23.11.2012 Views

Discrete Holomorphic Local Dynamical Systems

Discrete Holomorphic Local Dynamical Systems

Discrete Holomorphic Local Dynamical Systems

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>Discrete</strong> <strong>Holomorphic</strong> <strong>Local</strong> <strong>Dynamical</strong> <strong>Systems</strong> 37<br />

Theorem 5.17 (Brjuno, 1971 [Brj2–3]). Let f ∈ End(C n ,O) be a holomorphic<br />

local dynamical system such that f belongs to the Siegel domain, has no resonances,<br />

and d fO is diagonalizable. Assume moreover that<br />

+∞<br />

∑<br />

k=0<br />

1<br />

log<br />

2k Then f is holomorphically linearizable.<br />

1<br />

Ω f (2 k+1 )<br />

< +∞. (37)<br />

Theorem 5.18. Let λ1,...,λn ∈ C be without resonances and such that<br />

limsup<br />

m→+∞<br />

1<br />

m log<br />

1<br />

Ω λ1,...,λn<br />

(m) =+∞.<br />

Then there exists f ∈ End(C n ,O), with d fO = diag(λ1,...,λn), not holomorphically<br />

linearizable.<br />

Remark 5.19. These theorems hold even without hyperbolicity assumptions.<br />

Remark 5.20. It should be remarked that, contrarily to the one-dimensional case, it is<br />

not yet known whether condition (37) is necessary for the holomorphic linearizability<br />

of all holomorphic local dynamical systems with a given linear part belonging to<br />

the Siegel domain. However, it is easy to check that if λ ∈ S 1 does not satisfy the<br />

one-dimensional Brjuno condition then any f ∈ End(C n ,O) of the form<br />

f (z)= � λ z1 + z 2 1 ,g(z)�<br />

is not holomorphically linearizable: indeed, if ϕ ∈ End(C n ,O) is a holomorphic linearization<br />

of f ,thenψ(ζ)=ϕ(ζ,O) is a holomorphic linearization of the quadratic<br />

polynomial λ z + z 2 ,againstTheorem4.10.<br />

Pöschel [Pö] shows how to modify (36) and(37) to get partial linearization results<br />

along submanifolds, and Raissy [R1](seealso[Ro1]and[R2]) explains when<br />

it is possible to pass from a partial linearization to a complete linearization even in<br />

presence of resonances. Another kind of partial linearization results, namely “linearization<br />

modulo an ideal”, can be found in [Sto]. Russmann [Rü] and Raissy [R4]<br />

proved that in Theorem 5.17 one can replace the hypothesis “no resonances” by the<br />

hypothesis “formally linearizable”, up to define Ω f (m) by taking the minimum only<br />

over the non-resonant multiindeces. See also and Il’yachenko [I1] for an important<br />

result related to Theorem 5.18. Raissy, in [R3], describes a completely different<br />

way of proving the convergence of Poincaré-Dulac normal forms, based on torus<br />

actions and allowing a detailed study of torsion phenomena. Finally, in [DG] results<br />

in the spirit of Theorem 5.17 are discussed without assuming that the differential is<br />

diagonalizable.

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

Saved successfully!

Ooh no, something went wrong!