the symmetric form of Rouché's theorem

The usual Rouche theorem treats one function as the boss and the other as a small correction. But sometimes neither dominates cleanly, and you only know the two functions never point in exactly opposite directions on the boundary. The symmetric form captures precisely that weaker, more even-handed condition — and it turns out to be all you need.

The statement: let f and g be holomorphic on and inside a simple closed contour gamma, with neither vanishing on gamma, and suppose |f(z) - g(z)| is strictly less than |f(z)| + |g(z)| at every point of gamma. Then f and g have the same number of zeros inside gamma, counted with multiplicity. The geometric meaning of that inequality is sharp: |f - g| < |f| + |g| fails (becomes an equality) exactly when f and g point in opposite directions, that is when g / f is a negative real number. So the hypothesis simply says g / f never lands on the negative real axis along gamma. Then the curve g / f stays in the slit plane (plane minus the negative reals), which is simply connected and avoids 0, so it has winding number 0 about the origin — meaning f and g wind around 0 the same number of times, hence have equal zero counts.

Why prefer this version: the classical hypothesis |g| < |f| is asymmetric and can be annoying to verify when the two functions are comparable in size. The symmetric inequality is gentler and is implied by the classical one (if |g - f| < |f| then certainly |g - f| < |f| + |g|), so it never asks for more and often asks for less. It is the form to reach for when you are comparing two genuine competitors rather than a function and a tiny perturbation. The same honesty applies: the inequality must be strict and hold on the entire contour, and neither function may vanish on gamma.