In short
The nth roots of a complex number w = R(\cos\phi + i\sin\phi) are the n solutions of z^n = w. They are
These n roots are equally spaced on a circle of radius R^{1/n}, forming a regular n-gon. They differ from each other by multiplication by the roots of unity — once you find one root, the rest are that root times 1, \omega, \omega^2, \dots, \omega^{n-1}.
You know how to find the nth roots of 1 — they sit equally spaced on the unit circle. But what about the nth roots of an arbitrary complex number? What are the cube roots of 8i, or the fourth roots of -4?
The answer turns out to be almost as clean. The roots still form a regular polygon — just on a different circle, rotated to a different starting angle. The roots of unity are the special case; the general case is one small step further.
The formula
Start with a non-zero complex number w in polar form:
where R = |w| > 0 and \phi = \arg(w). You want all z such that z^n = w.
Write z in polar form too: z = r(\cos\theta + i\sin\theta). By De Moivre's theorem:
Set this equal to w = R(\cos\phi + i\sin\phi) and match modulus and argument:
Modulus: r^n = R, so r = R^{1/n} (the unique positive real nth root of R).
Argument: n\theta = \phi + 2k\pi for integer k, so \theta = \frac{\phi + 2k\pi}{n}.
As k runs from 0 to n-1, you get n distinct values of \theta, each separated by \frac{2\pi}{n}. At k = n, the angle wraps around to \theta_0 + 2\pi, which is the same point as k = 0.
$n$th roots of a complex number
Let w = R(\cos\phi + i\sin\phi) with R > 0. The n solutions of z^n = w are
In exponential form: z_k = R^{1/n}\, e^{i(\phi + 2k\pi)/n}.
Three observations jump out from this formula:
-
All n roots have the same modulus R^{1/n}. They lie on a circle of radius R^{1/n}, not necessarily the unit circle.
-
The roots are equally spaced — consecutive roots differ by an angle of 2\pi/n, the same spacing as the nth roots of unity.
-
They form a regular n-gon centred at the origin, rotated so that the first root (k = 0) sits at angle \phi/n.
The connection to roots of unity
Here is the cleanest way to think about it. Let z_0 be any one particular nth root of w — say the one with k = 0:
Then all the nth roots of w are:
where \omega = e^{2\pi i/n} is a primitive nth root of unity.
Check: (z_0 \omega^k)^n = z_0^n \cdot (\omega^k)^n = w \cdot \omega^{kn} = w \cdot 1 = w. Each one is indeed an nth root of w.
This is the key insight: finding the nth roots of any complex number reduces to finding one root and then multiplying by the nth roots of unity. The roots of unity act like a "rotational offset" that generates all the others from a single starting point.
Geometric interpretation
The n roots of a complex number w always form a regular n-gon:
- Centre: the origin (not w).
- Radius: R^{1/n}, where R = |w|.
- Orientation: the first vertex (k = 0) sits at angle \phi/n from the positive real axis, where \phi = \arg(w).
- Spacing: consecutive vertices are 2\pi/n radians apart.
This means you can see the roots before computing them. Given w, plot it on the complex plane. The roots sit on a smaller circle (radius = nth root of the distance), rotated so that the angle is divided by n, and replicated n times around the circle.
When w is a positive real number — say w = 8 — the angle \phi = 0, so the first root sits on the positive real axis at 8^{1/3} = 2 (for cube roots). The other roots are at 120° and 240°:
Only one of these roots is real. The other two are complex conjugates. This is a general pattern: a positive real number has exactly one real nth root (when n is odd) or two real nth roots (when n is even, \pm R^{1/n}), and the remaining roots come in conjugate pairs.
Applications
Solving polynomial equations
The equation z^4 = -16 is a fourth-degree polynomial equation. Writing -16 = 16(\cos\pi + i\sin\pi), the four roots are:
For k = 0, 1, 2, 3, the angles are \pi/4, 3\pi/4, 5\pi/4, 7\pi/4 — the four diagonal directions. In Cartesian form:
These are the vertices of a square with side length 2\sqrt{2}, rotated 45° from the axes.
Factoring polynomials
The identity z^n - w = (z - z_0)(z - z_1) \cdots (z - z_{n-1}) factors any binomial z^n - w into linear factors over the complex numbers. For z^3 - 8:
The quadratic factor z^2 + 2z + 4 is irreducible over the reals — its roots 2\omega and 2\omega^2 are complex conjugates with discriminant 4 - 16 = -12 < 0.
Interactive: exploring nth roots
Drag the red point along the curve below. The readouts show the coordinates (x, y) and the modulus r = \sqrt{x^2 + y^2}. As you move the point, think about where the cube roots of that number would sit — on a circle of radius r^{1/3} at one-third of the argument.
Worked examples
Example 1: Find the cube roots of $8i$
A complex number with a clean modulus and argument, giving roots with exact trigonometric values.
Step 1. Write 8i in polar form.
So 8i = 8\left(\cos\frac{\pi}{2} + i\sin\frac{\pi}{2}\right).
Why: 8i sits on the positive imaginary axis, at distance 8 from the origin and angle \pi/2 from the real axis.
Step 2. Apply the nth root formula with n = 3, R = 8, \phi = \pi/2.
Why: 8^{1/3} = 2. The angle \frac{\pi/2 + 2k\pi}{3} = \frac{\pi + 4k\pi}{6} for each k = 0, 1, 2.
Step 3. Compute each root.
k = 0: \theta_0 = \pi/6.
k = 1: \theta_1 = \pi/6 + 2\pi/3 = 5\pi/6.
k = 2: \theta_2 = \pi/6 + 4\pi/3 = 3\pi/2.
Why: each root is at angle \pi/6 + k \cdot 2\pi/3 on a circle of radius 2. The three angles \pi/6, 5\pi/6, 3\pi/2 are 120° apart.
Step 4. Verify z_0^3 = 8i.
Why: a direct computation confirms the formula. The real parts cancel (2\sqrt{3} - 2\sqrt{3} = 0), leaving only the imaginary part 8i.
Result: The cube roots of 8i are \sqrt{3} + i, -\sqrt{3} + i, and -2i.
Notice that two of the three roots have the same imaginary part (+i), while the third root -2i is purely imaginary. None of the cube roots of 8i are real — which makes sense, because a real number cubed is real, not purely imaginary.
Example 2: Find the square roots of $-3 + 4i$
A complex number with no clean polar form — you work algebraically instead of using explicit trigonometric values.
Step 1. Suppose z = a + bi satisfies z^2 = -3 + 4i. Expand:
Why: for square roots, the direct algebraic approach is often cleaner than the polar form, since you get a system of two real equations.
Step 2. Match real and imaginary parts.
From (ii): b = 2/a.
Why: two equations, two unknowns. From the imaginary equation, express b in terms of a and substitute into the real equation.
Step 3. Substitute into (i).
Multiply through by a^2:
This is a quadratic in a^2. Let u = a^2:
So u = 1 or u = -4. Since u = a^2 \ge 0, take u = 1, giving a = \pm 1.
Why: a is real, so a^2 cannot be negative. The root u = -4 is discarded.
Step 4. Find corresponding b values and write the roots.
If a = 1: b = 2/1 = 2. So z = 1 + 2i. If a = -1: b = 2/(-1) = -2. So z = -1 - 2i.
Why: the two square roots are negatives of each other, as always — if z^2 = w, then (-z)^2 = w too.
Step 5. Verify: (1 + 2i)^2 = 1 + 4i + 4i^2 = 1 + 4i - 4 = -3 + 4i. \checkmark
Result: The square roots of -3 + 4i are 1 + 2i and -1 - 2i.
The algebraic method (matching real and imaginary parts of z^2 = w) works cleanly whenever the resulting equation in a has rational or clean irrational roots. For higher powers, or when the algebra gets unwieldy, the polar form approach is more systematic. Both methods give the same answer.
Common confusions
-
"A complex number has only one nth root." Every non-zero complex number has exactly n distinct nth roots. The familiar notation \sqrt{9} = 3 is a convention for the principal (positive real) root. But z^2 = 9 has two solutions: 3 and -3. For complex numbers, z^n = w always has n solutions.
-
"The nth root formula gives different answers if I use \phi + 2\pi instead of \phi." If you replace \phi with \phi + 2\pi, then \frac{\phi + 2\pi + 2k\pi}{n} = \frac{\phi + 2(k+1)\pi}{n}, which shifts each root's label by one. You get the same set of n roots, just reindexed.
-
"I can take the nth root of 0." The formula requires R > 0. The only solution of z^n = 0 is z = 0 (with multiplicity n). There is no "regular polygon of roots" for w = 0.
-
"The roots form a polygon centred at w." The polygon is centred at the origin, not at w. The number w sits on its own, typically far from the roots. The roots relate to w through the power map z \mapsto z^n, not through proximity.
-
"To find the square root of a + bi, just take \sqrt{a} + \sqrt{b}\,i." This is wrong. Try it with -3 + 4i: \sqrt{-3} is not real, and the formula makes no sense. The correct method is either (a) the polar form approach or (b) setting (x + yi)^2 = a + bi and solving the real system, as in Example 2.
Going deeper
If you can find the nth roots of any complex number using the polar form, understand why they form a regular polygon, and connect the general roots to the roots of unity, you have the core. The following material explores where these ideas lead.
The fundamental theorem of algebra
Every polynomial of degree n with complex coefficients has exactly n roots in \mathbb{C} (counted with multiplicity). The nth root formula is a special case: z^n - w = 0 is a degree-n polynomial, and it has exactly n roots. The fundamental theorem guarantees this works for every polynomial, not just binomials — but for general polynomials, there is no closed-form formula (when n \ge 5, by the Abel-Ruffini theorem).
Principal roots and branch cuts
When you write w^{1/n}, which of the n roots do you mean? The convention is to take the principal root: the one with the smallest positive argument (equivalently, k = 0 in the formula). But this choice is artificial — the map w \mapsto w^{1/n} cannot be made continuous on the entire complex plane. The points where continuity breaks form a branch cut, usually placed along the negative real axis. This is the beginning of the theory of multi-valued functions and Riemann surfaces — a topic central to complex analysis.
Solving z^n = w geometrically
Before the algebraic formula, there is a compass-and-straightedge question: can you construct the nth roots of a given complex number? The answer depends on n. You can always construct z_0 = R^{1/n} when R is constructible, and you can subdivide the angle \phi into n equal parts only for certain values of n. Angle trisection is famously impossible in general, which means the cube roots cannot always be constructed with compass and straightedge — one of the great results of classical geometry, settled only with the algebraic theory of field extensions.
Where this leads next
- Roots of Unity — the special case w = 1, where the roots are the vertices of a regular n-gon on the unit circle.
- De Moivre's Theorem — the power formula that the nth root formula inverts.
- Polar Form of Complex Numbers — the representation that makes the root formula possible.
- Geometry with Complex Numbers — Basics — using complex coordinates to describe distances, midpoints, and sections on the plane.
- Polynomial Equations — the general theory of solving p(z) = 0, where the fundamental theorem of algebra guarantees n roots.