17.01.2013 Views

PDF (1016 kB)

PDF (1016 kB)

PDF (1016 kB)

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

Given periodic orbits x1 ∈ P(H1), x2 ∈ P(H2), and y ∈ P(H1#H2), the space M Λ Υ (x1, x2; y)<br />

consists of maps<br />

u : Σ Λ Υ → T ∗ M,<br />

solving the Floer equation ∂J,H(u) = 0, with asymptotics<br />

lim<br />

s→−∞ u(s, t − 1) = x1(t), lim<br />

s→−∞ u(s, t) = x2(t), lim u(s, 2t − 1) = y(t).<br />

s→+∞<br />

We can associate to a map u : Σ Λ Υ → T ∗ M the map v : Σ :=Ê×[0, 1] → T ∗ M 2 by setting<br />

v(s, t) := � −u(s, −t), u(s, t) � .<br />

The identifications (153) on the left-hand side of the domain of u are translated into the fact that<br />

v(s, t) is 1 periodic in t for s ≤ 0, or equivalently into the nonlocal boundary condition<br />

(v(s, 0), −v(s, 1)) ∈ N ∗ ∆ M 2 ∀s ≤ 0, (155)<br />

where ∆ M 2 denotes the diagonal in M 4 = M 2 × M 2 . The identifications (154) on the right-hand<br />

side of the domain of u are translated into the local boundary conditions<br />

v(s, 0) ∈ N ∗ ∆M, v(s, 1) ∈ N ∗ ∆M ∀s ≥ 0. (156)<br />

The map u solves the Floer equation ∂J,H(u) = 0 if and only if v solves the Floer equation<br />

∂J,K(v) = 0, where K :Ì×T ∗ M 2 →Êis the Hamiltonian<br />

K(t, x1, x2) := H1(−t, −x1) + H2(t, x2) ∀(t, x1, x2) ∈Ì×T ∗ M 2 ,<br />

which satisfies the growth conditions (H1) and (H2). The asymptotic conditions for u are equivalent<br />

to<br />

lim<br />

s→−∞ v(s, t) = (−x1(−t), x2(t)), lim<br />

s→+∞ v(s, t) = � −y((1 − t)/2), y((1 + t)/2) � . (157)<br />

Finally, since<br />

|∇v(s, t)| 2 = |∇u(s, −t)| 2 + |∇u(s, t)| 2 ,<br />

the energy of u equals the energy of v,<br />

�<br />

E(u) := |∇u| 2 �<br />

dsdt =<br />

Σ Λ Υ<br />

Σ<br />

|∇v| 2 dsdt =: E(v).<br />

We conclude that M Λ Υ (x1, x2; y) can be identified with the space of maps v : Σ → T ∗M2 which<br />

solve the Floer equation for the Hamiltonian K, satisfy the boundary conditions (155), (156), and<br />

the asymptotic conditons (157).<br />

By Nash theorem, we can find an isometric embedding M ֒→ÊN for N large enough. The<br />

induced embedding M2 ֒→Ê2N is such that<br />

∆M2 = M 2 ∩ ∆Ê2N, ∆M × ∆M = M 2 ∩ � �<br />

∆ÊN × ∆ÊN .<br />

Note that the linear subspaces ∆Ê2N and ∆ÊN × ∆ÊN are partially orthogonal inÊ4N . Therefore,<br />

the problem M Λ Υ reduces to the above setting, with Q = M2 , k = 1, s1 = 0, Q0 = ∆ M 2,<br />

Q1 = ∆M × ∆M, V0 = ∆Ê2N, and V1 = ∆ÊN × ∆ÊN.<br />

The energy of solutions in M Λ Υ (x1, x2; y) is bounded above from<br />

�H1(x1) +�H2(x2) −�H1#H2(y),<br />

so Proposition 6.2 implies that these solutions have a uniform L ∞ bound.<br />

90

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

Saved successfully!

Ooh no, something went wrong!