ....5)

This is known as Weddle's rule.

Obs. While applying (5), the numberof sub-intervals should be taken asa multiple of6
914 HIGHERENGINEERING MATHEMATIcs

Example 2827.Fvaluate
ted by using
) Trapezoidal rule, (i) Simpson's 1/3 rule (Madras, 2003; Andhra, 19991
ii) Simpson's 3/8 rule, (Bhopal, 2002 S
(u) Weddle's rule and compare the result with its actual value. (Rohtak, 2003)
Sol. Divide the interval (0, 6) into six parts each of width h = 1. The values of f(x) =

are given below

1 2 3 4 5
fix 1 0-5 02 01 0-0588 0 0385 0-027
1 y2 Y3 Y4 5
() By Trapezoidal rule,

J d
1+ 2 vo+6)+201 +2 +y3+4+ys)
=
(1 +0-027) + 2(05 +0,2 +0-1+0-0588 + 0-0385)) =
14108.
) By Simpson's 1/3 rule,

Jo o +s) +401 +ys +ys) +202 +yl

(1+0027)+4(05 +0-1+00385)+2(0-2+0-0588)) =1:3662.
iii) By Simpson's 3/8 rule,

Jo 2 0+y6)+ 301 +y2 +y4 +ys) +2y3

(w) By Weddle's rule,

=I1+0027)+3(0.5+0-2+0-0588+0-0385+ 20-1)) =13571.
3h
d=bo+5y1 +y2 *6y3+ y4 +5s+el
1+

=0311+505) +02+6(0
1)+00588+5(0-0385) +0-027) =13735
Also tan 6 = 14056

This shows that thevalue of the integral

value followed by its value given by Simpson'sfound by Weddle's rule
1/3 rule. is the nearest to
the actua
Obs. Applications of Simpson's rule. If the various ordinates in $ 2819 (
cross-sectional areas, then Simpson's rule gives the volume of the sold 3) represent equispaced
useful for civil engineers for calculating the amount of earth that musth c h , Simpson's rule is very
make a dam. Similarlyifthe ordinates denote velocities at equal intervals of t e dtofill a depression or
the distance travelled. The following examples illustrate these applicaton , the Simpson's rule gives