Nowtoo os poet ine sta
sos Me
Sovetigyat racy ato fea
se eal as LOS ss his A
won tabs ot terres
Newshs
warts ane often bran by
thot nanos
34 Dwvided Ditteronces 125
Suppose that Pc) is the ath,
the distinet numbers vy ietane polynomial that agrees with the function fat
inlpebrnie tera Ms Ais oo es Although this polynomial is unique, there are alternate
hte enc atons that are usetul in certain situations. The divided differences of f
Nutlerespect 10.49.44. ty ae used to eapress Py ta) inthe form
PAY = a9 and = a0) + ast = wl — ay) 4+ Hayy =A age GS)
{ex anpwoptiaty constants dy. ..adg, To determine the ist of these constants, ep, Mote
that 2) ts wat inthe form of Eq. (LS), then evaluating P¢4) at go leaves only the
omstant tena gs thats,
Hy = Pao) = F000)
Similarly, when P(4) is evaluated at), the only nonzero terms in the evaluation of
Pou.) are the constant and linea terms,
FA) + ayy = Wy) = Pn) = FOE
fou) = foo)
une
36)
We now introduce the divided-ditference notation, which is related 10 Aithen's A?
notation used in ‘The servth divided difference of the function f with respect
We
to va dlonoted tla} is simply the valne of
Phul = fay). a7
The remaining divide dite the first divided difference
of f with respect (04, and 1,41 isdenoted [4 1,.;] an defined
we defined recursiv
Sosa co)
‘The second divided diglerence, fits. peti 2] is defi
SUirtetieal = Fltintiend
PA ArT 7
Similarly, after the (& — Dt divided differences,
FW rigotigneccetenal and f[uy tages Maatsigele
have been determined, the Ath divided difference relative to Xi.tje164i436---s4ie4 iS
Atkin l= flietints
tive
Site) G9)
SU Nite tied atta
‘The process ends with the single nth divided difference,
Sim
okal = Slt.
Bay
PAI Ad = act]
Because of Fy, 3.6) we can write ay = F401} jst as ay can be expressed as ay
£9) = flap). Hence the interpolating polynomial in Eg, (3.5) i
PaAX) = fro) + flvo.e1 10 = x9) + a(x = x0) = xy)
pee baan(e = oe 01)
Spee corte am rate a an nee pt se hn dpe cant ae a be Da nd Chr
aged mv el nyse aves nan comatose Cee ogee
enon ston nunca marcoCHAPTER 3. Interpolation and Polynomial Approximation
“ ‘equired constants are
uation of ap and ay. the Fea
‘As might be expected from the evalu:
otis
aac = fleet tae ek sted Newton's Divide
for each k = 0.1 So P,(x) can be rewritten in a form ¢
for each & = O,1,....1 So Pelt
Difference: ain
Pax) = fl00) + SO Fle ye] 20) <0 AED
i as
Je Xty en hy
The value of /[p.x),....24] is independent of the order of the numbers 40,
shown in Exercise 21 and one
The generation of the divided differences is outlined in Table 3.9. Two fourth
fifth difference can also be determined from these data,
Table 3.9
Third
Fast ‘Second :
divided diterences Aivided dferences divided ditferences
fea) Leda fted
‘ avs : [Link].45) = Leal = fos]
2] = fhe, Rade (1.42.6) — fl
Poon = ASI fs) Mesa] = ese Lhe
& sles] - [Link]) = LUI Shin)
* Stel - Flrsexy. ny] = LUBED Sls)
Pye Taf Ht gn ee fli
She) “Peyxaty] = Le Slant)
Pc} = Leal= fd
% fle)
Newton's Divided-Difference Formula
{obtain the divided-difference coefficients of the interpolatory polynomial P onthe (mt)
distinct numbers x9,.1,...,%q for the function f:
INPUT numbers.x0, x1... Values (40), fi)... fq) as [Link],-
OUTPUT the numbers Foo, Fit... Faq where
Fag.
Pa) = Foot DFT] — 3). isis flrosxis..-.m))
0
Step1 Fori = 1,2,
Forj = 1,
id—1 — Fiangut
Set Fy = EN tay. xl)
Step2 OUTPUT (Foo, Fi,
STOP. ”Exampl
Table 310 met
: fa)
10 07651977
13 06200860
16 0.4554022
19 02818186
22 0.1103623
Table 3.11
33° Divided Differences 127
The form of the our
differences 'put in Algorithm 3.2 can be
+8 shown in Example? modified to produce all the divided
Complete the divided al
Saleem led difference table for the data used in Example 1 of Section 3.2, and
Able 3.10, and construct the intxpolting polynomial that uses al this dacs
Solution The frst divided difference involving xo and x, is
‘Flo.] = LEU = Fre) __ 0.620860 ~ 0.765197 a
a0 13-10
04837057,
‘The remaining frst divided differences ae found ina similar manner and are shown in the
fourth column in Table 3.11,
fal
ls
0 07651977
04837057
1 13 06200860 =0.1087339
05489460 o.06ss784
2 16 04554022 0.044433 0.0018251
05786120 00680685,
3 19 o2si8is6 oorsis3
05715210
4 0.1103623
‘The second divided difference involving xo, x1, and x) is
[Link] = [Link]} _ -0.5489460 — (—0.4837057)
Sas 16-10
The remaining second divided differences are shown in the Sth column of Table 3.11.
The third divided difference involving xy, 11,.12, and.x3 and the fourth divided difference
involving all the data points are, respectively,
[u1,%2.%3] — fLx0.x1.%3 —0.0494433 — (—0.1087339)
{snastn RR Res) OOH)
= 0.0658784,
Fl%.x133) = = -0.1087339.
and
Fler x2,43,44] — flxo.x1,42,23] _ 0.0680685 — 0.0658784
Pio 3.44] = — aa
= 0.0018251.
All the entries are given in Table 3.11.
‘The coefficients of the Newton forward divided-difference form of the interpolating
polynomial are along the diagonal in the table. This polynomial is
‘Pa(x) = 0.7651977 — 0.4837057(x — 1.0) ~ 0.1087339(x ~ 1.0)(x - 1.3)
+.0.0658784(x — 1.0)(x — 1.3)(x ~ 1.6)
+.0.0018251(x — 1.0)(x — 1.3)(x = 1.6) 1.9).
Notice that the value P,(1.5) = 0.5118200 agrees with the result in Table 3.6 for Example
2 of Section 3.2, as it must because the polynomials are the same, .
id 8 Lei ime yi ed al nlite Da ei, sep bp om ke th
van ely tpt ona nara orowsnenaepane peng noneoeRvlbind eer ap ine acne pear“Grempes Us-y Nece