JOVANA
Explore Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

Continuation Along a Path and Monodromy

Carry a function element along a road, disk by disk, and it travels with you. But two roads to the same town can deliver two different functions — and a loop can hand you back a stranger. The monodromy theorem says exactly when this cannot happen.

Walking a function element down a road

In the previous guide you met analytic continuation across overlapping disks: a power series convergent near z_0 agrees with a second series convergent near a nearby point, and where their disks overlap they must coincide, by the identity theorem. That pairing of a center with its convergent series is a function element — a tiny, fully self-contained piece of a holomorphic function, living on one disk. Now we set that element in motion. Pick a path, a continuous road gamma(t) from a start point a to an end point b, and try to carry the element all the way along it.

Here is the mechanism, and it is wonderfully concrete. Cover the road with a chain of overlapping disks D_0, D_1, ..., D_n: D_0 is centered at a and carries your starting element, D_n is centered at b, and each consecutive pair overlaps in a region the road passes through. On D_0 you have a series. Re-expand it about a point of the overlap that lies inside D_1 — that gives a new series, the element on D_1. Re-expand again into D_2, and so on, stepping the element forward one disk at a time until you arrive at b. This disk-by-disk relay is exactly what we call continuation along a path.

a = c_0 --- c_1 --- c_2 --- ... --- c_n = b   (centers along the road)

   ( D_0 )
        ( D_1 )
             ( D_2 )                  ... overlapping disks
                  ...      ( D_n )

element on D_0  ->  re-expand in overlap  ->  element on D_1  ->  ...  ->  element on D_n

The FINAL element on D_n is "the continuation of f along gamma".
Each disk overlaps the next; re-expanding the series in the overlap hands the element forward, one step at a time, all the way to b.

The path matters: two roads, two answers

Now the surprise that gives this whole subject its drama. The continuation you arrive at depends not just on where you start and where you stop, but on which road you took. Carry the same starting element from a to b along one path and you may land on one function element; carry it along a different path to the same b and you may land on a genuinely different element. The town is the same; the visitor you deliver is not. This is not a defect of the method — it is the method faithfully reporting something real about the function.

The cleanest example is the logarithm. Start near z = 1 with the element log z = 0 there, expanding the principal branch. Now carry that element once counterclockwise around the origin, along a loop that returns to z = 1. At each step you re-expand honestly, and the value of arg z creeps upward by a little on every disk. By the time the road closes back at z = 1, the argument has accumulated a full 2 pi, so the element you arrive with reads log 1 = 2 pi i, not 0. You walked a closed loop, came home, and came home with a different function. This is the headline meaning of the branch obtained around a loop.

Why does the loop change the answer here but not, say, for the function 1/z, which you can continue around the origin and recover unchanged? Because log z has no single-valued holomorphic extension to a punctured disk around 0 — it is intrinsically multivalued, and circling its branch point at the origin shifts you from one sheet of its values to the next. The path-dependence is the function telling you, through the relay, that the origin is special. Continuation never lies; it simply records the topology of where the singularities sit.

What stays fixed: small wiggles do not matter

Before the answer can depend on the path in any meaningful way, we need to know it does not depend on tiny, irrelevant details — how exactly you chose the disks, how finely you chopped the road. And indeed it does not. Two different chains of disks covering the same path produce the same final element at b. The proof is pure uniqueness of continuation: wherever two relays both have elements, those elements agree on the overlaps, and the identity theorem nails them together. So continuation along a fixed path is well-defined; the only thing that can change the outcome is changing the path itself.

Stronger still: you can deform the path continuously and the answer stays put — as long as you never drag it across a point where the continuation fails. If you slide gamma over to a nearby path gamma' by a small, continuous push, keeping both endpoints fixed and keeping the road inside the region where everything continues, the final element at b is unchanged. This is the same spirit as the homotopy invariance you met for contour integrals: continuous deformation through good territory leaves the result alone. Loosen a road; nothing breaks. Only when you are forced to sweep across a forbidden point does the value get the chance to jump.

The monodromy theorem

We can now state the result that organizes the whole picture. The monodromy theorem says: if a function element can be continued along every path in a region, and that region is simply connected (it has no holes — every loop can be shrunk to a point inside it), then the continuation gives a single, well-defined holomorphic function on the entire region. No matter which road you take to a point, you arrive at the same value. The multivaluedness simply cannot appear. The word "monodromy" comes from Greek for "running around once"; the theorem is precisely the statement that running around a loop in a hole-free region brings you back unchanged.

  1. Check the element can be continued along every path in the region — at no point along any road does the radius shrink to zero or a singularity block the way.
  2. Check the region is simply connected: every closed loop in it can be continuously shrunk to a point without leaving the region (no punctures, no holes).
  3. Take any loop. By homotopy invariance it can be shrunk to a point, and continuation around a tiny loop changes nothing — so continuation around the full loop also changes nothing.
  4. Conclude: the value at each point is independent of the path used to reach it, so the continuation patches together into one single-valued holomorphic function on the whole region.

Look back at the logarithm with this in hand and the mystery dissolves. The plane with the origin removed is not simply connected — the loop around 0 cannot be shrunk to a point without passing through the puncture. That single hole is precisely the loophole the monodromy theorem closes everywhere else, and it is exactly where log z escapes to a new branch. Cut the plane along a ray from the origin and you forbid the encircling loop; the slit region becomes simply connected, monodromy applies, and the principal logarithm becomes honestly single-valued there.

Sheets, and the promise of a better domain

Step back and see what the path-dependence is really telling us. Each time you loop the logarithm around the origin, you do not get nonsense — you get the next perfectly good branch, with arg shifted by another 2 pi. Loop again, another 2 pi. Loop the other way, subtract 2 pi. The branches are stacked in an orderly tower, each a single-valued sheet, and the loop is an elevator that moves you between floors. Collecting all of these elements together — every branch you can ever reach by every path — assembles what is called the complete analytic function: the function in its full, multivalued glory, holding all its sheets at once.

This is the doorway to the rest of the rung. If a single flat plane is too small to hold a single-valued logarithm — because one loop sends you to a different floor — then perhaps the honest domain is not the plane at all, but a surface built by stacking those floors and connecting them where the loops cross over. On that larger surface the elevator becomes an ordinary staircase, every value sits at exactly one address, and the function is single-valued again. Building that surface, and seeing the square root and the logarithm finally tamed on it, is exactly what the Riemann-surface guides ahead will do.

One honest caveat to carry forward. The monodromy theorem demands that the element continue along every path in the region — that hypothesis is doing real work and cannot be dropped. If even one obstruction lurks inside an otherwise simply-connected region (a point where the continuation simply fails), the conclusion can collapse. And there are functions that cannot be continued past a certain curve at all, no matter the path: their domain ends in a wall. That phenomenon, the natural boundary, is where the next guide picks up, asking what happens when continuation does not merely change the answer, but stops dead.