NUMERICAL DIFFERENTIATION
Itic the proceos of caleulating the valuc of the derivative of a function at some assigned value of x from the
given set of values (,9,). To compute dy/dx, we first replace the latter as many Limes as we desire, The choice of
the interpolation formula to be used, will depend on the assigned value of x at which dy/dr, is desired.
If the values of x are equi-spacod and dy/ds, is roquired near tho boginning of the table, we employ
‘Newton's forward formula. If it is required near the end of the table, we use Newton’s backward formula. For
values near the middle of the table, dy/dx, is calculated by means of Stirling's or Bessel’s formula. If the values
of x are not equi-spaced, we use Newton's divided difference formula to represent the function.
[EGE FORMULAE FOR DERIVATIVES _
Consider the function y = /(x) which is tabulated for the values x; (=x) + ih),#=0,1,2,..%.
(2) Derivatives using forward difference formula. Newton's forward interpolation formula (p. 958) is
+ BERD ORD aay,
(n=
Y=Iyt vayyt PILED ay,
Differentiating both sides w.r-t. p, we have
Since p
‘Now
Ate =xyp = 0. Hence putting p = 0,
(2), =f[on fet taty-
980
=A)LINUENEAL DFFERENTATON & INTEGRATION
Again differentiating (1) w.r.t. x, we get
4 (dy) dp
dp (dp) dx
1[2.2. ,6p-6 .9 , 12p?-36p+22 4 1
ie 9 SEES alyg 4 B36 BE ney, 5 |
Putting p = 0, we obtain
(§ 3) ] AB)
a
Similarly ( A)
Otherwise : We know that 1+ A= = ol
Lye, tas
AD =log (1+ 8)= 4-442 1 2a
* 2° '3
and
and
Now applying the above identities toy,, we get
; dy) 1 Lay ylasy Jaty ylas.
Dyy tens (2), M90 — 5 0 + 5 AN — GAN +5 A 0
dy) Afsana (Mat 8 5 | 187 Ao
(2), BLP Mog YF 199 S90
dy) Il, 34 ]
BY) 1 ary, 3 atyy
and (3), rr ae
which are the same as (2), (9) and (4) respectively.
(2) Derivatives using backward difference formula. Newton's backward interpolation formula
(p 958) is
pipe
2!
pins Dipr2)
Y=In t+ PVn +
Differentiating both sides w.r-t. p, we get
(5)
Atx=x,,p = 0. Hence putting p = 0, we get
(ay 1 lye. 1 ys.
ay vy, +4 A y%y,
(2), itp neg Int
Again differentiating (5) w.r-t. x, we have
ay 5 (B)P— [vs Eos Sei t1pti er, |
dx dp\dx)dx at " 12
i
4
1
5
Vivg +
peels ]
vy, +b Vy, be (6)
nt ZV | (6)82
Putting p = 0, we obtain
ay ty yi 4 ey 4187 y
vey, +V' +t > a +5 Sy A)
( es) - ie el 1 +VP 9, ‘In + OV Iy + Vn +
Similarly, ay [Pn +8 +5 Vint | (8)
dx?) “Fe
Otherwise + We know that 1-V= Fo} = 0°
‘ = AD = 10g 1-V)=-|v+iv?+tvstye
AD = log (1-V)= [r+dy tps |
ly t
D=2lvaiv
or ifesdy
dys ytyt,. |
3.4
2
Di= plydr dee |
WL "2 38
Similarly, Di= a shuts -]
ii z
Applying these identities toy, we get
dy 1 Lye pl ys, lye lys. 1 ye,
te, a) Alyy, tye sloty, stvty, eb vty, +by
Pn (2), il +3 PSV Int ZV In EV It GV Int |
d*y 2. By ally, . Sys , 137 ye. )
oe Vy, +¥8y, +2 vty, +2 Vy, +282
( ie al" Ya ag eg Ie ag Vn *
@. 1 3
a (22) -araedene]
s,
which are the same as (6), (7) and (8).
(8) Derivatives using central difference formulae. Stirling's formula (p. 964) is
2) (= +A a
a any aay Ga epee oeand
and
Solution. (a) The difference table is =
= y a 2 a a a 6
10 7.999
ai
11 8.403 0.036
0.378 0.006
12 7a 0.030 0.002
oma 0.004 0.001
13 9.129 ~~ 0.026 = 0.001 0.002
0.322 0.003 0.008
14 aot 0.028 0.002
0.299 0.005
15 9.750 = 0.018
0.28
16 10.031
‘We have
(2 -] 7
(@|
z
ae a7
0.1, x5 = L1, Ay
Substituting these values in (i) and (ii), we get
dx), O01
(3) = Fyos-Pe 0.03) +4 (0,008) - 4 (- 0.001) +=! 410.009)
Aa
dy 1 a 5
( aa owl 0.08 (0.004) » 15 ( 0.001) § coon) = a7
(b) We use the above difference table and the hackward difference operator V instead of A.
dx), h
dx?
Here h = 0.1, x, = 1.6, Vy, = 0.281, V*y,,
(43) - Fahad Vy Vy, + 8 vy, ST
5
Putting these values in (é) and (ti), we get
el
on
-0.716.
1 1 1
aa [oes + 0.018) + 5 (0.005) +
=~ 0.018 ete.
1
4
5
6
1
2) if lyay 41 lye ay 1
B&) =<) vy, +2 Vy, += Vy, += Vy, += V8 yy t= VOY, Hee
(2) ath v yn th yey, eh vty, shvty, + Evy, +]
137
180"
1 1
410.002)+4:0.009)+ 20.002]
n+]
2.75
7 [- 1018 0.005 +1 002) + os + F(0.002]Solution. The difference table is :
t v av
0 °
a
5 3
u
10 “
66
15 9
159
20 228
Hence the initial acceleration is 1 m/sec’.
difference table is :
t x a a
° 30.18
149
on 31.62 -0.24
1.25
oz 32.87 = 0.48
0.77
aa aa ad ~0.48
031
04 93.95 = 0.45
-0.14
vs 33.81 ~ 043
-0.57
8 a
‘As the derivatives are required near the middle of the table, we use Stirling’s formulae :
de
My
1
xy tae) 1
2
(*s 1+ Axy@e) 3 lary Late + Late,
(23), we [tt yah *2 bog 4 es ]
Here h = 0.1, 4, = 0.8, Ax, = 0.31, Ax
Putting these values in (i) and (ii), we get
(2) 1 [essere 1(o0trooe),, 2 (
dt}s OAL 2 2 30
ax 1 [ 1 1
£*) 21) 046-4 000+ 5 0.29)-.
(eI, oy 2 90
Honce the required velocity ic 5.83 em/coe and acceleration is 45.6 em/r0e2,
eae 9 a ne em
- ys
Drag id
Pi
1, find f’ (
Mind
Solution. Taking x,= 7.50, h = 0.1, we have p =
‘The difference table is :
x Pe » 4 a at at a af
TAT 8 0.198
0.002
748 -2 0.195 0.001
0.003 =0.001
1490-1 0.198 0.000 0.000
0.003 0.001 0.008
7.50 ° 0.201 = 0.001 0.003 0.01
‘2.002 0.002 =0.007
751 1 0.208 =0.008
0.003 0.002
152 2 0.206
0.002
758 a 0.208
Bessel's formula (p. 990) 1s
pip=v ayy +aty (Po g)PO? a
2! 2 3 7
(p-)@+ne@-ve-
p7¥o PAYG *
4 P+ DP(p-DD-2) My-24 Ay yg
2
(p42) p(p+D p(p-V(p-2)(p-3)_A*y-s +.4°y. .
+ 6! JT of)
i : , i
Since p= eb
Differentiating (i) w.r-t. p and putting p = 0, we get
dy) _1fay" -H{ al eaty., + A%y) 42 a8y + Lat “
(2), i$), Play Rear rate Bary a+ hatyoeaty
1 1
Tp 92 ay 5489.2]1 1 1
aon [vow = 7 0.001 + 0.001) + + (0.002)
(0.004 + 0.008) — wo 0.007 575 (0.010 0)
chase, An Soe valnen o's re nak sul epaced, we chal was Newtcts dskiad diffcenoeYarmmaia. a
divided difference table is
x fi) Ist div. diff. 2nd div. diff 3rd div. diff. 4th dv. diff
a -18
18
5 22 16
146 0.998
u 899, 39.96 0.0002
1025 Luvs
27 17815 69.04
2613
ey 6606
Fifth difference being zero, Newton's divided difference formula is
FL) = flg) + (x —) fry —1,) + (x — Hq) Ce 2,) fle Xy,%)
44) 44) (=H) [gy Zp Kq) + EM) (HM)
Xr) (9S) Flt, My, Hy Hy %)
Differentiating it w.r.t. x, we get
Fe) = fly, x4) + (20 — 4 — 4) ftp. Hy.)
4 (Bx? — Delay + ay +) + (agey + yey + E—UQN) * fg Ky Tye)
+x? — BaPlag + xy +g 42g) + De (ages + Hphe + Ealy + Tyhy + Hyp +R.)
T Fobra + Retest tata + Mortal fy Ay My Ay Hy)
x g= 11, x= 27 and x = 10, we obtain
8 + 12 x 16 + 23 x 0.998 - 426 x 0.0002 = 232.869.
Putting x, = 3,2,
re
MAXIMA AND MINIMA OF A TABULATED FUNCTION
Newton's forward interpolation formula is
v pi Dip ®»
Y= Iq + Pay + ug Big + apg +
Differentiating it wort. p, we get
dy 3p? —6p+2
eS + Ay + A)
ap Yo
For maxima or minima, dy/dp ~ 0. Henee equating the right hand oide of (1) to 2oro and retaining only
upto third differences, we obtain
: 2
a we 1 gry, 4 8? p42
Yo*
Ay,
. 1 A 1 1
we. (bat) ern-aboee(an-Batn eat)Substituting the values of Ayo, A, A°y, from the difference table, we solve this quadratic for p, Then the
corresponding values of x = x, + ph at which y is maximum or minimum.
Solution. The difference table is :
x y ay aty ay ay ay
2 2
~2.25
“1 0.28 25
025 -0
0 ° -05 6
0.25 3 0
1 0.25 25 6
025 ® °
° 2 M65 6
13.75 6
3 15.5 265,
40.25
4 56
= 0, Ay, ~ - 0.28, A¥y, ~ 2.5, Aly, ~ 9, Aty, ~ 6.
Newton's forward difference formula for the first derivative gives
Ap -18p" + 22p
con a
; (4a! 18 + 224-616)
=t [0.25 + 2.8 — 1.95 +4 Be? Or 4 3423 AG? + 5.5x 1.5] = 29-2
For y to be maximum or minimum, = 20ie, 8-220
ie, x=0,1-1
wy
Now @Y <32-1=—ve forx=0
dx
= tye for=1
1
Sinco yay tntyy+ EOD sty, sn 10)-0
‘Thus y is maximum for x = 0, and maximum value = y (0) = 0.
Also y is minimum for x = 1 and minimum value = y (U)
ve fora =
Eee2) Find tha AeA and eacand derivatives of the finction tabulated below, at the poidt 213 +
Sa RW Ate tae te ae Por tan
yi 0 (0128 0544 1.296. 2432 4.000 (WPT, 2010; Bhopal, 2009)
Given the following table of values of x and y
£6100, 105140 1.20 28 | 80
yd) 1000 1025 1.949 1072 1.095. .nis) 1.440
2,
fina ona 3 at (a) x = 1.05 (6) x = 1.25 (e) x = 1.16, W-T.U., 2008)
ie Diubeitg vliere ofy and, find the first derivative atx = 4.
mM, 1 2 4 8 y 10
yo 1 5 mo
. From the following table, find the values of dy/ds and dy/ds? at x = 2.
ae: 1.96 198 200 ° 202 204
y: 0.7825 0.7739 0.7651. 0.7563 0.7473
Find the value of cos 1.74 from the following table :
% YA: Big PORTS |. 2 ETB 182 1.86
sine: 0.9016. 0.9857 0.9781 0.9601 0.9584
‘The aistance covered by an athlete for the 50 metre is given inthe following table =
Time(se) 0 1 a 3 4 5 45
Distance (metre): 0 «25.85 18H. 90.5 BO
Determine the speed of the athlete at ¢ = 5 sec. correct to two decimals. (U.P-T.U,, 2009)
‘The following data gives corresponding values of pressive and specifie valume of a superheated stream.
oy 2 4 ° a 10
poy (M05) 427.) ea) a82) 118
Find the rate of change of
(@ pressure with respect to volume when v = 2,
i) volume with respect to pressure when p = 105,
‘Tho table below revealo the Yelocity » of a body during the opecifi time &, find ite ncccleration at = 1.1?
t: 10 1a 12 1a 14
2 aaa 47) gad sad en (1N.7.0,, 2009),
The clevation above a datum Tine of7 points of read is given below :
300, 6 1200 1500,
am 205
L.A rod is rotating in a plane. The following table gives the angle @ genianey bry which the rod hes turned for
various values ofthe time ¢ second.
t ° op 04 os 08 10 12
ee to 012049. a2 202 a0 467
Calculate the angular velocity and the angular acceleration af the rnd, when f = 0.6 serand
Find the value of f(x) at x = 0.4 from the following table using Bessel’s formula
x: 00Y 002) 003 0040.05 006
fi): 0.1028 | 0.104707! 0.1096»? ota
Ify = flx) andy, denotes Fosp + nh), prove that, if powers uth above A€ be meylecte
a
@)- Alou yi 1)-F09- yea! ME ocal] (UL. RU 2006)
[Hint: Differentiate Stirling’s formula wir-t. x, and put ~0}
15. Find the value of /"(8) froin the table given below :
ig 6 7 9 12
fia) 1,556 1.690 1.908 2.158 (Anna, 2007)
id the /* (6) from the following data:
2 3 7
26 58 466 (.N2.U., 2009 ; UPTO, 2008)NUMERICAL INTEGRATION
‘The process of evaluating a definite integral from a sct of tabulated values of the integrand f(x) is called
numerical integration. This process when applied to a function of a single variable, is known as quadrature.
‘Tho problem of numerical integration, like that of numerical differentiation, is solved by representing f(x)
by an interpolation formula and then integrating it between the given limits. In this way, we can derive quadra-
ture formula for approximate integration of a function defined by a set of numerical values only.
NEWTON-COTES QUADRATURE FORMULA
Let Is fi fl de 7
where f(x) takes the values.yp,).o5~-J,fOPX= yay,
yy wey (Fig. 30-1)
Let us divide the interval (a, 6) into n sub-in-
tervals of width h so that x9 = a, x =p + A,X, =X) +
2h, x, =4, +h =b. Then
9 tmh
= £ fxde =h [reo +rmar, oath
putting x =x, + rh, dx = hdr
oh (er ayy =D
4 LEmDE-De-BO-4) yy,
4
AY BL
4 Te Dlr ae le Ar -B) aby, |
[By Newton’s forward interpolation formula]
Integrating term by, we obtain
[O° pep ae nif ay eB ng + EM ta 0 2 as.
A"y0
2
nt 8n® 1n® o
* (g-2 ts 6
n° 15n® 4 4n®
+ ( 7 6 Vin’ aA)
‘This is known as Newton-Cotes quadrature formula. From this general formula, we deduce the following
important quadrature rules by taking n = 1, 2,3...
[EGEGN TRAPEZOIDAL RULE
Putting n = 1 in (A) § 30.5 and taking the curve through (x, yp) and (x,, ¥,) as a straight line ie. a
polynomial of first order so that differences of order higher than first become zero, we getfe rede=n (m2 F a0] # Gorrd
oF 2 2
+h 1 h
‘Similarly i reyax=n(y+Lan)—4o,+99
10) de =F 94)
in
ag +n
Adding these n integrals, we obtain
oa
£ P(x) de = to +Y,) +20, +99 tm Iq DI
‘This is known as the trapezium rule.
SIMPSON’S ONE-THIRD RULE
Putting n = 2 in (A) above and taking the curve through (xq, ¥9); (x1, 91) and (xg) yg) aa a parabola i.
polynomial of second order so that differences of order higher than second vanish, we get
Jo pede =2h 00 + Ave + ard 6+ AY, +99)
ah
Similarly, eo fle) dx =
0244549) hen
i +n h A
JOO flerde Boyt tig typ bing even,
sina
Adding all these integrals, we have (when n is even)
oh
Jer re de = A199 #9,) +404 +99) tn ABO HI HTT
‘This is known as the Simpson's one-third rule or simply Simpson's rule and is most commorly used.
SIMPSON’S THREE-EIGHTH RULE
Putting n = 3 in (A) above and taking the curve through (x,,.y
80 that differences above the third order vanish, we get
fo" reoar=an( + Sane + ats +2 avo)
a Zrorgr rg
), 1,2, 3 as a polynomial of third order
= ho + 87,481 +¥0)
‘larly,
ny th 3h
[EIT £0) as =F 05 + 944 85 +99) and 50 0n.
Adding all such expressions from x, to x) + nk, where n is a multiple of 3, we obtain
oh
g Fea) d= BM iyg 49) 430, +I HI HIGH 3, 1) 42005 4.954 ty a)
ASN as ee
which is known a5 Simpson's three eighth rule.[EGER boote’s RULE
‘Putting n = 4 in (A) above and neglecting all differences above the fourth, we obtain
ain ee Ban 2p Ta
fe reas Ah( sp +2ay Sarno +3.'vo +7 a's)
Be (ayy + By, + 12), + 82, + 7)
Similarly
[Ore ae - 2 (iy + 82y5+ 12y¢-+ 92); + Ty) and goon,
Adding all those integrals frou 2p 40:49 + nh, where 1 io a multiple of, we get
Joo ree ae = 2 cay, +829, + 1p, + Bg + Ly, + 8294+ 12Y¢t BBY, + Hy +o
‘This is known as Boole’s rule.
WEDDLE’S RULE
Pratingn = nA) abv od neleing ni dierenes bone thew, ween
[roar (v0 #3009 + $9 +400 #2 ty 0 FH gsy, 42, How)
we replace AY a%y, by Alyy the error made will be negligible.
a les Peay dte= M95 + By, +92 + Ory +H, + Bg +36)
Similarly
[oe renee Fey, + Oy HY + Wa +n + Oy, +p) and 60 on,
ea) where n is a multiple of 6, we get
en Pla) dic = 3 (yg + By, +95 + By #94 + 595 + 25+ 5b +4 +)
‘This is known as Weddle’s rule.
Solution. Divide the interval (0, 6) into six parts each of width h = 1. The values of flx) = we are given,
below() By ‘Trapezoidal rule,
1 A
GRe bloc rx)+ 20, +9: 495 +H +99
[C+ 0.027) + 2(0.5 + 0.2 + 0.1 + 0.0588 + 0.0385)] = 1.4108.
(i) By Simpson's 1/9 rule,
1 h
Phas blo ty +40 195490 +20 490)
1
= lO +0097) 4 400.5 4.01 + 0.0385) + 9(0.9 + 0.O8RR)) = 1 RHA?
Gi) By Simpson's 3/8 rule,
LA Boo t98 801 ttt rs)
(2 + 0.027) + 8(0.5 + 0.2 + 0.0588 + 0.0885) + 2(0.1)] = 1.9571.
(iv) By Weddle’s rule,
fs
Bh 4 Sy, 0+ Ore 494595 49
= 0.811 + 5(0.5) + 0.2 + 6(0.1) + 0.0588 + 5(0.0885) + 0.027] = 1.8785.
tan af = 1.4056
Solution, Let y = eh = 0.2 and n = 10.
The values of x and y are as follows :
Ry Trapezoidal rule, we have
+ A
[pee Flo ryt 201 1 tg tI AIS HE tI HIE HID
[G+ 54.5981) + 2(1.0408 + 1.1735 + 1.4333 + 1.8964
+2118 + 4.2206 + /.UYYS + 12.988 + 29.085)]
2
Hence fe" dx = 17.0821
Solution. Divide the interval (0, 0.6) into six parts each of width h = 0.1. ‘the values of y = f(x) =
given below :By Simpson's 1/3rd rule, we have
3
f de = Eig +35) +404 +95-495+ Hy, +01
Solution, Let y = sin x —log, x + e and h = 0.2,n = 6.
The values of y are as given below :
By Simpson's 2 th rule, we have
We ear
fade Flute + 209) +30, S949, +99)
= fo. 2) (7.7336 + 2(3.1660) + 3 (13.3247)] - 4.053
Hence f)*tsin x= log, x +") de = 4.058.
Solution. If km be the distance covered in ¢ (min), then =
| f = [Pode= 20x + 40 + 2, by Simpson's rue
=
5 ete.O=v, +0540, +0; + vy= 10 +25 + 324114280
Emu + uj det Ug= 18 + 29 +2015 =72
fo _2
== 44x 8042x792)
leo 3
= 309.33 km.
Solution. It secbe the time taken to traverse a distances (m) then a7
or Hx Ly (say)
Seay zy (sav),
-20
Hore ae Caan
1 1 doa ad
Also cE ee ne
ne dinedinah WY = qprdo = aged THI =y
c+ by Boole’s Rules, we have
[e- g ‘yds = 2A BE yg + 32, # Ly + 8p, + Lay, + 82y5 + Lye + By, + My)
an s5)*2*(z1)°22(z5) °(a5)-™(z2)
+ wo )+22( 4) +14(2)]
Solution. Here h = 0.25, yq = 1,9, = 0.9896, y = 0.9589, ete.
+. Required volume of the solid generated
= flip? dean: Bigd + 9+ 407 +99)+ 2981
= 0.25 F{(1 + (0.8415) + 4((0.9896)* + (0.9089)*) + 2(0.0589)"
= 025%S 1416 (3.7081 + 7.2216 + 1.699] = 0.2618 (10.7687) = 2.8192.INUMERICAL DiFrEReNTATION & INTEGRATION
CEP y
de "
1, Bvaluate [5S applying
(rapenical le NFL 300) Sng ete
Gi) Spt 8 le
de
a, Brats [1-55 sing taeda re abingh = 4
(ii) Simpson’s V/rd rule taking h = 1/4. (N-T.U, 2008)
(ii) Simpson's 3/8th rule taking h = 1/6. W.P.U, 2010; V-T.U,, 2007)
Ww) Wedale’s rule taking f = 1/6. (Bhopal, 2009)
Hence compute an approximate value of xin each ease.
S
5. Find an approximate value flog, 9 by calculating wo 4 decimal places, by Simpson's 16 rule, f° 2%, ivaing
the range into 10 equal parts.
6
4, Bvaluate [sec