Poincaré invariants are Seiberg-Witten invariants (Q1954327)

By the results of Donaldson, Li, and Morgan it is known that the famous Donaldson invariants for an algebraic surface \(X\) are the intersection numbers over the moduli spaces of rank 2 Gieseker semi stable sheaves on \(X\). On the other hand, when \(X\) is a Kähler surface, by the result of Kronheimer and Mrowka together with Witten's conjecture the Donaldson invariants are expressed in terms of the Seiberg-Witten invariants. An algebro-geometric approach to Seiberg-Witten theory is known in the special case \(b_1(X)=0\). Furthermore, in this case Witten's conjecture has been proven by Göttsche, Nakajima, and Yoshioka by using Mochizuki's formula. In a strive for giving an algebro-geometric definition for the Seiberg-Witten invariants in general, Dürr, Kabanov, and Okonek constructed a perfect obstruction theory on the Hilbert scheme of divisors on a smooth projective surface \(X\) and obtained a virtual fundamental class. This allowed them to define the so called Poincaré invariants of \(X\) by integrating cohomology classes against the virtual fundamental class, and they conjectured that the Poincaré invariants coincide with the Seiberg-Witten invariants of \(X\). They reduced the proof of their conjecture to proving that for a minimal surface \(X\) of general type with \(p_g(X) >0\) the degree of the virtual fundamental class of the Hilbert scheme of the canonical divisors is -1 to the power the holomorphic Euler characteristics of \(X\). The main result of the paper under review proves that Dürr-Kabanov-Okonek's conjecture is true by proving the last statement above. The main technique of the proof is the cosection localization principal of Kiem and Li. In the future, the authors of the paper under review plan to prove Witten's conjecture in general by using Mochizuki's formula.