Scribd
Upload
86 views5 pages

Newton's Divided Differences Formula

The document discusses Newton's divided differences formula for polynomial interpolation. It provides the formula for representing the Lagrange interpolation polynomial Pn(x) using Newton's divided differences. The coefficients a0, a1, etc. are found by substituting values into Pn(x) and setting it equal to the function values. These coefficients are called Newton's divided differences of order 0, 1, etc. Examples are provided to demonstrate computing divided differences and the Newton interpolation polynomial.

Uploaded by

Aqe Kitam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
86 views5 pages

Newton's Divided Differences Formula

The document discusses Newton's divided differences formula for polynomial interpolation. It provides the formula for representing the Lagrange interpolation polynomial Pn(x) using Newton's divided differences. The coefficients a0, a1, etc. are found by substituting values into Pn(x) and setting it equal to the function values. These coefficients are called Newton's divided differences of order 0, 1, etc. Examples are provided to demonstrate computing divided differences and the Newton interpolation polynomial.

Uploaded by

Aqe Kitam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

3.

3 Divided Differences
Suppose that 𝑷𝒏 (𝒙) is the Lagrange polynomial that
interpolates a continuous function 𝒇(𝒙) at {𝒙𝟎 , 𝒙𝟏 , 𝒙𝟐 , … , 𝒙𝒏 }.
Then 𝑷𝒏 (𝒙) can be represented in the following form with
fewer computations than Lagrange:
𝑷𝒏 (𝒙) = 𝒂𝟎 + 𝒂𝟏 (𝒙 − 𝒙𝟎 ) + 𝒂𝟐 (𝒙 − 𝒙𝟎 )(𝒙 − 𝒙𝟏 ) + …
+𝒂𝒏 (𝒙 − 𝒙𝟎 )(𝒙 − 𝒙𝟏 ) … (𝒙 − 𝒙𝒏−𝟏 )
It is called Newton’s divided differences formula
To find {𝒂𝟎 , 𝒂𝟏 , … , 𝒂𝒏 }
Substituting 𝒙 = 𝒙𝟎
→ 𝒇(𝒙𝟎 ) = 𝑷𝒏 (𝒙𝟎 ) = 𝒂𝟎 → 𝒂𝟎 = 𝒇(𝒙𝟎 ) ≡ 𝒇𝟎
Substituting 𝒙 = 𝒙𝟏
→ 𝒇(𝒙𝟏 ) = 𝑷𝒏 (𝒙𝟏 ) = 𝒂𝟎 + 𝒂𝟏 (𝒙𝟏 − 𝒙𝟎 )
𝒇(𝒙𝟏 ) − 𝒇(𝒙𝟎 )
→ 𝒂𝟏 = ≡ 𝒇𝟎𝟏
𝒙𝟏 − 𝒙𝟎
Substituting 𝒙 = 𝒙𝟐
→ 𝒇(𝒙𝟐 ) = 𝒂𝟎 + 𝒂𝟏 (𝒙𝟐 − 𝒙𝟎 ) + 𝒂𝟐 (𝒙𝟐 − 𝒙𝟎 )(𝒙𝟐 − 𝒙𝟏 )
𝒇𝟏𝟐 − 𝒇𝟎𝟏
→ 𝒂𝟐 = ≡ 𝒇𝟎𝟏𝟐
𝒙𝟐 − 𝒙𝟎
𝒇𝟎 , 𝒇𝟎𝟏 , 𝒇𝟎𝟏𝟐 are called Newton’s divided differences of
order zero, one and two respectively.

1
Newton’s divided differences formula
𝑷𝒏 (𝒙) = 𝒇𝟎 + 𝒇𝟎𝟏 (𝒙 − 𝒙𝟎 ) + 𝒇𝟎𝟏𝟐 (𝒙 − 𝒙𝟎 )(𝒙 − 𝒙𝟏 ) + …
+𝒇𝟎𝟏…𝒏 (𝒙 − 𝒙𝟎 )(𝒙 − 𝒙𝟏 ) … (𝒙 − 𝒙𝒏−𝟏 )
Note that this formula is equivalent to Lagrange
interpolation formula but with different form, thus the error
estimate is the same as for Lagrange:
𝒏
𝒇(𝒏+𝟏) (𝒛)
𝑬𝒏 (𝒙) ≅ �|𝒙 − 𝒙𝒊 |
(𝒏 + 𝟏)!
𝒊=𝟎

𝒘𝒉𝒆𝒓𝒆 𝒛 𝒊𝒔𝒂 𝒏𝒖𝒎𝒃𝒆𝒓 𝒘𝒊𝒕𝒉𝒊𝒏 𝒙𝟎 , 𝒙𝟏 , … , 𝒙𝒏 𝒂𝒏𝒅 𝒙

Divided Differences Table

𝒙𝒊 𝒇𝒊 𝒇𝒊,𝒊+𝟏 𝒇𝒊,𝒊+𝟏,𝒊+𝟐 𝒇𝒊,𝒊+𝟏,𝒊+𝟐,𝒊+𝟑


𝒙𝟎 𝒇𝟎
𝒙𝟏 𝒇𝟏 𝒇𝟏 − 𝒇𝟎
𝒇𝟎𝟏 =
𝒙𝟏 − 𝒙𝟎
𝒙𝟐 𝒇𝟐 𝒇𝟐 − 𝒇𝟏 𝒇𝟏𝟐 − 𝒇𝟎𝟏
𝒇𝟏𝟐 = 𝒇𝟎𝟏𝟐 =
𝒙𝟐 − 𝒙𝟏 𝒙𝟐 − 𝒙𝟎
𝒙𝟑 𝒇𝟑 𝒇𝟑 − 𝒇𝟐 𝒇𝟐𝟑 − 𝒇𝟏𝟐 𝒇𝟎𝟏𝟐𝟑
𝒇𝟐𝟑 = 𝒇𝟏𝟐𝟑 =
𝒙𝟑 − 𝒙𝟐 𝒙𝟑 − 𝒙𝟏

2
Example 1 Find the Newton’s divided difference formula
𝟏
that interpolate 𝒇(𝒙) = at {𝒙𝟎 = 𝟏, 𝒙𝟏 = 𝟐, 𝒙𝟐 = 𝟒}.
𝒙
Compute 𝑷𝟐 (𝟎. 𝟓), 𝑷𝟐 (𝟑) and 𝑷𝟐 (𝟓. 𝟓) . Find the exact error
and comment on the results.
Find 𝑷𝟑 (𝒙) if 𝒙𝟑 = 𝟓 and compute 𝑷𝟑 (𝟓. 𝟓).

3
Newton’s divided differences Algorithm
Input: 𝒏; 𝒙𝒊 , 𝒇𝒊 , 𝒊 = 𝟎. . 𝒏
To compute the divided differences:
Set 𝑸[𝒊, 𝟎]: = 𝒇𝒊 for 𝒊 = 𝟎. . 𝒏
𝑸[𝒊,𝒋−𝟏]−𝑸[𝒊+𝟏,𝒋−𝟏]
Calculate 𝑸[𝒊, 𝒋]: =
𝒙𝒊 −𝒙𝒊−𝒋
,

for 𝒋 = 𝟏. . 𝒊; 𝒊 = 𝟏. . 𝒏

To find Newton’s polynomial 𝑷𝒏 (𝒙):


Set 𝑷[𝟎] = 𝑸[𝟎, 𝟎] ∶ 𝒄𝒐𝒆𝒇[𝟎] = 𝟏
For 𝒊 from 1 to 𝒏 do
𝒄𝒐𝒆𝒇[𝒊]: = 𝒄𝒐𝒆𝒇[𝒊 − 𝟏] ∗ (𝒙 − 𝒙[𝒊 − 𝟏])
𝑷[𝒊] = 𝑷[𝒊 − 𝟏] + 𝑸[𝒊, 𝒊] ∗ 𝒄𝒐𝒆𝒇[𝒊]
end do:

Note: For any polynomial of degree 𝒏 the Newton’s divided


differences of order 𝒏 are constant.

4
Example 3: Complete the following divided difference
table and find the Newton’s polynomial of maximum
order:
𝒙 𝟎 = 𝟎 𝒇𝟎 =
𝒙𝟏 =. 𝟒 𝒇𝟏 = 𝒇𝟎𝟏 =
𝒙𝟐 =. 𝟕 𝒇𝟐 = 𝟔 𝒇𝟏𝟐 = 𝟏𝟎 𝒇𝟎𝟏𝟐 = 𝟓𝟎/𝟕

Example 4: The following data are given for a polynomial


𝑷(𝒙) of unknown degree. Determine the polynomial with
maximum degree if all third order divided differences are
1.
𝒙 𝑷(𝒙)
𝟎 𝟐
𝟏 −𝟏
𝟐 𝟒

You might also like