Complex Roots of Polynomials

A2 · P3 · 17 min

A polynomial with real coefficients can have non-real roots, but they never come alone: if a+bia + bi is a root then so is a−bia - bi. This one fact is what the syllabus means by "use the result that, for a polynomial equation with real coefficients, any non-real roots occur in conjugate pairs". In the exam it turns up as a cubic or quartic where you are given one complex root (or asked to verify it) and must find all the others, sometimes with unknown coefficients to determine first. The skills are the ones from the algebra unit (division and the factor theorem) combined with complex arithmetic.

The conjugate root theorem

Look back at a real quadratic with negative discriminant: its roots are −b2a±4ac−b22ai-\dfrac{b}{2a} \pm \dfrac{\sqrt{4ac - b^2}}{2a}i, a conjugate pair. That was not a coincidence of the quadratic formula. It holds for polynomials of every degree.

Conjugate pairs

If p(z)p(z) is a polynomial with real coefficients and α\alpha is a root, then α∗\alpha^* is also a root.

So non-real roots of real polynomials occur in conjugate pairs, and each pair a±bia \pm bi gives a real quadratic factor

(z−(a+bi))(z−(a−bi))=z2−2az+(a2+b2).(z - (a + bi))(z - (a - bi)) = z^2 - 2az + (a^2 + b^2).
Proof

Let p(z)=cnzn+⋯+c1z+c0p(z) = c_nz^n + \cdots + c_1z + c_0 with every ckc_k real, and suppose p(α)=0p(\alpha) = 0. Take the conjugate of both sides. The conjugate of a sum is the sum of conjugates, the conjugate of a product is the product of conjugates, so (αk)∗=(α∗)k(\alpha^k)^* = (\alpha^*)^k, and ck∗=ckc_k^* = c_k because ckc_k is real. Hence

0=0∗=(p(α))∗=cn(α∗)n+⋯+c1α∗+c0=p(α∗).0 = 0^* = \left(p(\alpha)\right)^* = c_n(\alpha^*)^n + \cdots + c_1\alpha^* + c_0 = p(\alpha^*).

So α∗\alpha^* is a root.

The proof shows exactly where "real coefficients" is used: in ck∗=ckc_k^* = c_k. If any coefficient is complex, the result fails.

Consequences for cubics and quartics

A polynomial of degree nn has exactly nn roots, counted with repetition (you may use this fact; it is not proved at A Level). Combined with conjugate pairs:

Degree (real coefficients)Possible numbers of real rootsPossible numbers of non-real roots
quadratic22 or 0000 or 22
cubic33 or 1100 or 22
quartic44, 22 or 0000, 22 or 44

In particular every real cubic has at least one real root, because the non-real roots pair off and three is odd. You can see this on a graph: a cubic y=p(x)y = p(x) goes from −∞-\infty to +∞+\infty (or the reverse), so it must cross the xx-axis.

y = x^3 - 5x^2 + 9x - 5 (1, 0)

The curve y=x3−5x2+9x−5y = x^3 - 5x^2 + 9x - 5 crosses the xx-axis only once, at x=1x = 1. Its other two roots, 2±i2 \pm i, are non-real and do not appear on the graph.

The sum and product shortcut

For the conjugate pair α=a+bi\alpha = a + bi and α∗=a−bi\alpha^* = a - bi:

  • the sum is α+α∗=2a\alpha + \alpha^* = 2a, which is 2Re⁡α2\operatorname{Re}\alpha;
  • the product is αα∗=a2+b2\alpha\alpha^* = a^2 + b^2.

So the quadratic factor is z2−(sum)z+(product)z^2 - (\text{sum})z + (\text{product}). For 2+i2 + i: sum 44, product 55, factor z2−4z+5z^2 - 4z + 5. Do this in your head and write it down; there is no need to expand (z−2−i)(z−2+i)(z - 2 - i)(z - 2 + i) each time, though it is a good check.

The three standard question types

Type 1: one complex root given, all coefficients known
  1. Write down the conjugate, stating that the coefficients are real.
  2. Form the real quadratic factor z2−2az+(a2+b2)z^2 - 2az + (a^2 + b^2).
  3. Divide the polynomial by it (long division, or compare coefficients in p(z)=(quadratic)(other factor)p(z) = (\text{quadratic})(\text{other factor})).
  4. Solve the other factor for the remaining roots.
Type 2: one complex root given, coefficients unknown

Either substitute the root, expand the powers, and equate real and imaginary parts to zero (two equations for two unknowns);

or form the quadratic factor and compare coefficients in p(z)=(quadratic)(other factor with unknowns)p(z) = (\text{quadratic})(\text{other factor with unknowns}).

Then finish as in Type 1.

Type 3: real root findable, complex roots wanted
  1. Use the factor theorem to find a real root rr (try ±1,±2,…\pm 1, \pm 2, \ldots, divisors of the constant term over divisors of the leading coefficient).
  2. Divide by (z−r)(z - r), or by (kz−m)(kz - m) if the root is mk\tfrac{m}{k}, to get a quadratic.
  3. Solve the quadratic, using ii when the discriminant is negative.

Comparing coefficients is usually quicker than long division for a cubic. For example, if z3−5z2+9z−5=(z2−4z+5)(z+k)z^3 - 5z^2 + 9z - 5 = (z^2 - 4z + 5)(z + k), the constant terms give 5k=−55k = -5 immediately.

Worked examples

Type 1: a cubic with one complex root given

Given that 2+i2 + i is a root of z3−5z2+9z−5=0z^3 - 5z^2 + 9z - 5 = 0, find the other two roots.

Solution

The coefficients are real, so 2−i2 - i is also a root. The sum of this pair is 44 and the product is 4+1=54 + 1 = 5, so z2−4z+5z^2 - 4z + 5 is a factor.

Write z3−5z2+9z−5=(z2−4z+5)(z+k)z^3 - 5z^2 + 9z - 5 = (z^2 - 4z + 5)(z + k).

Constant terms: 5k=−55k = -5, so k=−1k = -1.

Check another coefficient, z2z^2: k−4=−5k - 4 = -5. Correct.

The third factor is z−1z - 1, so the roots are 2+i2 + i, 2−i2 - i and 11.

Type 3: find the real root first

Solve the equation z3−4z2+14z−20=0z^3 - 4z^2 + 14z - 20 = 0.

Solution

Try small integers: p(2)=8−16+28−20=0p(2) = 8 - 16 + 28 - 20 = 0, so (z−2)(z - 2) is a factor.

Divide (by comparing coefficients): z3−4z2+14z−20=(z−2)(z2+bz+10)z^3 - 4z^2 + 14z - 20 = (z - 2)(z^2 + bz + 10), where the constant 1010 comes from −2×10=−20-2 \times 10 = -20. The z2z^2 coefficient gives b−2=−4b - 2 = -4, so b=−2b = -2. Check zz: 10−2b=1410 - 2b = 14.

Solve z2−2z+10=0z^2 - 2z + 10 = 0:

z=2±4−402=2±6i2=1±3i.z = \frac{2 \pm \sqrt{4 - 40}}{2} = \frac{2 \pm 6i}{2} = 1 \pm 3i.

The roots are 22, 1+3i1 + 3i and 1−3i1 - 3i.

Type 2: unknown coefficients by substitution

The equation z3+az2+bz−10=0z^3 + az^2 + bz - 10 = 0, where aa and bb are real, has a root 1+2i1 + 2i.

(a) Find aa and bb.

(b) Find the other two roots.

Solution

(a) Work out the powers:

z2=(1+2i)2=1+4i+4i2=−3+4i,z^2 = (1 + 2i)^2 = 1 + 4i + 4i^2 = -3 + 4i,z3=(−3+4i)(1+2i)=−3−6i+4i+8i2=−11−2i.z^3 = (-3 + 4i)(1 + 2i) = -3 - 6i + 4i + 8i^2 = -11 - 2i.

Substitute into the equation:

(−11−2i)+a(−3+4i)+b(1+2i)−10=0.(-11 - 2i) + a(-3 + 4i) + b(1 + 2i) - 10 = 0.

Real parts: −21−3a+b=0-21 - 3a + b = 0. Imaginary parts: −2+4a+2b=0-2 + 4a + 2b = 0, i.e. b=1−2ab = 1 - 2a.

Substitute: −21−3a+1−2a=0-21 - 3a + 1 - 2a = 0, so −5a=20-5a = 20, a=−4a = -4 and b=9b = 9.

(b) The equation is z3−4z2+9z−10=0z^3 - 4z^2 + 9z - 10 = 0. The coefficients are real, so 1−2i1 - 2i is a root and z2−2z+5z^2 - 2z + 5 is a factor. Then z3−4z2+9z−10=(z2−2z+5)(z−2)z^3 - 4z^2 + 9z - 10 = (z^2 - 2z + 5)(z - 2) (constant terms: 5×(−2)=−105 \times (-2) = -10).

The other roots are 1−2i1 - 2i and 22.

Type 2: a quartic by comparing coefficients

The equation z4+az3+bz2−8z+20=0z^4 + az^3 + bz^2 - 8z + 20 = 0, with aa and bb real, has a root 1−2i1 - 2i. Find aa and bb, and solve the equation.

Solution

The conjugate 1+2i1 + 2i is also a root. Sum 22, product 55: the factor is z2−2z+5z^2 - 2z + 5.

Write

z4+az3+bz2−8z+20=(z2−2z+5)(z2+pz+q).z^4 + az^3 + bz^2 - 8z + 20 = (z^2 - 2z + 5)(z^2 + pz + q).

Compare coefficients, starting with the ones that involve only known numbers:

  • constant: 5q=205q = 20, so q=4q = 4;
  • zz: 5p−2q=−85p - 2q = -8, so 5p=05p = 0 and p=0p = 0;
  • z3z^3: a=p−2=−2a = p - 2 = -2;
  • z2z^2: b=q−2p+5=9b = q - 2p + 5 = 9.

The other factor is z2+4z^2 + 4, so z=±2iz = \pm 2i.

a=−2a = -2, b=9b = 9, and the roots are 1+2i1 + 2i, 1−2i1 - 2i, 2i2i, −2i-2i.

Building a polynomial from its roots

Find the cubic equation with real coefficients and leading coefficient 11 that has roots 33 and −1+i2-1 + i\sqrt{2}.

Solution

Real coefficients means −1−i2-1 - i\sqrt{2} is also a root. That pair has sum −2-2 and product (−1)2+(2)2=3(-1)^2 + (\sqrt{2})^2 = 3, so its factor is z2+2z+3z^2 + 2z + 3.

(z−3)(z2+2z+3)=z3+2z2+3z−3z2−6z−9=z3−z2−3z−9.(z - 3)(z^2 + 2z + 3) = z^3 + 2z^2 + 3z - 3z^2 - 6z - 9 = z^3 - z^2 - 3z - 9.

The equation is z3−z2−3z−9=0z^3 - z^2 - 3z - 9 = 0.

Exam-hard: verify a root, then factorise a quartic

(a) Show that z=1+iz = 1 + i is a root of the equation z4+3z2−6z+10=0z^4 + 3z^2 - 6z + 10 = 0.

(b) Hence find the other three roots.

Solution

(a) z2=(1+i)2=2iz^2 = (1 + i)^2 = 2i and z4=(2i)2=−4z^4 = (2i)^2 = -4. Then

z4+3z2−6z+10=−4+6i−6−6i+10=0.z^4 + 3z^2 - 6z + 10 = -4 + 6i - 6 - 6i + 10 = 0.

(b) The coefficients are real, so 1−i1 - i is also a root. Sum 22, product 22: factor z2−2z+2z^2 - 2z + 2.

z4+0z3+3z2−6z+10=(z2−2z+2)(z2+pz+q).z^4 + 0z^3 + 3z^2 - 6z + 10 = (z^2 - 2z + 2)(z^2 + pz + q).

Constant: 2q=102q = 10, so q=5q = 5. z3z^3: p−2=0p - 2 = 0, so p=2p = 2. Check z2z^2: q−2p+2=5−4+2=3q - 2p + 2 = 5 - 4 + 2 = 3. Check zz: 2p−2q=4−10=−62p - 2q = 4 - 10 = -6.

Solve z2+2z+5=0z^2 + 2z + 5 = 0: (z+1)2=−4(z + 1)^2 = -4, so z=−1±2iz = -1 \pm 2i.

The other roots are 1−i1 - i, −1+2i-1 + 2i and −1−2i-1 - 2i.

Exam-hard: a non-monic cubic

Solve the equation 2z3−z2+5z+3=02z^3 - z^2 + 5z + 3 = 0, giving the complex roots in exact form.

Solution

The constant is 33 and the leading coefficient is 22, so try ±1,±3,±12,±32\pm 1, \pm 3, \pm\tfrac{1}{2}, \pm\tfrac{3}{2}.

p(−12)=2(−18)−14−52+3=−14−14−52+3=0p\left(-\tfrac{1}{2}\right) = 2\left(-\tfrac{1}{8}\right) - \tfrac{1}{4} - \tfrac{5}{2} + 3 = -\tfrac{1}{4} - \tfrac{1}{4} - \tfrac{5}{2} + 3 = 0.

So (2z+1)(2z + 1) is a factor. Write 2z3−z2+5z+3=(2z+1)(z2+bz+3)2z^3 - z^2 + 5z + 3 = (2z + 1)(z^2 + bz + 3). The z2z^2 coefficient gives 2b+1=−12b + 1 = -1, so b=−1b = -1. Check zz: b+6=5b + 6 = 5.

Solve z2−z+3=0z^2 - z + 3 = 0:

z=1±1−122=1±i112.z = \frac{1 \pm \sqrt{1 - 12}}{2} = \frac{1 \pm i\sqrt{11}}{2}.

The roots are −12-\tfrac{1}{2}, 12+112i\tfrac{1}{2} + \tfrac{\sqrt{11}}{2}i and 12−112i\tfrac{1}{2} - \tfrac{\sqrt{11}}{2}i.

On an Argand diagram the roots of a real polynomial are always symmetric about the real axis: each non-real root has its mirror image, and the real roots sit on the axis.

Common mistakes

Applying the rule to complex coefficients

The conjugate-pair result needs real coefficients. z2−(3+i)z+(4+3i)=0z^2 - (3 + i)z + (4 + 3i) = 0 has roots 2−i2 - i and 1+2i1 + 2i, which are not conjugates. Always check the coefficients before writing "the conjugate is also a root".

Wrong signs in the quadratic factor

The factor from a±bia \pm bi is z2−2az+(a2+b2)z^2 - 2az + (a^2 + b^2). For −3+2i-3 + 2i it is z2+6z+13z^2 + 6z + 13, not z2−6z+13z^2 - 6z + 13. The middle coefficient is minus the sum; the sum here is −6-6.

Equating only one part

When substituting a root, the result X+Yi=0X + Yi = 0 means X=0X = 0 and Y=0Y = 0. You need both equations to find two unknowns.

Sloppy powers

Build powers step by step: z2z^2, then z3=z2⋅zz^3 = z^2 \cdot z, then z4=(z2)2z^4 = (z^2)^2. Expanding (1+2i)3(1 + 2i)^3 in one go with the binomial theorem and powers of ii is where most sign errors happen.

Exam technique

Exam tip
  • State the reason when you use the conjugate: "since the coefficients are real, 2−i2 - i is also a root". This sentence is often worth a mark.
  • "Verify that α\alpha is a root" or "show that α\alpha is a root" needs the substitution with each power shown and a final line equal to 00. Showing X+YiX + Yi with X=0X = 0 and Y=0Y = 0 is ideal.
  • When comparing coefficients, use the highest and lowest powers first (they involve the fewest unknowns), then use a remaining coefficient as a check. Say "check" and show it.
  • Give all roots at the end, as a list. Do not leave a factorised polynomial as the answer to "solve".
  • If a question says "find the roots" and you have found a real root from a graph or calculator, you still need the factor theorem working to justify it.

Summary

Summary
  • For a polynomial with real coefficients, non-real roots come in conjugate pairs a±bia \pm bi.
  • Each pair gives the real quadratic factor z2−2az+(a2+b2)z^2 - 2az + (a^2 + b^2): minus the sum, plus the product.
  • A real cubic always has at least one real root; a real quartic has 00, 22 or 44 real roots.
  • Given one complex root: write the conjugate, form the quadratic factor, divide or compare coefficients.
  • Unknown coefficients: substitute the root and equate real and imaginary parts, or compare coefficients.
  • A real root can often be found by the factor theorem first, leaving a quadratic.
  • On an Argand diagram the roots of a real polynomial are symmetric about the real axis.

Practice

Question
  1. Given that 1+3i1 + 3i is a root of z3−4z2+14z−20=0z^3 - 4z^2 + 14z - 20 = 0, find the other roots without using the factor theorem.
  2. Find the quartic equation with real coefficients and leading coefficient 11 whose roots include 2i2i and 1+i1 + i.
  3. The equation z3+pz2+qz+26=0z^3 + pz^2 + qz + 26 = 0, with pp and qq real, has a root −3+2i-3 + 2i. Find pp, qq and the third root.
  4. Show that z=2iz = 2i is a root of z3−3z2+4z−12=0z^3 - 3z^2 + 4z - 12 = 0 and find the other roots.
  5. Solve z3−z2+z+3=0z^3 - z^2 + z + 3 = 0.
  6. The equation z3+az2+bz+6=0z^3 + az^2 + bz + 6 = 0, with aa and bb real, has a root 1−i1 - i. Find aa and bb by substituting the root, and find the other roots.
  7. Verify that 2i2i is a root of z4−4z3+9z2−16z+20=0z^4 - 4z^3 + 9z^2 - 16z + 20 = 0 and find all the roots.
  8. Find the quartic equation with real coefficients and leading coefficient 11 that has roots 1+i1 + i and 2−i2 - i, giving it in expanded form.
  9. The equation z3+kz+20=0z^3 + kz + 20 = 0, where kk is real, has a root 1+3i1 + 3i. Find kk and the other two roots.
Answers
  1. Real coefficients, so 1−3i1 - 3i is a root. Sum 22, product 1010: factor z2−2z+10z^2 - 2z + 10. Then z3−4z2+14z−20=(z2−2z+10)(z+k)z^3 - 4z^2 + 14z - 20 = (z^2 - 2z + 10)(z + k) with 10k=−2010k = -20, k=−2k = -2 (check z2z^2: k−2=−4k - 2 = -4). Roots 1±3i1 \pm 3i and 22.

  2. The roots must include the conjugates −2i-2i and 1−i1 - i. Factors z2+4z^2 + 4 and z2−2z+2z^2 - 2z + 2: (z2+4)(z2−2z+2)=z4−2z3+2z2+4z2−8z+8=z4−2z3+6z2−8z+8(z^2 + 4)(z^2 - 2z + 2) = z^4 - 2z^3 + 2z^2 + 4z^2 - 8z + 8 = z^4 - 2z^3 + 6z^2 - 8z + 8. So z4−2z3+6z2−8z+8=0z^4 - 2z^3 + 6z^2 - 8z + 8 = 0.

  3. −3−2i-3 - 2i is a root; sum −6-6, product 1313; factor z2+6z+13z^2 + 6z + 13. (z2+6z+13)(z+k)(z^2 + 6z + 13)(z + k) with 13k=2613k = 26, k=2k = 2. Expanding, z3+8z2+25z+26z^3 + 8z^2 + 25z + 26, so p=8p = 8, q=25q = 25, and the third root is −2-2.

  4. z2=−4z^2 = -4, z3=−8iz^3 = -8i: −8i−3(−4)+4(2i)−12=−8i+12+8i−12=0-8i - 3(-4) + 4(2i) - 12 = -8i + 12 + 8i - 12 = 0. With −2i-2i also a root, z2+4z^2 + 4 is a factor: z3−3z2+4z−12=(z2+4)(z−3)z^3 - 3z^2 + 4z - 12 = (z^2 + 4)(z - 3). Roots 2i2i, −2i-2i, 33.

  5. p(−1)=−1−1−1+3=0p(-1) = -1 - 1 - 1 + 3 = 0, so (z+1)(z + 1) is a factor: z3−z2+z+3=(z+1)(z2−2z+3)z^3 - z^2 + z + 3 = (z + 1)(z^2 - 2z + 3). Then z=2±4−122=1±i2z = \dfrac{2 \pm \sqrt{4 - 12}}{2} = 1 \pm i\sqrt{2}. Roots −1-1, 1+i21 + i\sqrt{2}, 1−i21 - i\sqrt{2}.

  6. z2=(1−i)2=−2iz^2 = (1 - i)^2 = -2i, z3=−2i(1−i)=−2−2iz^3 = -2i(1 - i) = -2 - 2i. Substitute: (−2−2i)+a(−2i)+b(1−i)+6=0(-2 - 2i) + a(-2i) + b(1 - i) + 6 = 0. Real: 4+b=04 + b = 0, b=−4b = -4. Imaginary: −2−2a−b=0-2 - 2a - b = 0, so −2a=−2-2a = -2, a=1a = 1. The equation is z3+z2−4z+6=0=(z2−2z+2)(z+3)z^3 + z^2 - 4z + 6 = 0 = (z^2 - 2z + 2)(z + 3). Other roots 1+i1 + i and −3-3.

  7. z2=−4z^2 = -4, z3=−8iz^3 = -8i, z4=16z^4 = 16: 16+32i−36−32i+20=016 + 32i - 36 - 32i + 20 = 0. With −2i-2i, the factor z2+4z^2 + 4: z4−4z3+9z2−16z+20=(z2+4)(z2−4z+5)z^4 - 4z^3 + 9z^2 - 16z + 20 = (z^2 + 4)(z^2 - 4z + 5) (constant 4×5=204 \times 5 = 20; z3z^3: −4-4; check z2z^2: 5+4=95 + 4 = 9; zz: −16-16). Then z2−4z+5=0z^2 - 4z + 5 = 0 gives z=2±iz = 2 \pm i. Roots ±2i\pm 2i, 2±i2 \pm i.

  8. Conjugates 1−i1 - i and 2+i2 + i are also roots. Factors z2−2z+2z^2 - 2z + 2 and z2−4z+5z^2 - 4z + 5: (z2−2z+2)(z2−4z+5)=z4−4z3+5z2−2z3+8z2−10z+2z2−8z+10=z4−6z3+15z2−18z+10(z^2 - 2z + 2)(z^2 - 4z + 5) = z^4 - 4z^3 + 5z^2 - 2z^3 + 8z^2 - 10z + 2z^2 - 8z + 10 = z^4 - 6z^3 + 15z^2 - 18z + 10. So z4−6z3+15z2−18z+10=0z^4 - 6z^3 + 15z^2 - 18z + 10 = 0.

  9. 1−3i1 - 3i is also a root, giving the factor z2−2z+10z^2 - 2z + 10. Write z3+0z2+kz+20=(z2−2z+10)(z+c)z^3 + 0z^2 + kz + 20 = (z^2 - 2z + 10)(z + c). The z2z^2 coefficient: c−2=0c - 2 = 0, so c=2c = 2; check constant: 10×2=2010 \times 2 = 20. The zz coefficient: k=10−2c=6k = 10 - 2c = 6. So k=6k = 6 and the roots are 1+3i1 + 3i, 1−3i1 - 3i, −2-2.

How well do you know this?

Builds on

Where this leads

Console

Search notes, courses and tools, or run an action