Let $X$ be a projective variety and let $D$ be a simple normal crossings divisor on $X$
Does $$IH^*(X;\mathbb C)\cong H_{(2)}^*(X\setminus D;\mathbb C)$$ hold true for each Kähler metric on $X\setminus D$? Is there any counterexample?
What about the Fubini-Study metric on $X\setminus D$? (This is a conjecture of Cheeger) What about the case when the Kähler metric on $X\setminus D$ has a conic model?, a cusp model(Is OK), or a combination of these two models?
-
3$\begingroup$ What is a reference for Cheeger's conjecture ? $\endgroup$– BS.Commented Dec 12, 2015 at 9:24
-
1$\begingroup$ conjecture 9 mathunion.org/ICM/ICM1990.1/Main/icm1990.1.0735.0746.ocr.pdf . He said something about conical metric. But it is far than the definition of conical model of Tian-Yau $\endgroup$– user21574Commented Dec 12, 2015 at 18:05
1 Answer
I'm not an expert, but you don't seem to be getting any answers.
First of all, I would be surprised if holds for any Kähler metric (e.g. for one with really bad singularities along $D$) but I don't have a counterexample*. Regarding the case of Fubini-Study metric, for isolated singularities, I believe it was settled positively by Ohsawa: "Cheeger-Goreski [Goresky]-MacPherson's conjecture for the varieties with isolated singularities." Math. Z. 206 (1991). I don't know the status in general. I believe that there was a gap in a later paper by author on this.
There has been a lot of work with other metrics. For example, you can look at work on Zucker's conjecture, which was proved by Looijenga and Saper-Stern. Also for Poincaré type metrics, allowing coefficients in a variation of Hodge structure, there is the work by Cattani-Kaplan-Schmid and Kashiwara-Kawai. This provides the analytic basis for Saito's theory of Hodge modules.
*(added later) Here's a counterexample. Choose a diffeomorphism $f$ between $\mathbb{C}$ and the disk $D$. Pulling back the Poincaré metric gives a Kähler metric such that $\dim H^1_{(2)}(\mathbb{P}-\{\infty\})=\infty$ because $f^*(z^ndz)$ gives an infinite family of harmonic $L^2$ forms. However intersection cohomology is finite dimensional.