14.08.2013 Views

View PDF - Project Euclid

View PDF - Project Euclid

View PDF - Project Euclid

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CURVATURE AND COMPLEX SINGULARITIES 509<br />

(iii) We will conclude with some observations about curvature integrals in<br />

the real and complex cases. In real differential geometry one generally encounters<br />

two types of curvature integrals, illustrated for a Riemannian surface by<br />

the two expressions<br />

(5.28) Is KdA, .ts IK]dA<br />

In general the first type includes characteristic classes and Weyl’s coefficients<br />

(1.10), and reflects intrinsic metric properties of the manifold which may even<br />

be of a topological nature. The second type usually describes extrinsic properties<br />

of the manifold in <strong>Euclid</strong>ean space. For a complex manifold M, C<br />

because of sign properties such as<br />

(_ 1)kcc(t) / n-k _> 0<br />

the distinction between the two types of curvature integrals illustrated in (5.28)<br />

seems to disappear, and the situation may be said either to be simpler or more<br />

rigid, depending on one’s viewpoint.<br />

Finally, along similar lines it would seem interesting to try and draw conclusions<br />

on the curvature of real algebraic curves acquiring a singularity. For<br />

example, one might examine the real curves Ct,IR IR C where Ct c is<br />

defined by f(x, y) with f(x, y) being a weighted homogeneous polynomial<br />

having real coefficients. Superficial considerations suggest that the curvature of<br />

Ct,R in IR tends to , while on the Riemann surface Ct the geodesic curvature<br />

of the (purely imaginary) vanishing cycles remains bounded.<br />

(iv) This paragraph is an afterthought. Upon reviewing the preceeding discussion<br />

about "topologically invariant curvature integrals associated to a singularity"<br />

it seems to me that some clarification is desirable.<br />

To begin with let us consider the pedogogical question of how one might best<br />

prove the Gauss-Bonnet theorem<br />

(5.29) ct-e l KdM x(M)<br />

in a course on differential geometry. Chern’s intrinsic proof (cf. footnote (8) in<br />

section 1) is probably the quickest but in my experience leaves some mystery<br />

as to the origin of his formulas. The Allendoerfer-Weil proof (cf. footnote (5) in<br />

section 1) is intuitively appealing and explains the origin of the formula, but<br />

relies on either the Nash embedding theorem or an unpleasant construction to<br />

show that the left hand side of (5.29) is independent of the metric. Finally, the<br />

proof by characteristic classes, interpreting the Gauss-Bonnet integrand as a<br />

de Rham representative of the Euler class of the tangent bundle, is conceptually<br />

satisfying but involves establishing a fairly elaborate machine. The following<br />

proof, in four steps, represents a compromise"<br />

a) Prove (5.29) for oriented real hypersurfaces using the Hopf theorem, as

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!