1: \begin{abstract}

2: Quantum systems subject to time periodic fields of  {\em finite} 

3: amplitude $\lambda$ have conventionally been handled either by

4: low order perturbation theory, for $\lambda$ not too large, or

5: by exact diagonalization within a finite basis of $N$ states.

6: An adiabatic limit, as $\lambda$ is switched on arbitrarily

7: slowly, has been assumed.  But the validity of these procedures 

8: seems questionable in view of the fact that, as $N \rightarrow

9: \infty$, the quasienergy spectrum becomes dense, and numerical

10: calculations show an increasing number of weakly avoided 

11: crossings (related in perturbation theory to high order resonances).

12: This paper deals with the highly non-trivial

13: behavior of the solutions in this limit. 

14: The Floquet states, and the associated quasienergies, become highly

15: irregular functions of the amplitude $\lambda$. The

16: mathematical radii of convergence of perturbation theory in $\lambda$ 

17: approach zero. There is no adiabatic limit of the wave functions

18: when $\lambda$ is turned on arbitrarily slowly.  However, the

19: quasienergy becomes independent of time in this limit.  We introduce 

20: a modification of the adiabatic theorem.  We

21: explain why, in spite of the pervasive pathologies of the

22: Floquet states in the limit $N \rightarrow \infty$, the

23: conventional approaches are appropriate in almost all

24: physically interesting situations.

25: 

26: 

27: \end{abstract}

28: