Abstract. This paper is a handyman's manual for learning how to resolve the singularities of algebraic varieties defined over a field of characteristic zero by sequences of blowups.Three objectives: Pleasant writing, easy reading, good understanding.One topic: How to prove resolution of singularities in characteristic zero.
Statement to be proven (No-Tech):The solutions of a system of polynomial equations can be parametrized by the points of a manifold.
Statement to be proven (Low-Tech):The zero-set X of finitely many real or complex polynomials in n variables admits a resolution of its singularities (we understand by singularities the points where X fails to be smooth). The resolution is a surjective differentiable map ε from a manifoldX to X which is almost everywhere a diffeomorphism, and which has in addition some nice properties (e.g., it is a composition of especially simple maps which can be explicitly constructed). Said differently, ε parametrizes the zero-set X (see Figure 1).
Figure 1. Singular surface Ding-dong:The zero-set of the equation x 2 + y 2 = (1 − z)z 2 in R 3 can be parametrized by R 2 via (s, t) → (s(1 − s 2 ) · cos t, s(1 − s 2 ) · sin t, 1 − s 2 ). The picture shows the intersection of the Ding-dong with a ball of radius 3. You will agree that such a parametrization is particularly useful, either to produce pictures of X (at least in small dimensions), or to determine geometric and topological properties of X. The huge number of places where resolutions are applied to prove theorems about all types of objects (algebraic varieties, compactifications, diophantine equations, cohomology groups, foliations, separatrices, differential equations, D-modules, distributions, dynamical systems, etc.) shows that the existence of resolutions is really basic to many questions. But it is by no means an easy matter to construct a resolution for a given X.
Puzzle:Here is an elementary problem in combinatorics -the polyhedral game of Hironaka. Finding a winning startegy for it is instrumental for the way singularities will be resolved. Each solution to the game can yield a different method of resolution. The formulation is simple.Given are a finite set of points A in N n , with positive convex hull N in R n (see Figure 2),(0,0) There are two players, P 1 and P 2 . They compete in the following game. Player P 1 starts by choosing a non-empty subset J of {1, . . . , n}. Player P 2 then picks a number j in J.After these "moves", the set A is replaced by the set A obtained from A by substituting the j-th component of vectors α in A by the sum of the components α i with index i in J, say α j → α j = i∈J α i . The other components remain untouched, α k = α k for k = j (see Figure 3). Then set N = conv(A ) + R n ≥0 . The next round starts over again, with N replaced by N : P 1 chooses a subset J of {1, . . . , n}, and player P 2 picks j in J as before. The polyhedron N is replaced by the corresponding polyhedron N . In this way, the game continues.
THE HIRONAKA THEOREM ON RESOLUTION OF SINGULARITIES 325Player P ...