D2 - Transforming polynomial roots

D2 - Transforming polynomial roots

Move every root of a polynomial by the same scaling and translation, and you can find the new equation without solving the old one. Learn to reverse the root transformation, substitute safely and check the result using symmetric sums.

Linear changes of roots

If α,β,γ\alpha,\beta,\gamma solve x3−4x2+x+6=0x^3-4x^2+x+6=0, what equation has roots 2α+1,2β+1,2γ+12\alpha+1,2\beta+1,2\gamma+1? We will transform the equation itself, rather than first finding the three roots.

Start with one root: if α=−1\alpha=-1, its new value is 2(−1)+1=−12(-1)+1=-1. If β=2\beta=2, its new value is 55. The same rule acts on each root occurrence.

Affine root transformation

A transformation u=px+qu=px+q, with constants p≠0p\ne0 and qq, scales each root by pp, then adds qq. Its inverse is x=(u−q)/px=(u-q)/p. Distinct old roots give distinct new roots, and multiplicities are preserved.

Why must the substitution be the inverse? The original polynomial expects an old value xx. Given a proposed new root uu, undo the transformation to recover that old value.

Worked example 1: mapping roots and reversing the map

Question. The roots of a cubic are −1,2,3-1,2,3. Find the transformed roots under u=2x+1u=2x+1, and express xx in terms of uu.

Plan. Apply the same rule to each root, then solve the transformation for the old variable.

−1↦−1,2↦5,3↦7u=2x+1⟹u−1=2x⟹x=u−12\begin{aligned} -1&\mapsto -1,&2&\mapsto5,&3&\mapsto7\\ u=2x+1&\Longrightarrow u-1=2x \Longrightarrow x=\frac{u-1}{2} \end{aligned}

Answer. The roots become −1,5,7-1,5,7; the inverse substitution is x=(u−1)/2x=(u-1)/2.

Check. Undo the map on 77: (7−1)/2=3(7-1)/2=3, the root from which it came.

Change only the sign of the scale factor. A negative scale reverses the order of real roots, but it remains invertible.

Worked example 2: a negative scale

Question. Use the same old root list −1,2,3-1,2,3, now with u=−2x+1u=-2x+1. Find the new roots and the inverse substitution.

Plan. Multiply by −2-2 before adding 11, then isolate xx.

−1↦3,2↦−3,3↦−5u−1=−2x⟹x=1−u2\begin{aligned} -1&\mapsto3,&2&\mapsto-3,&3&\mapsto-5\\ u-1&=-2x \Longrightarrow x=\frac{1-u}{2} \end{aligned}

Answer. The roots are 3,−3,−53,-3,-5, and x=(1−u)/2x=(1-u)/2.

Check. The new root −5-5 maps back to (1+5)/2=3(1+5)/2=3; applying the forward map again gives −5-5.

If p=0p=0, every value maps to qq, so no inverse exists. One could construct (u−q)n=0(u-q)^n=0 to list nn transformed root occurrences, but the inverse-substitution method below requires p≠0p\ne0.

Watch out

Wrong: to obtain roots 2α+12\alpha+1, substitute x=2u+1x=2u+1. That is the forward map in the wrong direction. Solve u=2x+1u=2x+1 first: substitute x=(u−1)/2x=(u-1)/2.

Practise reversing a new transformation.

Forming the transformed polynomial

If f(x)=0f(x)=0 has degree nn and leading coefficient a≠0a\ne0, replace xx by (u−q)/p(u-q)/p. Multiplying by pnp^n clears the powers of the scale factor from the denominators.

Inverse substitution

For p≠0p\ne0, an equation whose roots are pαi+qp\alpha_i+q is

pnf ⁣(u−qp)=0p^n f\!\left(\frac{u-q}{p}\right)=0

Its degree is still nn, and its leading coefficient is aa. A further nonzero constant multiple gives an equivalent equation. This method is not supplied in the formula booklet: learn the inverse substitution.

Completeness matters. If f(x)=a∏i=1n(x−αi)f(x)=a\prod_{i=1}^{n}(x-\alpha_i), then

pnf ⁣(u−qp)=a∏i=1n(u−(pαi+q))p^n f\!\left(\frac{u-q}{p}\right) =a\prod_{i=1}^{n}\bigl(u-(p\alpha_i+q)\bigr)

These are exactly the required nn root occurrences, with no extra roots and no lost repeated roots. The proof works over the complex numbers too.

Worked example 3: scale and translate a cubic

Question. Form an equation with integer coefficients whose roots are 2α+1,2β+1,2γ+12\alpha+1,2\beta+1,2\gamma+1, where the old roots solve x3−4x2+x+6=0x^3-4x^2+x+6=0.

Plan. Substitute x=(u−1)/2x=(u-1)/2 and multiply the whole equation by 88.

(u−12)3−4(u−12)2+u−12+6=0(u−1)3−8(u−1)2+4(u−1)+48=0u3−3u2+3u−1−8u2+16u−8+4u−4+48=0\begin{aligned} \left(\frac{u-1}{2}\right)^3 -4\left(\frac{u-1}{2}\right)^2+\frac{u-1}{2}+6&=0\\ (u-1)^3-8(u-1)^2+4(u-1)+48&=0\\ u^3-3u^2+3u-1-8u^2+16u-8+4u-4+48&=0 \end{aligned}

The constant term comes from four contributions: −1−8−4+48=35-1-8-4+48=35.

Answer. u3−11u2+23u+35=0u^3-11u^2+23u+35=0.

Check. The old polynomial factors as (x+1)(x−2)(x−3)(x+1)(x-2)(x-3). Transforming those roots gives −1,5,7-1,5,7, and (u+1)(u−5)(u−7)(u+1)(u-5)(u-7) expands to the answer.

Change only the translation, from +1+1 to −1-1. The substitution now contains u+1u+1, which changes the collected coefficients.

Worked example 4: changing the translation

Question. Form an equation with integer coefficients whose roots are 2α−1,2β−1,2γ−12\alpha-1,2\beta-1,2\gamma-1, for the same cubic.

Plan. Use x=(u+1)/2x=(u+1)/2, then clear denominators before expanding.

(u+1)3−8(u+1)2+4(u+1)+48=0u3+3u2+3u+1−8u2−16u−8+4u+4+48=0u3−5u2−9u+45=0\begin{aligned} (u+1)^3-8(u+1)^2+4(u+1)+48&=0\\ u^3+3u^2+3u+1-8u^2-16u-8+4u+4+48&=0\\ u^3-5u^2-9u+45&=0 \end{aligned}

Answer. u3−5u2−9u+45=0u^3-5u^2-9u+45=0.

Check. The new roots are −3,3,5-3,3,5. Their factor product (u+3)(u−3)(u−5)=(u2−9)(u−5)(u+3)(u-3)(u-5)=(u^2-9)(u-5) gives exactly this cubic.

For a quartic, multiply by p4p^4; do not use p3p^3 just because the worked examples were cubic. If integer coefficients are requested and fractions remain, multiply the entire equation by a suitable nonzero integer.

Watch out

Wrong: multiplying only the leading term by 88. Every term of the equation must be multiplied, including the old constant. Finish with =0=0: an expression alone is not the requested equation.

Apply the cubic method with a fresh polynomial.

Checking with roots and coefficients

Symmetric sums provide an alternative route and a quick independent check. For three roots, a common translation appears three times in their sum, but its effect on the pairwise sum and product needs careful expansion.

Let the old cubic have e1=α+β+γe_1=\alpha+\beta+\gamma, e2=αβ+αγ+βγe_2=\alpha\beta+\alpha\gamma+\beta\gamma, e3=αβγe_3=\alpha\beta\gamma. For new roots pα+q,pβ+q,pγ+qp\alpha+q,p\beta+q,p\gamma+q, expansion gives

E1=pe1+3qE2=p2e2+2pqe1+3q2E3=p3e3+p2qe2+pq2e1+q3\begin{aligned} E_1&=pe_1+3q\\ E_2&=p^2e_2+2pqe_1+3q^2\\ E_3&=p^3e_3+p^2qe_2+pq^2e_1+q^3 \end{aligned}

For example, each old root appears in two of the three pair products, explaining 2pqe12pqe_1. These are derived expressions, not extra formula-booklet entries. The monic new cubic is u3−E1u2+E2u−E3=0u^3-E_1u^2+E_2u-E_3=0.

Worked example 5: a pure scaling by symmetric sums

Question. The roots of x3−2x2−5x+6=0x^3-2x^2-5x+6=0 are scaled by 22. Find their equation using coefficient relations.

Plan. Read the old sums, scale each product according to its number of factors, then rebuild the cubic.

(e1,e2,e3)=(2,−5,−6)(E1,E2,E3)=(2e1,4e2,8e3)=(4,−20,−48)\begin{aligned} (e_1,e_2,e_3)&=(2,-5,-6)\\ (E_1,E_2,E_3)&=(2e_1,4e_2,8e_3)=(4,-20,-48) \end{aligned}

Answer. u3−4u2−20u+48=0u^3-4u^2-20u+48=0.

Check. Inverse substitution gives 8f(u/2)=u3−4u2−20u+488f(u/2)=u^3-4u^2-20u+48, independently confirming all coefficients.

Keep the scaling and add a translation of 11. The mixed terms now matter; this is why substitution is often shorter for cubic and quartic questions.

Worked example 6: adding a translation by symmetric sums

Question. For the same old cubic, find an equation whose roots are 2α+1,2β+1,2γ+12\alpha+1,2\beta+1,2\gamma+1, using coefficient relations.

Plan. Use p=2,q=1p=2,q=1 in the three derived sums.

E1=2(2)+3=7E2=4(−5)+4(2)+3=−9E3=8(−6)+4(−5)+2(2)+1=−63\begin{aligned} E_1&=2(2)+3=7\\ E_2&=4(-5)+4(2)+3=-9\\ E_3&=8(-6)+4(-5)+2(2)+1=-63 \end{aligned}

Answer. u3−7u2−9u+63=0u^3-7u^2-9u+63=0.

Check. The inverse-substitution expression is (u−1)3−4(u−1)2−20(u−1)+48(u-1)^3-4(u-1)^2-20(u-1)+48. Expanding gives the same cubic, including constant −1−4+20+48=63-1-4+20+48=63.

For a quartic, the sum check is E1=pe1+4qE_1=pe_1+4q; the 44 counts four root occurrences. Full expansion of all quartic symmetric sums is possible, but inverse substitution avoids managing many mixed terms. Choose the shorter route, then use a sum or product as a check.

Watch out

Wrong for a translation: E3=p3e3+q3E_3=p^3e_3+q^3. The product also contains mixed terms. Expand it completely, or use inverse substitution. When using symmetric sums, group expressions as p(α+β+γ)+3qp(\alpha+\beta+\gamma)+3q so the coefficient relation can actually be used.

Check a proposed equation by two symmetric sums.

Quick Recap

  • Reverse u=px+qu=px+q to x=(u−q)/px=(u-q)/p, with p≠0p\ne0.
  • Substitute the inverse into the old degree-nn equation and multiply by pnp^n.
  • The factor identity proves that degree, completeness and multiplicities are preserved.
  • Symmetric sums offer a second route and useful coefficient checks; translations create mixed products.

The opening equation is u3−11u2+23u+35=0u^3-11u^2+23u+35=0, whose roots are exactly 2α+1,2β+1,2γ+12\alpha+1,2\beta+1,2\gamma+1.