Meromorphic Functions

One small permission, granted carefully

By now you can read any isolated singularity off its Laurent series and sort it into one of three boxes: a removable singularity (no negative powers — the function was secretly fine), a pole (finitely many negative powers — it blows up to infinity in a clean, controlled way), or an essential singularity (infinitely many negative powers — pure chaos). A holomorphic function on an open set has none of these; it is singularity-free by definition. The idea of this guide is to relax that purity by exactly one notch.

Here is the relaxation. A function f is meromorphic on a region if it is holomorphic on that region except at a set of isolated points, and at each of those points the singularity is a pole — never essential. Removable singularities do not even count: we simply fill them in and forget them. So a meromorphic function is 'holomorphic except for some poles.' The name says it: 'mero' from the Greek for 'part,' as in part-holomorphic, in deliberate contrast to 'holo,' whole.

Locally a quotient: holomorphic over holomorphic

There is a second way to picture a meromorphic function that is often more useful than the definition: near every point, it looks like one holomorphic function divided by another. To see why, recall how zeros and poles trade places under reciprocation — the lesson of the previous guide. If a holomorphic g has a zero of order m at z_0, then 1/g has a pole of order m there, and vice versa. So a quotient g/h, with g and h holomorphic and h not identically zero, is holomorphic wherever h is nonzero and has a pole exactly where h has a zero (unless g cancels it). That is precisely the meromorphic shape.

f(z) = g(z) / h(z)        g, h holomorphic, h not identically 0

  h(z_0) != 0   ->  f holomorphic at z_0
  h has zero of order m at z_0, g(z_0) != 0
                ->  f has a POLE of order m at z_0

  example:  f(z) = (z + 3) / (z - 1)^2
            pole of order 2 at z = 1,  zero at z = -3

This 'quotient' picture immediately populates the world with examples. Every rational function p(z)/q(z) — a polynomial over a polynomial — is meromorphic on the whole plane, with poles at the roots of q. The tangent tan z = sin z / cos z is meromorphic, with poles at the odd multiples of pi/2 where cos z vanishes. The function 1/sin z is meromorphic, with simple poles at every integer multiple of pi. The Gamma function and the Riemann zeta function are meromorphic too. The class is enormous and contains nearly every named function you will meet.

Why the class is so comfortable to live in

Holomorphic functions are closed under addition and multiplication but NOT under division — divide by one and you may manufacture poles, which escapes the holomorphic world. Meromorphic functions repair exactly that gap. Add two meromorphic functions and you get a meromorphic function; multiply them, same; and now divide one by another (as long as the bottom is not identically zero) and you STILL get a meromorphic function. The dividing zeros just become new poles, which are allowed. In algebraic language, the meromorphic functions on a region form a field — the smallest closed, self-sufficient world containing the holomorphic ones.

Reading the values at a pole: the principal part

Even at a pole, a meromorphic function is not truly undefined — it has a perfectly definite value, namely infinity, in a sense that becomes literal if you adopt the Riemann sphere. There the point at infinity is an honest point, and a function with a pole of order m at z_0 maps z_0 smoothly to that point. From this viewpoint a meromorphic function is just a holomorphic map into the sphere instead of the plane: at a pole it does not 'break,' it simply lands on the one extra point that the plane was missing. The pole is a feature, not a failure.

Globally, a meromorphic function carries a beautiful bookkeeping near each pole, packaged in its principal part — the finite tail of negative-power terms in the Laurent series there. Knowing every pole together with its principal part captures all the singular information of f at once. For nice meromorphic functions you can even reassemble the whole function from those pieces: a rational function is literally the sum of its principal parts plus a polynomial, which is the elementary partial-fraction expansion you met long ago, now revealed as a statement about poles.

1. Find every pole of f and its order — these are the only places f leaves the holomorphic world.
2. At each pole z_k, read off the principal part: the negative-power terms b_1/(z - z_k) + b_2/(z - z_k)^2 + ... up to the order of the pole.
3. Subtract all the principal parts from f; what remains has no poles left, so it is holomorphic (entire, if f was meromorphic on the whole plane).
4. For a rational function that holomorphic remainder is just a polynomial — add it back and you have rebuilt f from its poles: the partial-fraction expansion.

What meromorphy lets you count

Because a meromorphic function has only isolated zeros and isolated poles, each of a finite order, you can attach an integer to every point: positive at a zero (its order), negative at a pole (minus its order), zero everywhere else. This single tally of zeros-minus-poles turns out to control an astonishing amount. It is the heart of the residue calculus, and through the argument principle of a later rung, an integral of f'/f around a loop literally counts (zeros inside) minus (poles inside). The good behaviour 'only poles, all of finite order' is exactly what makes this counting work.

There is even a converse craftsmanship: you can BUILD a meromorphic function with poles prescribed wherever you like, with whatever principal parts you choose, as long as the poles do not accumulate at a finite point. That is the content of the Mittag-Leffler theorem, the additive cousin of the partial-fraction idea. So meromorphic functions are not only abundant — they are buildable to order, a designed object whose singular skeleton you get to specify in advance.