Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
U.
In other words, the sets of + together cover S, or each point of S belongs to at least one of the sets of + . A subcovering of S is then a subcollection of + that still covers S. More precisely, * is a subcovering of S if * + and S
U *
U.
Heine-Borel Theorem Any open covering of a closed and bounded set in En has a finite subcovering. For a proof see a book on Real Analysis or a book on elementary Topology. Topology books take this covering property as the definition of a compact set, but for M337 this Theorem can be stated as: Any open covering + of a compact set K in ? has a finite subcovering ., that is, . is a finite collection of sets from + that cover K. Simple examples 1. The collection of open discs + = {Da, b : a, b } , where Da, b = {z : |z (a + ib)| < 1} is an open covering of the plane ? , but no finite number of these open discs cover ? . Of course, ? is not compact as it is not bounded. 2. The collection of open rectangles + = {Rn : n } , where Rn = {z : 1 / n < Re (z) < 3 / n and 1 < Im (z) < 2} is an open covering of the open unit square S = {z : 0 < Re (z) < 1 and 0 < Im (z) < 1} , but no finite number of the rectangles cover S. This time S is not compact because it is not closed. 3. If we add the open rectangle R0 = {z : 1 < Re (z) < 1 / 106 and 1 < Im (z) < 2} to the collection in example 2, then the new collection + = {Rn : n and n = 0} covers the closure of S, S = {z : 0 Re (z) 1 and 0 Im (z) 1} which is compact. There is a finite subcover of S consisting of 1,000,001 of the rectangles, but no fewer will cover. 4. An open covering of a non-compact set may have a finite subcovering as, for example, the finite subcover in the previous example also covers S ; but for a non-compact set there will always be some open covering of the set that has no finite subcovering. 1
M337 2013
, if and only if it is a Theorem A function is meromorphic in the extended complex plane, ? rational function. Proof. Clearly, a rational function is meromorphic (see above and Section 3.1 of Unit D1). . Since f either has a pole at or is analytic at Conversely, let f be a meromorphic function in ? , there exists a punctured open disc at on which f is differentiable. So, for some R > 0 ,
{z ? : |z| > R} . f is differentiable on [By the definition and method on p12 in Section 1.3, Unit D1, g (w) = f (1 / w) is differentiable on a punctured open disc centred at 0 , say {w ? : 0 < |w| < r}, and this corresponds to the above punctured open disc at by taking R = 1 / r.] So, all the poles of f , other than any pole at , are in the closed disc
K = {z ? : |z| R} . Since poles are isolated singularities, for each pole K , there exists r > 0 such that the open disc D = {z ? : |z| < r} contains no other poles. Also, for each point K at which f is analytic, there exists r > 0 such that f is differentiable on the open disc D = {z ? : |z| < r} and so D contains no poles. The collection of all these open discs, D and D , is an open covering of K. Since K is closed and bounded, it is compact, and so a finite collection . of these discs covers K. As none of the D contain any of the poles and each D contains only the pole , . must include all the D . Hence there are only a finite number of poles. Let these poles be 1 , 2 , . . . . . . . . , n and let their orders be m1 , m2 , . . . . . . . . , mn , respectively. Define the function h (z) = (z 1)m1 (z 2)m2 . . . . . . . . . (z n)mn f (z) . By Theorem 3.2 (A) and (B) of Unit B4, the function h (z) is differentiable on ? (after removing the removable singularities at the i ) and so analytic on ?. By Taylors Theorem and Corollary, in Unit B3, h (z) has a Taylor Series valid on ? ,
k=0
akzk ,
for all z ?.
p (z) = (z 1)m1 (z 2)m2 . . . . . . . . . (z n)mn has a pole of order M = m1 + m2 + . . . . . . . . . . + mn at , because the function
m1 m2 mn 1 1 1 1 q (w) = p = 1 2 . . . . . . . . . n w w w w has largest negative power term wM and so a pole of order M at 0. Let l 0 be the order of any pole of f at , where l = 0 covers the case of no pole at . Then h (z) has a pole of order N = M + l at . Now
() ( ()
)(
1 h = w
k=0
akwk ,
and, as this has a pole of order N at 0 , ak = 0 , for all k > N . This means that h (z) =
k=0
akzk ,
for all z ? ,
is a polynomial. Consequently, f (z) = h (z) / p (z) is a rational function. I A function meromorphic only in ? may not be rational. For example, the only singularities of f (z) = 1 / sin z in ? are poles at k , for all k , but, being infinite in number, f cannot be a rational function. In fact f has an essential singularity at (consider g (w) = 1 / sin (1 / w) at 0). 2