Вы находитесь на странице: 1из 5

Central difference interpolation

newton's forward and backward interpolation formula are applicable for interpolation near the
beginning and the end of the tabulated arguments respectively.now in this lesson we discuss
interpolation near the center of tabulated arguments.for the purpose we use central difference
interpolation formula.Guass norward,Bessel's formula,laplace Euertt's formula are some of the various
central difference interpolation formulas.

Let us consif=der some equidistant arguments with interval of difference.say g and corresponding
function values are given.let xn be the central point among the arguments

For interpolation at the point x near the central value

Let f(xo)=yo,f(xo-h)=y-1,f(xo+h)=y1,

F(xo-2h)=y-2,f(x0+2h)=y2,

F(xo-3h)=y-3,f(xo+3h)=y3 and soon.

Lagrange interpolation
Let the values of the function f:[a,b]< 𝑅 → 𝑅be given at x1,x2,…..xn ∈ [a,b].choose the
function g(x)from the th set of the polynomials of degree at most (n-1)

{𝑔(𝐴, 𝑥) = 𝐴1 + 𝐴2𝑥 + 𝐴3𝑋 2 +An𝑥 𝑛−1 :A∈ 𝑅 𝑛 }

Which satisfies the condition :

G(xi)=f(xi),1,2,…..,n.

In other words, we would like to determine the coefficients A1,A2,…An such that the chosen
function g(x)to be fitted to the table.

X …. X1 X2 …. Xn
F(x) …. F(x1) F(x2) …. F(xn)
That is, the graph of g(x)to be gone through the given points ;
The simplest idea for determining coefficients is to solve the system of the linear equations.

A1+A2xi+A3𝑥𝑖 2 +…….+An𝑥 𝑛−1 =f(xi),i=1,2,3…n.

But we can obtain the result from the so-called Lagrange formula more easily. for this reason
introduce the functions
(𝑥−𝑥1)(𝑥−𝑥2)…(𝑥−𝑥𝑖−1)(𝑥−𝑥𝑖+1)…(𝑥−𝑥𝑛)
Li(x)=(𝑥𝑖−𝑥1)(𝑥𝑖−𝑥2)…(𝑥𝑖−𝑥𝑖−1)(𝑥𝑖+𝑥𝑖−1)…(𝑥𝑖−𝑥𝑛),i=1,2,…n

Spline interpolation

All Lagrange interpolation ist has been mentioned that the piecewise use of polynomials of low
degree instead of the so-called global interpolation is often advantageous.generaliging
this thought by requiring the smoothness (the existence of the derivatives of some
order)at the "connecting points”. we can obtain of the spline function.

Definition

Let the portion of the interval [a,b]be given by the points

a=x1< 𝑥2 < 𝑥3 <….xn=b

the function s:[a,b]< 𝑅 → 𝑅 is called a spline function of degree m and order k (with respect to
the given partion)if it is polynomial of degree at most m and it is continuously
defferntiable at the connecting points to the order k.

interpolation by linear spline order 0

l let the values of the function f:[a,b]< 𝑅 → 𝑅be given at a=x1< 𝑥2 < 𝑥3 < ⋯ 𝑥𝑛=b
and choosen that function (x)from the set (go'(x))of linear spines of order 0 which
satisfies the condition.

G(xi)=f(xi),i=1,2…n
1
In other words, we would like to obtain a linear saline of order 0 such that its graph to be
fitted to the given points .this piecewise linear function g(x)can be given separately for
each subinterval. here we give a "global formula "introduce the functions.

Which can be given by(multiple) formulas easy, too. the piecewise linear continuous function
li(x),j…..n takes the value 0 at every point differing xj and the value 1 at xj .hence there
function g(x)can be given by the linear combination of the so-called basis function lj(x)
the concerning functions values f(xj)

g(x)=f(x1)l1(x)+f(x2)l2(x)+…..+f(xn)ln(x).

*interpolation by cubic spline of order 1

Let the function table

X X1 X2 …. Xn
F(x) F(x1) F(x2) …. F(xn)
f'(x) f'(x1) f'(x2) …. f'(xn)
Be given. choose that function g(x) from the {𝑔13 (𝑥)}of the cubic splines of order 1 which
satisfies the condition:

G(xi)=f(xi),g1(x1)=f1(xi) i=1,2……n

This piecewise cubic function g(x) can be given separately for every subinterval easily by

Applying cubic hermit interpolating formulas for two-two appropriate points a “global
formulas”

Can be given by the following piecewise polynomial of degree at most 3

The represented functions can be given by (multiple) formulas easily enough if we use up their

Characteristic properties. these properties are as follows :

The function uj(x),j=1,2…..n has the value lat xj and the value 0 at every point furthermore its

Derivatives has the value 0 at each given point the function uj(x) takes the value 0at each given
point furthermore its first derivative is 1 at xj and 0 at other given points hence the
function g(x) can be given be the by the linear combination of the basic functions
uj(x),uj(x) in the form

2
g(x)=f(x1)u1(x)+……..+f(xn)un (x)+f1(x1)vu(x)+……+f1(xn)vu(x)

Interpolation by cubic spine of order2

it the number of the given point in n then we it the number of the given point in n then we
obtain a concrete cubic spline of order 2 after suitable determination of 4(n-1)
coefficients. Since the order of the spline is two therefore the cubic polynomials and
their first and second derivatives have the same value at the connecting points. So we
have 3(n-2) condition and the number of the ”free coefficient” is 4(n-1)-3(n-2)=n+2

So we can prescribe two extra conditions besides prescribing function values. mostly it is
required that the second derivative be 0 at the end points. After this define the
interpolating problem. Let the function table

X A=x1 X2 …… b=xn
F(x) f’(x1) f(x2) …... f(xn)
Be given. choose the function g(n) from the set:

{g(x): g(x)E(g32(x) , gn(a)=0 , gn(b)=0}

Of the cubic spline of second order which satisfies the conditions

g(xi)=f(xi) , i=1,2,3…,n

Introducing the notation

g1(x)=A11+A12x+A13x2+A14x3 ,if x𝜖[x1,x2)

g(x)= g2(x)=A21+A22x+A23x2+A24x3 ,if x𝜖[x2,x3)

gn-1(x)=An-1,1+An-1,2x+An-1,3x2+An-1,4x3, if x∈ [xn-1,xn]

we obtain the following equations for determining the coefficients prescription for function

values:

gi(xi)=fi

gi(xi+1)=fi+1,i=1,2………n-1

Equality of derivatives at connecting points:

gi’(xi+1)=gi’+1(xi+1)

3
gi”(xi+1)=gi”+1(xi+1) i=1,2,……n-1

Boundary conditions:

g’1(x1)=0

g”n-1(xn)=0

It can be proved that linear system of the 4n-4 equations has a unique solution for the

Coefficents.

Formulation of the problem

The problems to be saved are as follows:

Y’(x)=f(x,y(x)), y(xo)=yO;

Where x∈[xo,x0+T],y(x)=[y1(x)=[y1(x),y2(x),……..yn(x)]:R->Rn

Yi(x)∈cp+1([x0,xo+T]),i=1……n for a given p and f(x,y(x))=[f1(x,y(X)),f2(x,y(x))],…..fn(X,Y(X))]T

Let us introduce the following notations:

Y(x)=[x,y1(x),…….,yn(x)]T.f(y(x))=[1,f1(y(x)],………fn(y(x))]T

By this notation the equation (1) is as follows:

Y(x)=f(y(x)), y(x0)=[x0,y1(xo),y2(x0)……..yn(x0)]T

Assume that the functions fi(x,y(x)),i=1….n are (p+1) times continuously differentiable functions

with respect x.then the method of taylor series is based on the following taylor’s series

expansion of the solution using our compact notation the taylor series of the solution of (2) is :
1
y(xo+3)=y(xo)+y’(xo)3+2!1 y’’(x0)s2+3!1 y’’(xo)s3+…..+(𝑝+1)!y(p+1)(x0+Qs)

where Q could be different for the components and the values of p might be chosen depending

on the smoothness of the right n and side in (1) and the desired order of the method to be

obtained.

Оценить