MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm working on a project and I've used the Picard-Fuchs equation at a maximally unipotent monodromy point for a certain 1-dimensional family of Calabi-Yau 3-folds to calculate the A-model Yukawa coupling, and thus the generating function of the genus 0 GW invariants on the suspected mirror. I'm wondering if it's possible to have negative GW invariants even in genus 0. Also is it possible to have $n_d=0$ for all odd $d$, where the $n_d$ are defined to be the unique rational numbers fitting into the formula $N_l=\sum_{d|l} n_{l/d}d^{-3}$, where $N_d=I_{0,0,d}$ is the $d$-th genus 0, unpointed, GW invariant in degree d on a Calabi-Yau 3-fold with Picard rank 1. These numbers are also known as the instanton numbers.

To be more precise about my exact question: Suppose you have a CY 3-fold with Picard rank 1, so you can write any cohomology class in codim 2 as $dH$. Then we can write the GW invariants $$N_d=I_{0,0,dH}=\sum_{k|d} n_{d/k} k^{-3}.$$ Is it possible for these $n_{d/k}$'s to be negative for a few low degree terms, and zero for all odd degrees.

share|cite|improve this question
up vote 5 down vote accepted

It is certainly possible to have $n_d$ be negative or zero. For example if $X=K_{\mathbb{P}^2}$, the total space of the canonical bundle over $\mathbb{P}^2$, then $n_2=-6$. Your $n_d$ are genus 0 Gopakumar-Vafa invariants (or BPS numbers) and they have a description in terms of sheaves on $X$ which is (conjecturally) equivalent to your formula in terms of GW invariants. $n_d$ is given as the topological Euler characteristic (weighted by the Behrend function) of the moduli space of semi-stable sheaves $F$ on $X$ whose support is in the curve class dual to "$d$" and such that $\chi(F)=1$. The Behrend function is a integer valued constructible function on any scheme over $\mathbb{C}$ which takes the value $(-1)^n$ at non-singular points of dimension $n$. My point is that if the curves of degree $d$ move in a family, then this weighted Euler characteristic might easily be negative or zero. In the case of $X=K_{\mathbb{P}^2}$, $d=2$, the sheaves are given by the structure sheaves of conics in the plane and so the moduli space is a $\mathbb{P}^5$ (whose Behrend function is the constant -1) and so the weighted Euler characteristic is minus the topological euler characteristic of $\mathbb{P}^5$, so $n_2=-6$.

share|cite|improve this answer
the reference for this description of genus zero BPS invariants is the paper by Sheldon Katz: – Jim Bryan Jan 30 '13 at 0:27
Two questions: 1) is your $K_{\mathbb P^2}$ Calabi-Yau? If you mean by the total space of the canonical bundle the scheme $Spec_{\mathbb P^2} Sym(\omgea_{\mathbb P^2})$ then I believe this has canonical bundle $\pi^*\mathcal O_{\mathbb P^2}(-6)$. So did you mean the total space of the line bundle which has sheaf of sections $\omega_{\mathbb P^2}$, which I agree would be Calabi-Yau. 2)Also, aren't the GV invariants conjectured to be nonnegative (for example see p. 19 of, so is that simply not believed anymore, or are they talking about different things? – HNuer Jan 30 '13 at 14:27
The total space of a locally free sheaf $F$ on $X$ is given by $Spec(Sym(F^\vee))$ --- you are missing the dual in your comment above. The invariants $n^0_d$ are only expected to be positive integers when the number of genus 0 curves in the class $d$ is finite. – Jim Bryan Jan 30 '13 at 17:51
This example, often called "local $\mathbf{P}^2$, is extensively studied in both the math and the physics literature. Here is one paper which analyses this example from the mirror symmetry / Picard-Fuchs point of view: Note that even though local $\mathbb{P}^2$ is non-compact, any compact CY3 $X$ which contains a $\mathbb{P}^2$ will have part of its GW theory identical to that of local $\mathbb{P}^2$. – Jim Bryan Jan 30 '13 at 17:51

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.