A complex number is a pair of ordinary numbers with one particular multiplication rule. The word “imaginary” is a historical accident that has damaged more confidence than any other term in mathematics.
In polar form the multiplication rule is simple: multiply the magnitudes, add the angles. So multiplying scales and rotates. That is why complex numbers are the natural language for anything that oscillates. And j² = −1 stops being a strange axiom — j has magnitude one and angle 90°, so multiplying by it twice is two quarter turns, which is a half turn, which is multiplying by −1.
A sinusoid is the shadow of something rotating
Interactive
ejθ = cos θ + j sin θMagnitude 1Constellation aligned
Drag the angle and watch the arrow’s two shadows: horizontally it traces a cosine, vertically a sine. Neither is fundamental — the rotation is, and it carries both at once. That is all Euler’s formula says. The second slider rotates the 8-PSK constellation the way a carrier frequency offset does in the HF lab; push it past 22.5° and points cross into their neighbour’s decision region, which is exactly the failure carrier recovery exists to prevent.
Why radio needs this
One complex number replaces an amplitude and a phase. Combining sinusoids becomes addition; delaying one becomes multiplication. All the trigonometric identities collapse into ordinary arithmetic. And complex exponentials are the eigenfunctions of linear systems: feed one in, and the same function comes out multiplied by a single complex constant whose magnitude is the gain and whose angle is the phase shift.
For most engineering the complex numbers are scaffolding — erected to make the algebra easy, taken down at the end. In communications they are not. A real signal cannot distinguish a frequency above a reference from one below it; both land on top of each other when mixed down, which is the image problem. A complex signal can, because its spectrum need not be symmetric.
That is why every modern radio works with I and Q as the real and imaginary parts of one complex baseband signal. A constellation diagram is complex numbers plotted on the complex plane; carrier offset makes it rotate; carrier recovery is measuring and undoing that rotation. Here the complex signal is the object, and treating it as two real signals is the awkward description.
Key points
A complex number is a pair of ordinary numbers with one particular multiplication rule. The word "imaginary" is a historical accident and has damaged more confidence than any other term in mathematics.
Complex numbers live on a plane rather than a line. The horizontal axis is real, the vertical is imaginary, and a complex number is a point or an arrow from the origin.
Engineers write j rather than i, because i was taken by electric current.
Addition is componentwise, exactly like vectors. Multiplication is where the structure lives.
In polar form, multiplication is simple: multiply the magnitudes and add the angles. Multiplying by a complex number scales and rotates. Multiplying by one of magnitude one is pure rotation.
j has magnitude one and angle ninety degrees, so multiplying by j is a quarter turn. j squared equals minus one is just the statement that two quarter turns make a half turn.
Euler's formula says e to the j x is the point on the unit circle at angle x. The exponential of an imaginary number is a rotation.
So e to the j omega t is an arrow rotating steadily — a phasor. A sinusoid is its shadow. The rotation is the more fundamental object, since it contains both the sine and the cosine with no arbitrary choice between them.
One complex number replaces an amplitude and a phase. Combining sinusoids becomes complex addition; delaying one becomes complex multiplication. All the trigonometric identities collapse into ordinary arithmetic.
Complex exponentials are the eigenfunctions of linear time invariant systems: feed one in and the same function comes out multiplied by a single complex constant, whose magnitude is the gain and whose angle is the phase shift.
Solution by substitution replaces a real problem with a complex one, solves it easily, and takes the real part at the end. The complex numbers are scaffolding, and the answer contains nothing imaginary.
The complex DFT takes N complex inputs to N complex outputs with one formula and no special cases, unlike the real DFT with its two arrays and its exceptions at zero and Nyquist.
Output indices zero through N over two are positive frequencies; indices above N over two are negative frequencies, with the most negative in the middle. Libraries provide a shift function to reorder them for plotting, and forgetting it is a common bug.
For a real input, the negative frequency half mirrors the positive half and carries no new information. For a genuinely complex signal the two halves are independent, which is where negative frequencies become meaningful.
A real signal cannot distinguish a frequency above a reference from one below it — the image problem. A complex signal can, because its spectrum need not be symmetric.
That is why radio systems use I and Q channels as the real and imaginary parts of a complex baseband signal. A constellation diagram is complex numbers plotted on the complex plane, carrier offset makes it rotate, and carrier recovery is measuring and undoing that rotation.