The complex derivative of a complex function at is

provided the limit exists. The definition looks identical to the real-variable case. The decisive difference: is a complex number, and the limit must give the same value along every direction of approach in the plane.

This requirement is much stronger than two-dimensional real differentiability of the components. It forces a deep regularity on the functions that pass: they turn out to be infinitely differentiable, equal to their Taylor series, and rigid in ways real functions never are.

A computation that succeeds

. Difference quotient:

The limit is regardless of how approaches . So everywhere — same as the real-variable formula.

A function with no complex derivative anywhere

. Difference quotient:

Along the real axis (, real): the quotient equals .

Along the imaginary axis (): the quotient equals .

Different directional limits, so is complex differentiable nowhere.

This is striking because is smooth in every real-variable sense — and are linear and infinitely differentiable. The function fails complex differentiability not because it’s badly behaved as a function of , but because its behavior is orientation-reversing (a reflection), while complex differentiability requires the function to behave locally like rotation-and-scaling.

More diagnostic failures

. Difference quotient . Along reals: . Along imaginaries: . Differentiable nowhere.

. Write with and . Computing along the real and imaginary axes:

  • Along real: the difference quotient is .
  • Along imaginary: the quotient is .

These agree only when , i.e., when and . So is differentiable only at the origin, and not analytic anywhere — analyticity requires differentiability on a whole open set, but a single point isn’t open.

Differentiation rules

Despite the stringent condition, the standard rules carry over for differentiable and complex constant :

RuleStatement
Sum
Constant multiple
Product
Quotient
Chain
Power

Proofs don’t use that the variable is real — only that has field structure. So polynomials, rational functions, , , all differentiate as in real calculus, with the same formulas.

The genuinely complex content is hidden in the question: is the function differentiable to begin with? Polynomials and rationals are; products, quotients, compositions of differentiables are; but plenty of perfectly smooth-looking real-variable functions (, , , ) are not, anywhere.

The test: Cauchy–Riemann

The clean test for complex differentiability is the Cauchy-Riemann equations. Writing , is complex differentiable at iff and at , plus continuity of the partials in a neighborhood. The derivative is then .

Differentiable on an open set: analyticity

A function differentiable on an open neighborhood of is analytic at . The big payoffs of complex analysis — Cauchy’s theorem, the Cauchy integral formula, the residue theorem — are statements about analytic functions, not merely differentiable ones. See Analytic function.