Is the argument of a complex number always arctan(y/x)?

No. arctan(y/x) equals the argument only when the real part of z is positive, the right half-plane. Elsewhere the ratio y/x cannot tell one quadrant from another, so you must place z by its quadrant first and read the principal value in the range from -pi to pi. For example -1 - i root 3 sits in the third quadrant, so its principal argument is -2pi/3, while the naive arctan of (-root 3)/(-1) = arctan(root 3) = +pi/3 is off by pi. This is exactly the structure of JEE Advanced 2025 Paper 2 Q13.

Does changing the sign of a root of unity leave its powers unchanged?

No. A factor of modulus 1 such as -1 = e to the i pi leaves the modulus alone but adds pi to the argument, and that reshapes the cycle of powers. If w is a primitive cube root of unity, its order is 3, but the order of -w is 6, because (-w) to the n = (-1) to the n times w to the n equals 1 only when n is even and a multiple of 3. In particular (-w) cubed = -1, not 1.

What is the principal argument of a number built from squares, conjugates and reciprocals?

Track it by the rules, not by arctan of the original coordinates. Conjugating negates the argument, squaring doubles it, and taking a reciprocal negates it. So for z = root 3 + i with argument pi/6, both the conjugate-square and the reciprocal-square have argument -2 times pi/6 = -pi/3. This shared-argument structure is the surface of JEE Advanced 2022 Paper 2 Q11.

Why does a sum of powers of -w not collapse like a cube root?

Because -w has order 6, not 3. For a genuine cube root of unity three consecutive powers sum to zero, but 1 + (-w) + (-w) squared = 1 - w + w squared = -2w, which is not zero. Evaluating 1 + (-w) + (-w) squared + (-w) cubed + (-w) to the fourth term by term gives w squared, whose principal argument is -2pi/3. The tail survives precisely because the sign factor doubled the order.

JEE Advanced Maths · Complex Numbers

Is arg z always arctan(y/x)?

Not in general. $\arctan(y/x)$ equals the argument only when $\mathrm{Re}\,z > 0$ , the right half-plane. Everywhere else the ratio $y/x$ cannot tell one quadrant from another, so you must place $z$ by its quadrant first and read the principal value in $(-\pi, \pi]$ .

The same care governs periodicity. A factor of modulus $1$ , such as $-1 = e^{i\pi}$ , is never cosmetic: it adds $\pi$ to the argument, and that reshapes the cycle of powers. If $\omega$ is a primitive cube root of unity, the order of $-\omega$ is $6$ , not $3$ . Both slips are tested across JEE Advanced 2022 and 2025.

Ready to fix this? The Complex Numbers lesson works through this misconception and the others in Complex Numbers, one altitude at a time.

How to spot it in your own work

You wrote $\arg z = \arctan(y/x)$ for a point with $\mathrm{Re}\,z < 0$ without correcting by $\pi$ .
You reported an argument outside $(-\pi, \pi]$ , or chased "all the arguments" instead of the principal one.
You claimed $(-\omega)^3 = 1$ , treating the sign as something cubing erases.
You reduced the exponent of $-\omega$ modulo $3$ , as if its order were still $3$ .

An exam question that triggers it

Here is a question in the style of JEE Advanced 2025 Paper 2 Q13, which tests this misconception head-on:

Find the principal argument, in $(-\pi, \pi]$ , of

z = -1 - i\sqrt{3}.

The misconception answer reaches for $\arctan(y/x)$ : $\arctan\!\big(\tfrac{-\sqrt{3}}{-1}\big) = \arctan(\sqrt{3}) = +\tfrac{\pi}{3}$ . The two minus signs cancel inside the ratio, so $\arctan$ reports a first-quadrant angle for a third-quadrant point.

Place $z$ first: $\mathrm{Re}\,z = -1 < 0$ and $\mathrm{Im}\,z = -\sqrt{3} < 0$ , so $z$ is in the third quadrant, with reference angle $\arctan(\sqrt{3}) = \tfrac{\pi}{3}$ . The principal argument is therefore

\arg z = -\Big(\pi - \tfrac{\pi}{3}\Big) = -\tfrac{2\pi}{3},

off from the naive $+\tfrac{\pi}{3}$ by exactly $\pi$ .

Why students fall for this

The belief underneath is that $\arg z$ is a formula you evaluate, $\arctan(y/x)$ , rather than a direction you locate. In real-variable work $\arctan$ really does invert the tangent, and it happens to give the argument whenever $\mathrm{Re}\,z > 0$ , so the shortcut feels safe until a point crosses into the left half-plane.

The periodicity half has the same root: a sign change looks too small to matter. But $-1 = e^{i\pi}$ adds $\pi$ to the argument, and multiplying $\omega = e^{2\pi i/3}$ by it gives $-\omega = e^{-i\pi/3}$ , whose powers need six steps to return to $1$ . What dislodges both habits is computing $(-\omega)^3$ once and watching it come out $-1$ .

The fix: Fix the quadrant, and treat a sign flip as an argument shift

$\arg z = \arctan(y/x)$ only when $\mathrm{Re}\,z > 0$ ; otherwise place $z$ by its quadrant and read the principal value in $(-\pi, \pi]$ . In the second quadrant add $\pi$ to the reference-angle result; in the third, take $-(\pi - \text{reference angle})$ ; in the fourth, take the negative reference angle.

For powers, a factor of modulus $1$ shifts the argument and can change the order. Write $(-\omega)^n = (-1)^n \omega^n$ : it equals $1$ only when $n$ is even and a multiple of $3$ , so the order of $-\omega$ is $6$ . Track built-up expressions by the rules too: conjugating negates the argument, squaring doubles it, a reciprocal negates it.

Worked example

Cube once, watch the sign survive, then use the structure that explains it. Take $\omega$ a primitive cube root of unity.

Cube $-\omega$ . $(-\omega)^3 = (-1)^3 \omega^3 = (-1)(1) = -1$ , not $1$ . An odd power keeps the sign.
Find the order. $(-\omega)^n = (-1)^n \omega^n = 1$ needs $n$ even and a multiple of $3$ , so $n = 6$ . The powers run $-\omega,\ \omega^2,\ -1,\ \omega,\ -\omega^2,\ 1$ , closing only after six steps.
Read a principal argument. For $-1 - i\sqrt{3}$ (third quadrant), $\arg = -(\pi - \tfrac{\pi}{3}) = -\tfrac{2\pi}{3}$ , while $\arctan(y/x) = +\tfrac{\pi}{3}$ is off by $\pi$ .
Track a built-up argument. For $z = \sqrt{3} + i$ with $\arg z = \tfrac{\pi}{6}$ , both $\bar z^{2}$ and $1/z^{2}$ have argument $-2\cdot\tfrac{\pi}{6} = -\tfrac{\pi}{3}$ .
Sum a tail that survives. With $\zeta = -\omega$ , $1 + \zeta + \zeta^2 + \zeta^3 + \zeta^4 = 1 - \omega + \omega^2 - 1 + \omega = \omega^2$ , whose principal argument is $-\tfrac{2\pi}{3}$ . It does not cancel, because $\zeta$ has order $6$ .
Conclude.
$\text{fix the quadrant for } \arg; \quad \text{a modulus-1 factor shifts the argument and can double the order.}$

The same reasoning settles JEE Advanced 2025 Paper 2 Q13 (a third-quadrant principal argument, not the $\arctan$ value) and 2022 Paper 2 Q11 (a shared argument tracked through a conjugate and a reciprocal).

Find out if this is costing you marks

The 10-minute diagnostic checks for this pattern (and four others) using AQA-style GCSE Higher items. Free, no signup, anonymous.

Take the diagnostic →

Common questions

When is $\arg z = \arctan(y/x)$ actually correct?: Only in the right half-plane, $\mathrm{Re}\,z > 0$ . There the point lies in the first or fourth quadrant, where $\arctan(y/x)$ already falls in $(-\tfrac{\pi}{2}, \tfrac{\pi}{2})$ and matches the principal argument. For $\mathrm{Re}\,z < 0$ you must add or subtract $\pi$ to reach the correct quadrant.
What is $(-\omega)^3$ , and why does it matter?: $(-\omega)^3 = (-1)^3\omega^3 = -1$ , not $1$ . It matters because it shows $-\omega$ is not a cube root of unity: its order is $6$ . A sum or product that would collapse for $\omega$ does not collapse the same way for $-\omega$ .
How do I find the argument of $\bar z^{2}$ or $1/z^{2}$ ?: Use the rules. If $\arg z = \theta$ , then $\arg(\bar z^{2}) = -2\theta$ (conjugate negates, square doubles) and $\arg(1/z^{2}) = -2\theta$ (square doubles, reciprocal negates). For $z = \sqrt{3} + i$ with $\theta = \tfrac{\pi}{6}$ , both are $-\tfrac{\pi}{3}$ .
Does a sum of powers of $-\omega$ cancel like a cube root?: No. Three consecutive powers of a genuine cube root of unity sum to zero, but $1 + (-\omega) + (-\omega)^2 = 1 - \omega + \omega^2 = -2\omega \ne 0$ . Summed out to $(-\omega)^4$ , the total is $\omega^2$ , with principal argument $-\tfrac{2\pi}{3}$ . The order- $6$ structure keeps the tail alive.

Related misconceptions

Does an equation in z pin down a few points?No: a single modulus or argument condition is one real equation on a complex z, so it carves out a whole curve; an Apollonius circle is centred outside the segment, never at the midpoint.
Is a fraction real when its top is real?No: w is real is the single equation w = w-bar you impose and solve, not something you read off a numerator; z-bar = 1/z holds only on the unit circle.

← All GCSE Maths Higher misconceptions