Powers and Roots of Complex Numbers

Finding powers and roots of complex numbers is a two-step process that rewards fluency in polar form. The one-dimensional number line that works for real numbers stops being enough once you square a number and get a negative. Complex numbers, written as a + bi with real part a and imaginary part b, live in a two-dimensional plane. The distance from the origin to the point (a, b) is the modulus, written r. The angle from the positive real axis to the point is the argument, written θ. For multiplication, division, powers, and roots, polar form turns what would be algebraic drudgery into repeated angle arithmetic.

The single most useful tool here is De Moivre's theorem, which states that [r(cos θ + i sin θ)]^n = r^n(cos(nθ) + i sin(nθ)). It works for integer n, positive, negative, and zero, and it is the backbone of every root calculation. If you can convert a complex number from rectangular to polar form, and back again, you can raise it to any power or extract any root. If you cannot convert fluently, stop and practice that first, because every example below assumes you can.

De Moivre's theorem is not a formula you memorize and then forget. It is the reason powers of complex numbers become tractable. For any complex number z = r(cos θ + i sin θ) and any integer n, the theorem gives z^n = r^n(cos(nθ) + i sin(nθ)). The modulus is raised to the power, and the argument is multiplied by the power. That is the entire content. There is no hidden condition about the size of n, no caveat about the quadrant, nothing.

The theorem holds for negative integers too. For z^-n, you get r^-n(cos(-nθ) + i sin(-nθ)), which is the same as taking the reciprocal of the modulus and negating the angle. For n = 0, provided z is not zero, you get z^0 = 1, consistent with the real-number rule. A common mistake is to think the theorem requires a positive exponent or a principal argument. It does not. As long as n is an integer, the formula is exact. This is a proven theorem, not a pedagogical preference, and it appears in standard treatments such as OpenStax Precalculus 2e, section 8.5.

Take the complex number z = 1 + i. In rectangular form, a = 1 and b = 1. The modulus r = sqrt(1^2 + 1^2) = sqrt(2). The argument, because both parts are positive, is θ = π/4. In polar form, z = sqrt(2)(cos(π/4) + i sin(π/4)).

Now raise it to the fifth power. Using De Moivre's theorem, z^5 = (sqrt(2))^5 (cos(5·π/4) + i sin(5·π/4)). The modulus becomes (sqrt(2))^5 = 2^(5/2) = 4·sqrt(2). The angle becomes 5π/4. So z^5 = 4·sqrt(2)(cos(5π/4) + i sin(5π/4)). Since cos(5π/4) = -sqrt(2)/2 and sin(5π/4) = -sqrt(2)/2, the result is 4·sqrt(2)·(-sqrt(2)/2 - i·sqrt(2)/2) = 4·(-2/2 - i·2/2) = -4 - 4i.

The alternative, expanding (1+i)^5 by binomial theorem, is possible but slow and error-prone. Polar form is faster because it converts the exponentiation into a multiplication of angles. If you find yourself expanding brackets for a power above 3, you are working too hard. The modulus is always non-negative, and it is what you compare when checking if two complex numbers are equal, not the angle alone.

The phrase nth roots of complex numbers refers to a specific fact: for any nonzero complex number z and any positive integer n, there are exactly n distinct complex numbers w such that w^n = z. Not one, not two, exactly n. This is a consequence of the fundamental theorem of algebra, which guarantees that the polynomial w^n - z = 0 has n complex roots counting multiplicity, and for z ≠ 0 they are all distinct.

The formula for these roots is direct. If z = r(cos θ + i sin θ), then the n roots are given by r^(1/n) (cos((θ + 2πk)/n) + i sin((θ + 2πk)/n)) for k = 0, 1, ..., n-1. The modulus of each root is the positive real nth root of r, written r^(1/n). The angles are equally spaced around the circle, separated by 2π/n radians. The k = 0 root is the principal root, the one with the smallest non-negative argument. The other roots come from adding full rotations to the angle before dividing.

A common failure is to stop at the principal root. Your calculator, depending on the model, may return only that one value. The Casio fx-991EX, for instance, gives a single result for a fractional power, not the full set. For exam questions that ask for all roots, you must add the k values yourself. The distinction matters in AC circuit analysis, where the phase angle of a root determines the timing of a waveform, and in polynomial factorization, where missing a root means missing a factor.

The square root of a complex number is the n = 2 case of the general formula. For z = r(cos θ + i sin θ), the two square roots are r^(1/2)(cos(θ/2) + i sin(θ/2)) and r^(1/2)(cos((θ + 2π)/2) + i sin((θ + 2π)/2)). These are separated by π radians, so they are negatives of each other, which is expected because squaring either gives z back.

Take the square root of i. Here r = 1 and θ = π/2. The principal root, k = 0, is cos(π/4) + i sin(π/4) = sqrt(2)/2 + i·sqrt(2)/2. The second root, k = 1, is cos(5π/4) + i sin(5π/4) = -sqrt(2)/2 - i·sqrt(2)/2. These are the two answers. If your calculator returns only one, it is returning the principal root, which is a convention, not the full solution.

There is also an algebraic method that avoids trigonometry entirely. For w = a + bi, you want (a + bi)^2 = x + yi. Expanding gives a^2 - b^2 = x and 2ab = y. Solving these simultaneously for a and b yields a system that, while more laborious, works when you cannot or will not convert to polar form. The algebraic method is slower for most people, but it has one advantage: it forces you to track signs explicitly, which reduces the quadrant errors that plague the trigonometric approach. For a square root, either method works; for higher roots, polar form is the only practical route.

The roots of unity are the solutions to z^n = 1 for a positive integer n. They are a special case of the nth roots of complex numbers where the modulus r = 1 and the argument θ = 0. The formula gives z_k = cos(2πk/n) + i sin(2πk/n) for k = 0, 1, ..., n-1. These roots always lie on the unit circle in the complex plane, equally spaced by an angle of 2π/n.

Two properties are worth memorizing because they appear in exam questions and in signal processing. First, the sum of all n roots of unity is 0 for n > 1. For n = 3, the roots are 1, -1/2 + i·sqrt(3)/2, and -1/2 - i·sqrt(3)/2; their sum is 1 - 1/2 + i·sqrt(3)/2 - 1/2 - i·sqrt(3)/2 = 0. Second, the product of all n roots of unity equals (-1)^(n+1). For n = 3, the product is 1 · (-1/2 + i·sqrt(3)/2) · (-1/2 - i·sqrt(3)/2) = 1 · (1/4 + 3/4) = 1, which matches (-1)^(3+1) = 1.

The picture is a regular n-gon inscribed in the unit circle. The cube roots of 1 are an equilateral triangle with one vertex at 1. The fourth roots of 1 are the points 1, i, -1, -i, a square. The fifth roots are a regular pentagon. Drawing these by hand clarifies why the angles are evenly spaced: you are dividing the full circle into n equal slices. This geometric view is what makes the roots of unity useful in digital signal processing, where they are called twiddle factors, and in polynomial factorization, where they generate cyclic groups under multiplication.

To see the full process, find the fourth roots of -16. In polar form, -16 has modulus r = 16 and argument θ = π (or any odd multiple of π). The fourth roots have modulus r^(1/4) = 16^(1/4) = 2. The angles are (π + 2πk)/4 for k = 0, 1, 2, 3.

For k = 0, the angle is π/4, giving 2(cos(π/4) + i sin(π/4)) = sqrt(2) + i·sqrt(2). For k = 1, the angle is 3π/4, giving 2(cos(3π/4) + i sin(3π/4)) = -sqrt(2) + i·sqrt(2). For k = 2, the angle is 5π/4, giving -sqrt(2) - i·sqrt(2). For k = 3, the angle is 7π/4, giving sqrt(2) - i·sqrt(2).

Check the first one: (sqrt(2) + i·sqrt(2))^2 = 2 + 2·sqrt(2)·i·sqrt(2) + (i·sqrt(2))^2 = 2 + 4i - 2 = 4i. Squaring again, (4i)^2 = 16i^2 = -16. It works. The four roots form a square rotated 45 degrees from the axes. If you only got the k = 0 root, you missed three-quarters of the answer, which is exactly the kind of error that costs marks on a Further Maths paper.

Quadrant Checks Before Arctan

The most frequent error is taking arctan(b/a) as the argument without checking the quadrant. That formula is only correct when a > 0 and b > 0. For a < 0 or b < 0, the angle is off by π. A quick rule: draw the point. If it is in the second quadrant, add π; third, add π; fourth, add 2π or use a negative angle. Your calculator returns a principal value in the range (-π/2, π/2) for arctan, which is not the same as the argument for most points. This is a known trap in textbooks that state the formula without the quadrant condition.

Don't Stop at One Root

A second failure is trusting the calculator's single output for a root. The Casio fx-991EX and the TI-84 both give one result for a fractional power, typically the principal root. The problem asks for all roots, so you must generate the rest using the + 2πk term. The calculator is not wrong; it is following a convention. You are wrong if you stop there.

Integer Exponents Only for De Moivre

A third error is misapplying De Moivre's theorem to fractional n. The theorem as stated is for integer n. For rational n, it gives one root, the principal one, not all of them. The full set of n roots requires the k-index formula. Textbooks often state the integer case and leave the fractional case as an exercise, which is why students miss roots in exams. Remember: for z^(1/n), there are n distinct answers, and the principal root is the one with the smallest non-negative argument.

Finally, do not confuse the modulus with the absolute value. For real numbers they coincide, but for complex numbers the modulus is always non-negative and is the distance from the origin. The notation is what distinguishes them: |z| for complex, |x| for real, but the value is the same only when the imaginary part is zero.

Roots of complex numbers suit anyone preparing for A-level Further Maths, IB Analysis and Approaches HL, or a first university course in engineering or physics. If you need to solve polynomial equations, analyze AC circuits, or understand signal processing, the ability to find all n roots quickly is non-negotiable. The subject rewards practice with polar form and punishes a lazy quadrant check.

It does not suit a casual learner who wants to avoid angle arithmetic. If you are not comfortable converting between rectangular and polar form, or if you expect every calculation to come out as a nice integer, you will fight the material. The algebraic method for square roots is a fallback, but it does not scale to higher powers or roots. For those readers, the time is better spent mastering polar form first.