Scribd
Upload
39 views6 pages

Cubic Approximation

The cubic approximation to a function f(x) at a point x=a is a cubic polynomial f3(x) that matches the value, first derivative, second derivative, and third derivative of f(x) at x=a. The example shows using Taylor's formula to find the cubic approximation of the function f(x)=1/(1-x) at x=0, and calculates the maximum error of this approximation when |x|<0.1.

Uploaded by

Anis Yulia
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
39 views6 pages

Cubic Approximation

The cubic approximation to a function f(x) at a point x=a is a cubic polynomial f3(x) that matches the value, first derivative, second derivative, and third derivative of f(x) at x=a. The example shows using Taylor's formula to find the cubic approximation of the function f(x)=1/(1-x) at x=0, and calculates the maximum error of this approximation when |x|<0.1.

Uploaded by

Anis Yulia
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 6

CUBIC APPROXIMATION x=a

The cubic approximation to f at x =a is a cubic, f3 (x), which has the same value,
derivative, second derivative and third derivative as f at x = a:

EXAMPLE
Taylor Series cubic approximation
Using Taylor's formula with a=0 and n=3, how would you find the cubic approximation of f(x)= 1/(1-x)
at x=0, given the upper bound for the magnitude of the error in the approximation when lxl< 0.1 ?

The derivatives are

The Taylor polynomial

and the Taylor

remainder

where c is between 0 and

we are trying to determine the maximum of


domain
maximum error would be

and so this would occur at

. Since

then

on the
and the

1. CODING NATURAL CUBIC SPLINE


%Natural cubic spline based on Cubic Hermite Formula
clear
clf reset
x=[.25 1 1.5 2 2.4 5]
y=[23.1 1.68 1 .84 .83 .26]
n=length(x);
for i=1:n-1,
h(i)=x(i+1)-x(i); delta(i)= (y(i+1)-y(i))/h(i);
end
%Construct the Tridiagonal System
hp=h(2:n-1); hm=h(1:n-2);
deltap=delta(2:n-1); deltam=delta(1:n-2);
U=[1,hm]; L=[hp,1]; D=2*[1,hp+hm,1];
B=3*[delta(1), hp.*deltam+hm.*deltap,delta(n-1)];
dy = tridiag(L,D,U,B);
%this is just the Cubic Hermite Graphing Code
t=[];
u=[];
for i=1:n-1,
r=h(i)/10; a=delta(i);
b(i)=(a-dy(i))/h(i); c=(dy(i+1)-a)/h(i); d(i)=(c-b(i))/h(i);
p=x(i):r:x(i+1);
s=y(i)+(p-x(i)).*(dy(i)+(p-x(i)).*(b(i)+(p-x(i+1)).*d(i)));
t=[t, p];
u=[u, s];
end
plot(t,u,'b',x,y,'*r')
grid on

2. NATURAL CUBIC SPLINES

% find out coefficent b


s=size(x,1);
j=zeros(s,1);
m=zeros(s,1);
n=zeros(s,1);
p=zeros(s,1);
q=zeros(s,1);
for i=1:s
j(i,1)=i-1;
if i==1
p(i,1)=0;
q(i,1)=0;
m(i,1)=x(i+1,1)-x(i,1);
n(i,1)=y(i+1,1)-y(i,1);
elseif i==s
m(i,1)=0;
n(i,1)=0;
p(i,1)=0;
q(i,1)=0;
else
m(i,1)=x(i+1,1)-x(i,1);
n(i,1)=y(i+1,1)-y(i,1);
p(i,1)=2*(m(i-1,1)+m(i,1));
q(i,1)=3*(n(i,1)/m(i,1)-(n(i-1,1)/m(i-1,1)));
end
end
A=[j,x,y,m,n,p,q];
A
P=zeros(s-2,s-2);
q1=zeros(s-2,1);
for i=1:s-2
for j=1:s-2
if i==j
P(i,j)=p(i+1,1);
elseif abs(i-j)==1
P(i,j)=m(i,1);
else
P(i,j)=0;
end
end
q1(i,1)=q(i+1,1);
end
b1=zeros(s-2,1);
a=zeros(s,1);
b=zeros(s,1);
c=zeros(s,1);
b1=P\q1;
for i=1:s-2
b(i+1,1)=b1(i,1);
end

for i=1:s
if i==s
a(i,1)=0;
c(i,1)=0;
else
a(i,1)=n(i,1)/m(i,1)-m(i,1)*(b(i+1,1)+2*b(i,1))/3;
c(i,1)=(b(i+1,1)-b(i))/(3*m(i,1));
end
end
B=[a,b,c];
B;
le=round((x(length(x),1)-x(1,1))*100);
Mat=zeros(le,1);
YTM=zeros(le,1);

% maturity
% yield

t=1;
for i=1:s
if i==s
else
k=round(x(i+1,1)*100)-round(x(i,1)*100);
%
%
%
%
%
%
%

if k==1
Mat(t,1)=x(i,1);
YTM(t,1)=y(i,1);
Mat(t+1,1)=x(i+1,1);
YTM(t+1,1)=y(i+1,1);
t=t+1;
else
for r=0:(k)
Mat(t+r,1)=x(i,1)+r/100;
if r==0
YTM(t+r,1)=y(i,1);
else

YTM(t+r,1)=y(i,1)+a(i,1)*(r/100)+b(i,1)*(r/100)^2+c(i,1)*(r/100)^3;
end
end
t=t+k;
% end
% YTM(x(length(x),1),1)=y(length(x),1);
end
end
[Mat,YTM];
plot(Mat,YTM,'-',x,y,'ro',dataT(:,3),dataT(:,2),'o');

%zy=zeros(x(length(x),1),1);
%fy=zeros(x(length(x),1),1);
%'zero coupon yield curve (zy)'

%for i=1:s
%
if coupon(i,1)==0
%
zy(i,1)=y(i,1);
% else
% end
%end
%for i=1:C(x(length(x),1))
% if zy(i,1)==0
%
d=0;
%
for j=1:i-1
%
d=d+C1(i,1)/((1+C1(j,1)/100)^(j));
% end
%
%zy(i,1)=(((100+C1(i,1))/(100-d))^(1/i)-1)*100;
%else
% end
%end
%for i=1:C(x(length(x),1))
%

if i==length(C)
%
fy(length(C),1)=zy(length(C),1);
% else
%
fy(i,1)=(((1+zy(i+1,1)/100)^2)/(1+zy(i,1)/100)-1)*100;
%end
%end

%[C,C1,zy,fy]
%plot(C,zy,C,C1,C,fy);

You might also like