1: \begin{abstract}

2: 

3: 

4: Hoover and Posch studied a single deterministic compressible time-reversible version of Eberhard Hopf's

5: ``Baker Map''.  This particular version was selected to serve as a ``Toy Model'' of nonequilibrium molecular

6: dynamics.  Iterating the map generates an ergodic fractal distribution with measure everywhere in a square

7: domain. Two versions of the map are isomorphic to a third model, a one-dimensional random walk confined within the unit

8: interval $(0 < y < 1)$ with leftward steps twice as likely as rightward. This correspondence between

9: two-dimensional time-reversible deterministic $(q,p)$ and $(x,y)$ maps and the stochastic bounded

10: one-dimensional $y$-direction walk is used here to characterize discrepancies between three different

11: formulations of the fractal information dimension $D_I$ of the compressible Baker Map.

12: 

13: We compare the Kaplan-Yorke conjecture, believed valid for these models, $D_I\stackrel{?}{=} D_{KY}$, to

14: brute-force computations of either a few, or as many as a trillion iterations of the one- and

15: two-dimensional Maps. The $(q,p)$ Baker Mapping contains square roots. These irrational numbers provide

16: sufficient detail, or ``noise'',

17: to avoid the periodic orbits that would otherwise short-circuit the convergence of the Baker Map to the steady

18: state solutions found here. We find that the Kaplan-Yorke Lyapunov dimension, $D_{KY}$, accepted as correct

19: for simple maps like ours based on the mapping of smooth density functions, differs from the simplest of several

20: information dimension estimates.  The latter estimate is based on repeated mappings of an initial point, as

21: measured on a series of meshes ranging from (1/3) in width down to $(1/3)^{19}$.  Other families of meshes support

22: the conventional wisdom that the small-mesh limiting information dimension is given {\it exactly} by the

23: 40-year-old Kaplan-Yorke ``conjecture''.

24: 

25: \end{abstract}

26: