Numerical Methods II - Polynomial Approximation
Numerical Methods II - Polynomial Approximation
və Sənaye Universiteti
Numerical
Methods II
Polynomial Approximation
When a complicated function in a practical problem
must be evaluated many times, it is often more
convenient to produce an efficiently computable
approximation.
Continuous Least-Squares
Letting an approximation to from be
is minimized.
According to (see Multivariable Calculus), for to
attain minimum at , it is necessary that
for .
As before, Eqs. describe what are called the .
for which .
Therefore, it is also true that for this vector .
If we define the polynomial
then we obtain
Thus, because
Since
.
Let the least-squares straight line fit be given by
with .
and .
.
For example, for the coefficient matrix is
𝒃
𝐤
𝐢 .∫| 𝒙| ∙ 𝝎 ( 𝒙 ) 𝕕 𝒙 <∞ 𝐟𝐨𝐫 𝐚𝐥𝐥 𝐤 ≥ 𝟎
𝒂
ii. For a continuous function ,
for all .
Otherwise, they are said to be .
. Show that the functions
for
are linearly independent.
for all .
form a for .
for .
Orthogonal Polynomials
The general statement
so that and .
. Two continuous functions and are said to be
orthogonal if .
Then we write
so that
Since , and since forms a basis for polynomials, we
may write
as a generalization of .
with
.
Since , we have
and .
Next we find
with
Hence . ◄
.
Generate the first four members of the sequence of
orthogonal polynomials for the weight function on the
interval and the inner product
with
.
Since
we have
and .
Next we compute
.
Hence
and .
Finally
Hence . ◄
Let be a sequence of polynomials in which is a
polynomial of degree .
where
For example,
• alone will define the least-squares constant
approximation to the given function ;
• and will define the least-squares straight-line
approximation;
• , , and will define the least-squares quadratic
approximation, and so on.
.
Find for the following data , , that will minimize for :
1 2 3 4 5
1 1.2 1.5 1.8 2.1
Letting
Thus we get
.
Notice that has been obtained simply by adding
another term to .
In addition, we have
.
Let the least-squares approximation using Legendre
polynomials, or simply, the Legendre least-squares
polynomials for , be denoted by
where
Thus the minimization of on is equivalent to the
minimization of on .
shows and on .
. Approximation to by on .
Thank you for attention!
text
text
text
.
text
text
text
.
text
text
text
. text
text
text