1. Argument Principle
A rational function is a quotient of a complex polynomial by a nonzero complex poly- nomial1. In other words, if R(z) is a rational function, then there exist polynomial P (z) and Q(z) in C[z] with Q(z) 6= 0 such that R(z) = P (z)/Q(z). When deg P (z) > deg Q(z), division algorithm implies P (z) = Q(z)g(z) + r(z) for some g(z) r(z) in C[z] such that deg r(z) < deg Q(z) or r(z) = 0. Hence
R(z) = g(z) + r(z) Q(z).
Using division algorithm, we will simply consider rational function R(z) = P (z)/Q(z) with deg P (z) < deg Q(z). Denote n = deg P (z) and m = deg Q(z). Also assume that P (z), Q(z) are relatively prime.
Rational functions define meromorphic functions on C. Assume that p1, · · · , pn and q1, · · · , qm are zeros of P (z) and Q(z) respectively. Then R(z) is holomorphic on U = C \ {q1, · · · , qm} and
R0(z)
R(z) = P0(z)
P (z) −Q0(z)
Q(z), z ∈ U.
Let D be a closed disk containing {p1, · · · , pn} ∪ {q1, · · · , qm} whose boundary C = ∂D is an oriented circle with positive orientation. Then R0(z)/R(z) is defined on C and
Z
C
R0(z) R(z)dz =
Z
C
P0(z) P (z)dz −
Z
C
Q0(z) Q(z)dz.
The function P0(z)/P (z) has the following expansion P0(z)
P (z) = 1 z − p1
+ 1
z − p2
+ · · · + 1 z − pn
.
Let Ci be the small circle with positive orientation center at pi of radius δ > 0 such that Ci are disjoint. Using the property of path integration, we obtain
I
C
P0(z) P (z)dz =
n
X
i=1
Z
Ci
dz z − pi
= 2πin.
Similarly, we have
I
C
Q0(z) Q(z)dz =
m
X
j=1
Z
Cj0
dz
z − qi = 2πim.
where Cj0 is the small circle with positive orientation center at qi of radius δ0 > 0 for 1 ≤ j ≤ m such that Cj0 are disjoint. This implies that
I
C
R0(z)
R(z)dz = 2πi(n − m).
Let f (z) be a nonzero meromorphic function on a region (open connected set) Ω. Let D be a closed disk in Ω whose boundary C is an oriented circle with positive orientation. Let ZC(f ) and PC(f ) be the set of zeros and the set of poles of f (z) lying in the interior D◦ of D. Since D is a compact, f can only have finitely many zeros and poles in D. Both of ZC(f ) and ZP(f ) are finite sets. Let us denote ZC(f ) = {p1, · · · , pr} and PC(f ) = {q1, · · · , qs}
1We only consider rational function over C.
1
2
where pi and qj are all distinct. Denote the order of pi by ni and the order of qj by mj. We construct a rational function R(z) by
R(z) = Qr
i=1(z − pi)ni Qs
j=1(z − qj)mj.
Then the zeros and poles of R(z) lie inside D◦. Define a new function g(z) = f (z)/R(z).
Then pi are removable singularities of g and hence g can be extended to a meromorphic function on Ω such that g is holomorphic in D◦ and g is nonzero on D. Write f (z) = R(z)g(z). Let U be any open subset of Ω such that f is holomorphic. For z ∈ U,
f0(z)
f (z) = R0(z)
R(z) +g0(z) g(z).
Since g is nonzero on D◦, g0(z)/g(z) is holomorphic in D◦. By the previous argument, 1
2πi I
C
f0(z)
f (z)dz = 1 2πi
I
C
R0(z)
R(z)dz + 1 2πi
I
C
g0(z) g(z)dz
= 1 2πi
I
C
R0(z) R(z)dz
= (n1+ · · · + nr) − (m1+ · · · + ms)
Let us denote #ZC(f ) = n1 + · · · + nr and #PC(f ) = m1+ · · · + ms. Then #ZC(f ) and
#PC(f ) are numbers of zeros and numbers of poles in D◦ counting multiplicities.
Theorem 1.1. (Argument Principle) Let f be a meromorphic function on a region Ω. Let D be a closed disk whose boundary C is a oriented circle with positive orientation. Let
#ZC(f ) and #PC(f ) be numbers of zeros and numbers of poles of f in D◦with multiplicities counted. Then
1 2πi
I
C
f0(z)
f (z)dz = #ZC(f ) − #PC(f ).
If f is a holomorphic function on a region Ω, we would like to study the number of solutions to the equation
(1.1) f (z) = ω
in the closed disk D in Ω; here we require that ∂D contains no solution of (1.1). When ω = 0, the numbers of solution to (1.1) can be calculated using argument principle. In fact, if we denote F (z) = f (z) − ω, then (1.1) is equivalent to F (z) = 0. Using argument principle, the number of solutions to F (z) = 0 in D is given by the count our integral
1 2πi
I
C
F0(z)
F (z)dz = 1 2πi
I
C
f0(z) f (z) − ωdz.
2. Generalization of Argument Principle
Let h(z) be a holomorphic function on a region Ω and R(z) be a rational function whose zeros and poles are contained in a closed disk D in Ω. Assume R(z) is nonzero on C and C is equipped with the positive orientation. Then we would like to compute
1 2πi
I
C
h(z)R0(z) R(z)dz.
3
Let R(z) and P (z) and Q(z) be as the previous section. Using Cauchy’s integral formula, 1
2πi I
C
h(z)P0(z) P (z)dz =
n
X
i=1
1 2πi
Z
C
h(z) (z − pi)dz
=
n
X
i=1
h(pi).
When R(z) = P (z)/Q(z), we have 1
2πi I
C
h(z)R0(z)
R(z)dz = 1 2πi
I
C
h(z)P0(z)
P (z)dz + 1 2πi
I
C
h(z)Q0(z) Q(z)dz
=
n
X
i=1
h(pi) −
m
X
j=1
h(qj).
Let f (z) be a meromorphic function on Ω. Write f (z) = R(z)g(z) where R(z) is a rational function and g(z) is a meromorphic function on Ω so that g(z) is holomorphic on D◦ and nonzero on D. Let U be any open subset of Ω so that f (z) is holomorphic on U. For any z ∈ U
h(z)f0(z)
f (z) = h(z)R0(z)
R(z) + h(z)g0(z) g(z). Since h(z)g0(z)/g(z) is holomorphic on D◦, we obtain
1 2πi
I
C
h(z)f0(z)
f (z)dz = 1 2πi
I
C
h(z)R0(z)
R(z)dz + 1 2πi
I
C
h(z)g0(z) g(z)dz
= 1 2πi
I
C
h(z)R0(z) R(z)dz
=
n
X
i=1
h(pi) −
m
X
j=1
h(qj).
Theorem 2.1. Let h be a holomorphic function on a region Ω and f be a meromorphic function on Ω. Let D be a closed disk whose boundary C does not contain any of zeros or poles of f. Let p1, · · · , pn be zeros of f in D and q1, · · · , qm be poles of f in D. Then
1 2πi
I
C
h(z)f0(z) f (z)dz =
n
X
i=1
h(pi) −
m
X
j=1
h(qj).
A univalent function on an open set U is a holomorphic injection f : U → C. For each ω ∈ f (U ), the solution to f (z) = ω is unique. Let us denote the solution by z(ω). Choose a closed disk D such that the point z(ω) is contained in the interior of D. We obtain
z(ω) = 1 2πi
I
C
z f0(z) f (z) − ωdz.
Here C = ∂D.