Skip to content

Mathematics · Ch 2 — Basic Algebra

Method of Undetermined Coefficients

2.6.3

Method of Undetermined Coefficients

Given information about a polynomial's zeros and/or its value at specific points, we can construct it by writing it with unknown ('undetermined') coefficients and using the equality test of §2.6 -- match same-power coefficients on both sides of an equation -- to pin those coefficients down.

Two equivalent approaches. To build a quadratic f(x)=ax2+bx+cf(x)=ax^2+bx+c satisfying f(0)=1, f(−2)=0, f(1)=0f(0)=1,\ f(-2)=0,\ f(1)=0: either substitute all three conditions directly and solve the resulting linear system for a,b,ca,b,c; or, since x=−2,1x=-2,1 are already known zeros, write f(x)=d(x+2)(x−1)f(x)=d(x+2)(x-1) for an unknown constant dd and use the remaining condition f(0)=1f(0)=1 to solve for dd -- the zero-factor form is usually the faster route whenever the zeros are already given.

Constructing from a mix of real and irrational zeros. If an irrational zero like 1+31+\sqrt3 is given for a polynomial with rational coefficients, its conjugate 1−31-\sqrt3 must also be a zero (§2.6.1's Note); build the factor [(x−1)2−3][(x-1)^2-3] from the conjugate pair before bringing in any other given zero and an extra condition (like a specified function value) to fix the overall scale constant.

Using the coefficient-matching idea for a divisibility identity. E.g. to prove ap+q=0ap+q=0 given f(x)=x3−3px+2qf(x)=x^3-3px+2q is divisible by g(x)=x2+2ax+a2g(x)=x^2+2ax+a^2: since deg⁡f=3,deg⁡g=2\deg f=3,\deg g=2, the quotient must be linear, f(x)=(x+b)g(x)f(x)=(x+b)g(x); expand and match same-power coefficients on both sides to solve for b,p,qb,p,q in terms of aa, then combine to get the required identity.

Undetermined coefficients also finds closed forms for sums. To find S(n)=1+2+⋯+nS(n)=1+2+\cdots+n: first bound it above by n2n^2 crudely; then, since the growth pattern looks quadratic, POSIT S(n)=a+bn+cn2S(n)=a+bn+cn^2 for unknown constants, use the fact that S(n+1)−S(n)=n+1S(n+1)-S(n)=n+1 to match coefficients and solve for b,cb,c, and use S(1)=1S(1)=1 to solve for aa -- giving the closed form S(n)=n(n+1)2S(n)=\dfrac{n(n+1)}2.

Roots and multiplicity read off from a factored form. For f(x)=(x−1)3(x+1)2(x+5)=0f(x)=(x-1)^3(x+1)^2(x+5)=0, the roots are 11 (multiplicity 3), −1-1 (multiplicity 2), and −5-5 (multiplicity 1, a simple root) -- multiplicities are visible directly from the exponents once ff is fully factored, without any further work. …