8 Interpolation With Unequal Intervals ss #1, Introduction In the previous chapters on Interpolation we had the intervals of differencing to be a constant h. In other words, we had x;-x;_;=h constant, for =1,2,...n. If the values of x’s are given at unequal j intervals, our Newton’s forward, backward formulae and central difference interpolation formulae will not hold good. Hence, we introduce anew idea of divided differences. These divided differences take into consideration the changes of the values of the function f(x) and also the changes in the values of the arguments x. 82. Divided differences Let the function y=f(x) assume the values f(x), f(x), ...f(%,) conesponding to the arguments X,x,,...x, respectively where the intervals 2) ~ Xp, X— 1, ...) X,—%q-1 need not be equal. Definitions — The first divided difference of f (x) for the arguments x9, x, is defined £0) ~F) es It is denoted by f (xp, x) oF [Xo X)] or Af(%). In other | et) Nor, eats Fe) -Fe) 5) 5) = Pei «ee | In the Same notation, we have Fx) = AF) = % wo =f, Hy. 5) = AS = OO one ha cy ae lets Sf) -fO) 5%a ‘ 5h Numerical Hiei) With Unequal Intervals ‘The second divided difference of f (x) for three argument , We form the table below Syn ae 4 defined as a td ey Ry fy F(%, %2) =F XX) = Pca F (Xp X12) = an uF (0) %=% j 40 Fy This shows that to find a second divided difference, w, continuous arguments. In the same way, we define the third divided differences the four arguments %p,.x),.x,.x as © Tecuie ty Of f(x) or M169 =F a2) =P AY Fl.) x a0) dinate (), Q), @) refer to divided differences of order one, and three respectively. We will see below the divided difference table. ind ie divided differences of Flx)=x' +242 for the eT. form below the divided difference table. Pests Argument Entry First divided Second divided Third died z meer) a?) a f(a) x Vi@)niiee le difference &f@) e419 22A=32 _ 64 ceil 20=10_ | so8 ae Be i eo 6-3 2246s all * 1344 ~ 224 Me Foy raved tt 224 n & 11+6 Baas Example BaF oe show that Inlike the wy i co % 2 * where r is any positive integer. seen in the alsa f@n-fe) 4% i imag =X Us prove the result by induction. AD, (2)Inequal Intervals 261” Numerical Meth E Anti) = Xx), vided differences 2 alue of any divided difference is independent of uments. That is, the divided differences are | their arguments. LG) =Fl%o) _ fo) (pect \ X-Xp eH =F(x1,%9) ; FG) _ f(s) _ 160 f(x) gain, J _ XX £() pare 4 In the same WAY, J Ha% %—% » X1) =F (%1, 0) fle) \_( £00) X)-X Xy-% X—X) x XX | Soe fs | (x4 — Xp) 1 - yi ea) =x %-% ) iG), iG Go) (%)—%) (1-2) =%9) @-*1) (4)Numerical Meth } Unequal Intervals ae .r.t, any two arguments. Therefore, the divg r.t. any two arguments. " A is linear. are two functions and a, 8 are cong th order divided differences of x" will be a polynomial of vith leading coefficient |. That is, its value is 1 iv =0, for i=1,2, FC) + BE Ce) = fof (2) + Be (5) ie +44] Say A +a, A" x" + +A" a, shee *0 N=), , £04) -8 0) Te x0 =a. 1+0+0+--+0=a, Conversely, if the nth divided difference of a polynomial is constant, the polynomial is of degree n. 84, Relation between divided differences and forward differences If the arguments me ... are equally spaced, then we have, y-meH Xj =%— == sh. AFG) =F05.50)= iy an FOy)=A Fo) Fifa) Al Xx — Xo 2h : ‘Note: ( . Pip Sak Fle) AFG) 3k f ) € Fegas. ee Beeeats |) = Gao! xo + aS : s ; [heorem: Newton’s interpolation formula for unequal intervals =X (or) me Newton’s divided difference formula) = R=2(y. = Xo - eli aM pg sie ry me Jor values Fo) f(x) fOr) corresponding to the (x, X9) en) differences of am * ae = £66.84) =f) *=x, yf (% X04)164 Numericai ee Unequal Intervals “Using this value of f(x, %) in (1), we have 1 (2) =F G0) + = 0) f oy 1) + (X= 0) (x —x,) F(x, x, m oy sot) eet = Lf G6 Xp X1) =F Op, 10 ¥2) + = 20) F(R, X44, 25) Using this value in (2), we get Xo) + C= %) (=X) F @,.x,, x) FH) OX) F(X, %, x) u(u=1) By + wu= 0-2} ayy (6) is exactly Newton's forward interpolation formula that the divided difference of second order can be lent of two determinants of third order. (B.Sc. BR. Ap. 1995) n. We have seen already, eo fe) fey H—*1) H—%) %—%0) -) (—H) (Q--m) (— AFG0) + Co- )F (4) + 1-0) &) ee VG-Da-) A == 41) (1-4) @-%0) aa! | | Fle) Fea) Fd eT po) = Xp) f (Xo, X1 Xp X3) +> Ha) (EX q— 1) F oy Ki Xp ll = =x) FO Xo Ay 4) (x= %)) f & %) E ee can be extended to the nth order divided differences also, w(l)equal Intervals 267 4) 52+ (x- 4) (x5) 15 + (x4) (x-5) (x-7) | +90-30=4 52 + (4) (3) 15 + (4) (3) (1) 1= 448 Za rere 10x 5+ 11x 10x8=3150-~ 2 7 " 5 A F poe gfe) 5) ded difference formula, (Gq x1) + (&— 40) Geom) fm x08) *" 1 yea [-s}ra-ne-ne-” i : +321x—224) ! gis . ¥ % : ev 6x 58+ 1926-224] ). Find the function f(*) from the following rable ' 7 2 4 5 6 : i : i if be of j 6 data are given, We assume the polynomia an = -4) is d f (4) =, itis a clear zee“ DO 4) | ow sa i | y= 0,xy=11,%5=! 5, fo *1 2al Intervals sion With Unequ s 2) es) (2) 4 ince the third difference 9G)= on 4 so ebb tei 25 three times and i Os "7 : Jiu E LO. 8 5 pet at bet. —7 HM" F(a)” 126 Now we will find i ivi TA id (x) using divided difference formula of Ng on 1B2=9a= 194 8 of Neg a q=P ) i soo) F oi dob ej yde2ls-—4 *? : pe I-9 hed oj eokeal?. 1=2 9 o i 58 Now take 5 as the origin at fy =f) += 5) (x *1) By Newton's formula, § (8) = (49) + (4H) Gp Hy) + =A) CH) F (Xo) 3+ (x= 2)9+ (x= 2) (v-5) 1 : ex’ +2r-5 wat (eS) qe (x- 5) Example 11, Find the equation 3 are const = 194498 (x = 5) + 17 (x= 5) 4 (x= 5)’. 1, 21), (1, 15) (2, 12), 169 ants (= 1), we extend the table ntroducing unknowns from the last 228 cam 23 => p = 182 => q=98. nd proceed. 5) f Xo 1%) (x= 5) f toy 1,209) + (x= 5) (e= kt (x= 5) x1 degree and passing ) of least = 0, (3, 3), Find also y at * mn divided ny af (x N ; carta icen = 1) (v4) (x? + 2-5). J irough the points ( fraule dG) a ri adopted to minimise the work. i Solution, Since the intervals are not equal We will for in powers of (x= 5), ¢ following table, obtain f(x) as a polynonll ference table and use Newton's formula i ob ae 3 4 my, e £@) ie) a) hal using ae "s method, 58 mm 4S ; - 2 f : lution, We - 1 i form the divided difference table below: 2 12 a2 =f, 0 y 4 f() & fs) ase) 3 7 “9 \ ; We will ri , 7 \ { a use Newton's divided difference formula 3 i » i ' annie aaa 4 2 I 1 = 984 1D +6 12 54 A, Setting x=0, 5 5 0) 4 =6. 6 i. ge beBel Bs sanible 12. The following table sive’ same M alg eye (= 03706) (y-0-4433) * (0.4068 — 03706) (0.4068 — 0.4433) (y= 03706) (y -0-4068) * (0.4433 — 03706) (0.4433 — 0.4068) (03887 — 0.4068) (03887 0.4439 (03706 — 0.4068) (03706 — 0.4433) (0.3887 — 0.3706) (0:3887- 04433) a for inverse interpolation 4068 = 03706) (0-4068 - 04433) Eo) O— 1991) ; (03887-03706) (03887-04068) ' * (0.4433 — 0.3706) (04433 - 04068) 885832 + 17-202739 — 3.086525 = 22-0020" I. Find the age corresponding to the annuity value le x (23) x25 x (21) x (23) 30 a St eee hs 15: 149 9 (BSc. BR. Nov. #4) 3.6 = 14-1) (13-6 — 133) (136-125) . 39 ) (15-9 — 133) (159 — 12:5) 14-1) (13-6 — 133) (13.65 2) es = 141) (149 - 133) (149- 5 : = 13:3) (13-6= 12°" , 40 14-9) (13:6 — 13:3) 3x 49) (141-133) 4 “ae. 2728 Numerical ey F 136-159) 136 - 14:9)(13.6= 14:1) (136~ 125) 4 033= 159) 033 = 149) (133-141) (13312545 © (136-159) (136 = 149) (136=141) (136-133, + (25 — 15.9) (125 — 149) (12.5 = 14.1) (125 = 133) “50 2 xQ=136)=43. j ; gual Intervals 06 03332 Pie log (304)=2:4829, iog (305) = 2.4843 «i 7 10g (307) en y (x)= 19 given 1 2 1 20 EXERCISE 8.2 (BSc. BR. Nov, 94) Using Lagrange’s interpolation formula, do the problems below: ae 0-737 0738 : 12832974 1.284103. 2g49085 0-740 0-741 0742 1-2873348 1.281461 following table 12 14 16 18 0 1942 1497 110979549 6, u4= 12 find uy, = Be =33 and f(7)=—- 15 find f(2), 100 given 3 5 7 9 it 6 24 58 108 174 30, f (G4) =— 13, f(38)=3 and f(42)=18 find x so that G+ 6) — 03 (y1 + ys) +0-75 2 +9) 33° by using Lagrange’s formula of interpolation given B27 oom 38° 5, 0-6249 0-702 07813 i 4 i = 3 91 = 64, yg=81 and y= 100, find yy 0-48 ‘ A ii osil éb 5 8 05027498 E a a 2 a, fit a polynomial to the data i 1 3 ot i 12 } 1 made on a curve 120 o (KU 1972) ’ acne ) sing 1), (1,3) and G, 55)Numerical 280 Mey se Et ms “a Bu Giciiga|=on0 9. ANSWERS EXERCISE 8.1 Page 270 (abe + bed + acd + abd) eS aie a ax (a+b) ab+berca,_ 1 ay ae Pod: +0 erect? S 6. 31° 5x° + 627 + 1ar4g 7 @-)@-de-9e ++) 8. Joye +s4a-9+T1E-F HEH 9. yex! = 10x +36¢ - 36-5 10. 3-8986688 12, 0777298926 fia) =H (38 + 41x —84) 15, 25099063 ato yates @—+2@-1 +@-1)- EXERCISES 8.2 Page 278 16. 636 47. e+e txt] 1 3515 3. 493 4, 294 5. 1052 7. «147 8. 32:9 9. 101 42, 08932 13. 007575 14. 247 0 17. 129518. 4 19.9 an. 7 + Bipeear+, 24. 0014 * one M. y=" - 1x + 18x-12;4 28. yet 6+ 0