lecture notes on polynomial interpolation


1 Summary of notation and terminonlogy

2 The polynomial interpolation problem

Definition 1.

Let 𝐱n be given. Define ev𝐱:𝒫mn, the multi-evaluation mapping, to be the linear transformation given by

ev𝐱:p[p(x1)p(xn)],p𝒫m.
Problem 2 (Polynomial interpolation).

Given (x1,y1),,(xn,yn)2 to find a polynomial p𝒫m such that p(xi)=yi,i=1,,n. Equivalently, given 𝐱,𝐲n to find the preimageMathworldPlanetmath ev𝐱-1(𝐲).

Theorem 3.

Fix xRn. The multi-evaluation mapping evx:Pn-1Rn is an isomorphismPlanetmathPlanetmathPlanetmathPlanetmath if and only if the componentsPlanetmathPlanetmath x1,,xn are distinct.

3 Vandermonde matrix, polynomial, and determinant

Definition 4.

For a given 𝐱n, the following matrix

𝖵𝖬(𝐱)=𝖵𝖬(x1,,xn)=[1x1x12x1n-11x2x22x2n-11x3x32x3n-11xnxn2xnn-1]

is called the Vandermonde matrix. The expression,

𝖵(𝐱)=𝖵(x1,,xn)=1i<jn(xj-xi)

is called the Vandermonde polynomial. Note: we define 𝖵(x1)=1; the usual convention is that the empty product is equal to 1.

Proposition 5.

The Vandermonde matrix is the transformation matrix of evx with the monomial basis [1,x,,xn] as the input basis and the standard basis [e1,,en+1] as the output basis.

Theorem 6.

The Vandermonde polynomial gives the determinantMathworldPlanetmath of the Vandermonde matrix:

|1x1x1n-11x2x2n-11xnxnn-1|=1i<jn+1(xj-xi), (1)

or more succinctly,

det𝖵𝖬(𝐱)=𝖵(𝐱).
Proof.

We will prove formulaMathworldPlanetmathPlanetmath (1) by inductionMathworldPlanetmath on n. Note that

𝖵𝖬(x1)=[1],𝖵(x1)=1.

Evidently then, the formula works for n=1. Next, suppose that we believe the formula for a given n. We show that the formula is valid for n+1. For x1,,xn and a variable x, and consider the nth degree polynomial

p(x)=det𝖵𝖬(x1,,xn,x)=|1x1x1n-1x1n1x2x2n-1x2n1xnxnn-1xnn1xxn-1xn|

By the properties of determinants, x1,,xn are roots of p(x). Taking the cofactor expansion along the bottom row, we see that the coefficient of xn is 𝖵(x1,,xn). Therefore,

p(x)=𝖵(x1,,xn)(x-x1)(x-xn)=𝖵(x1,,xn,x),

as was to be shown. ∎

4 Lagrange interpolation formula

Let x1,,xn be distinct. We know that ev𝐱:𝒫n-1n is invertiblePlanetmathPlanetmathPlanetmath. Let pol𝐱:n𝒫n-1 denote the inversePlanetmathPlanetmathPlanetmath. In principle, this inverse is described by the inverse of the Vandermonde matrix. Is there another way to solve the interpolation problem? For i=1,,n let us define the polynomial

pi(x)=jix-xjxi-xj.

These n-1 degree polynomials have been “engineered” so that pi(xi)=1 and so that pi(xj)=0 for ij

Theorem 7 (Lagrange interpolation formula).

Let x1,,xnR be distinct. Then,

pol𝐱(𝐲)=y1p1++ynpn.

5 Underdetermined and overdetermined interpolation

Definition 8.

Let T:𝒰𝒱 be a linear transformation of finite dimensional vector spaces. A linear problem is an equation of the form

T(𝐮)=𝐯,

where 𝐯𝒱 is given, and 𝐮𝒰 is the unknown. To be more precise, the problem is to determine the preimage T-1(𝐯) for a given 𝐯𝒱. We say that the problem is overdetermined if dim𝒱>dim𝒰, i.e. if there are more equations than unknowns. The linear problem is said to be underdetermined if dim𝒱<dim𝒰, i.e., if there are more variables than equations.

Remark 9.

By the rank-nullity theoremMathworldPlanetmath, an underdetermined linear problem will either be inconsistent, or will have multiple solutions. Thus, an underdetermined system arises when T is not one-to-one; i.e., the kernel ker(T) is non-trivial. An overdetermined linear system is inconsistent, unless 𝐯 satisfies a number linear compatibility constraints, equations that describe the image Im(T). To put it another way, an overdetermined system arises when T is not onto.

Definition 10.

Let x1,,xn be distinct and let ev𝐱:𝒫mn be the corresponding multi-evaluation mapping. Let 𝐲n be given. The linear equation

ev𝐱(p)=𝐲,p𝒫m

is called an underdetermined interpolation problem if mn. If mn-2, we call the above equation an overdetermined interpolation problem. Note that if m=n-1, the interpolation problem is “just right”; there is exactly one solution, namely the polynomial pol𝐱(𝐲) given by the Lagrange interpolation formula.

Proposition 11 (Underdetermined interpolation).

Let x1,,xnR be distinct. Define

q(x)=(x-x1)(x-xn)

to be the n𝑡ℎ degree polynomial with the xi as its roots. Suppose that mn. Then, the multi-evaluation mapping evx:PmRn is not one-to-one. A basis for the kernel is given by q(x),xq(x),,xm-nq(x). Let y1,,ynR be given. The solution set to the interpolation problem evx(p)=y, is given by

ev𝐱-1(𝐲)={r+sq:s𝒫m-n},

where r=polx(y) is the n-1𝑠𝑡 degree polynomial given by the Lagrange interpolation formula. To put it another way, the general solution of the equation

ev𝐱(p)=𝐲

is given by

p=r+sq,s𝒫n-m free.
Proposition 12 (Overdetermined interpolation).

Let x1,,xnR be distinct. Suppose that mn-2. Then, the multi-evaluation mapping evx:PmRn is not onto. If m=n-2, then yR4 belongs to Im(evx) if and only if

|1x1x1my11x2x2my21xn-1xn-1myn-11xnxnmyn|=0

More generally, for any mn-2, the interpolation problem evx(p)=y,yRn has a solution if and only if rankM(x,y)=m+1 where

M(𝐱,𝐲)=[1x1x1my11x2x2my21xm+1xm+1mym+11xnxnmyn]
Remark 13.

The compatibility constraints on y1,,yn for an overdetermined system amount to the condition that 𝐲 lie in the image of ev𝐱, or what is equivalentMathworldPlanetmathPlanetmathPlanetmathPlanetmath, belong to the column space of the corresponding transformation matrix. We can therefore obtain the constraint equations by row reducing the augmented matrix M(𝐱,𝐲) to echelon formMathworldPlanetmath. The back entries of the last n-m-1 rows will hold the constraint equations. Equivalently, we can solve the interpolation problem for (x1,y1),,(xm,ym) to obtain a p=y1p1+ym+1pm+1𝒫m. The additional equations

yi=y1p1(xi)++ym+1pm+1(xi),i=m+2,,n

are the desired compatibility constraints on y1,,yn.

Title lecture notes on polynomial interpolation
Canonical name LectureNotesOnPolynomialInterpolation
Date of creation 2013-03-22 16:51:59
Last modified on 2013-03-22 16:51:59
Owner rmilson (146)
Last modified by rmilson (146)
Numerical id 7
Author rmilson (146)
Entry type Topic
Classification msc 41A05
Related topic EvaluationHomomorphism
Related topic LagrangeInterpolationFormula
Related topic VanderMondeMatrix
Defines multi-evaluation mapping
Defines interpolation mapping
Defines polynomial interpolation