20.07.2013 Views

Gauge theory for embedded surfaces, II

Gauge theory for embedded surfaces, II

Gauge theory for embedded surfaces, II

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Gauge</strong> <strong>theory</strong> <strong>for</strong> <strong>embedded</strong> <strong>surfaces</strong>, <strong>II</strong> 15<br />

on the sign of the self-intersection number n; we will assume hence<strong>for</strong>th that n<br />

is positive and will write the <strong>for</strong>mula as<br />

dim Mκ(X o ; R s )=8κ−3(b + (X) − b 1 (X)+1)+η(R). (3.6)<br />

We choose to express this in terms of the invariants of the closed manifold X.<br />

Note also that 8κ is not necessarily an integer. The term η(R) involvesanηinvariant<br />

of the boundary data as well as the genus and self-intersection number<br />

of the surface. To use the dimension <strong>for</strong>mula, we need to know this term:<br />

Proposition 3.7. For positive self-intersection number n, thetermη(R)in the<br />

dimension <strong>for</strong>mula is given by<br />

(a) η(R+)=0;<br />

(b) η(R−)=0;<br />

(c) η(Rm)=−(8m2 /n − 4m +2g−3). ⊓⊔<br />

Remarks. These <strong>for</strong>mulae, or rather, the equivalent calculations <strong>for</strong> the index<br />

of the Kuranishi complex with exponential decay, are essentially worked out<br />

in [T] (section 12), though the results are stated only modulo 8. The mod 8<br />

ambiguity can be resolved by going through the proofs, though in the case of (c)<br />

the proof is only briefly discussed. We shall give an independent verification of<br />

(c) in subsection (vi) by using the gluing theorem (described below) to relate the<br />

dimension <strong>for</strong>mula here to the dimension <strong>for</strong>mula <strong>for</strong> the twisted moduli spaces<br />

M α k,l . An alternative treatment is given in [MMR].<br />

It is helpful to think of the term η(R) as the ‘efficiency’ of the boundary<br />

condition R, in that a larger value of η means a larger dimension of the moduli<br />

space <strong>for</strong> given action. In these terms, the next lemma gives the condition under<br />

which the components R+ and R− are more efficient than all the components<br />

Rm.<br />

Lemma 3.8. If the genus g and self-intersection number n satisfy the inequality<br />

2g − 2 > 1<br />

2n +1,withn>0,thenη(Rm)is negative <strong>for</strong> all m. Ifthestronger<br />

inequality 2g − 2 > 1<br />

2 n +2 holds, then η(Rm) is always less than −1.<br />

Proof. We only have to look at the expression in part (c) of (3.7) as a quadratic<br />

function of m and calculate the discriminant b 2 − 4ac,<br />

4 2 − 4·(8/n)(2g − 3) = 16<br />

(n − 4g +6)<br />

n<br />

= 32<br />

n<br />

1 ( 2n +1−(2g − 2)),<br />

to see that η(Rm) is always negative under the stated condition. The rider to<br />

the lemma is an immediate corollary. Note that η(Rm) achieves its maximum<br />

when m = n/4, which is in the middle of its range of validity. ⊓⊔

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

Saved successfully!

Ooh no, something went wrong!