Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
We can pass to local boundary conditions by considering the map<br />
v :Ê×[0, 1] → T ∗ (Q × Q) ⊂�2N , v(z) := (u(z/2), u(z/2 + i)).<br />
The map v satisfies the boundary conditions<br />
Moreover,<br />
v(s, 0) ∈ N ∗ Qj ⊂ N ∗ Vj, if s ∈ [2sj−1, 2sj], v(s, 1) ∈ N ∗ ∆Q ⊂ N ∗ ∆ÊN, ∀s ∈Ê. (150)<br />
∂v(z) = 1<br />
�<br />
�<br />
∂u(z/2), ∂(z/2 + i) ,<br />
2<br />
so by (149) and by the fact that XH(t, q, p) has quadratic growth in |p| by (29), there is a constant<br />
c such that<br />
|∂v(z)| ≤ c(1 + |v(z)| 2 ). (151)<br />
Let χ be a smooth function such that χ(s) = 1 for s ∈ [0, 1], χ(s) = 0 outside [−1, 2], and<br />
0 ≤ χ ≤ 1. Given h ∈�set<br />
w(s, t) := χ(s − h)v(s, t).<br />
Fix some p > 2, and consider the norm � · �X p introduced in section 5.3, with S = {2s1, . . .,2sk}.<br />
The map w has compact support and satisfies the boundary conditions (150), so by Proposition<br />
5.10 we have the elliptic estimate<br />
Since<br />
we obtain, together with (151),<br />
�∇w�Xp ≤ c0�w�X p + c1�∂w�X p.<br />
∂w = χ ′ (s − h)v + χ(s − h)∂v = χ′<br />
w + χ(s − h)∂v,<br />
χ<br />
�∇w�X p ≤ (c0 + c1�χ ′ /χ�∞) �w�X p + c1�χ(· − h)∂v�X p<br />
≤ (c0 + c1�χ ′ /χ�∞) �w�X p + c1c�χ(· − h)(1 + |v| 2 )�X p.<br />
Therefore, we have an estimate of the form<br />
�∇w�Xp ≤ a�w�Xp + b�χ(· − h)(1 + |v|2 )�Xp. (152)<br />
Since w has support in the set [h − 1, h + 2] × [0, 1], we can estimate its Xp norm in terms of its<br />
X1,2 norm, by Proposition 5.13. The X1,2 norm is equivalent to the W 1,2 norm, and the latter<br />
norm is bounded by Lemma 6.1. We conclude that �w�Xp is uniformly bounded. Similarly, the<br />
Xp norm of χ(· − h)(1 + |v| 2 ) is controlled by its W 1,2 norm, which is also bounded because of<br />
Lemma 6.1. Therefore, (152) implies that w is uniformly bounded in X1,p . Since p > 2, we deduce<br />
that w is uniformly bounded in L∞ . The integer h was arbitrary, hence we conclude that v is<br />
uniformly bounded in L∞ , and so is u.<br />
Let us explain how all the solution spaces M considered in this paper can be viewed in terms<br />
of the above general setting. We describe explicitly the reduction in the case of the spaces M Λ Υ<br />
associated to the pair-of-pants product as described in sections 3.2 and 3.3, the argument being<br />
analogous for all the other solution spaces. The pair-of-pants Riemann surface Σ Λ Υ<br />
is described<br />
as the quotient of the disjoint union of two stripsÊ∪[−1, 0] andÊ×[0, 1] with respect to the<br />
identifications<br />
(s, −1) ∼ (s, 0−), (s, 0+) ∼ (s, 1) ∀s ≤ 0, (153)<br />
(s, −1) ∼ (s, 1), (s, 0−) ∼ (s, 0+) ∀s ≥ 0. (154)<br />
89