1: \begin{abstract}

2: 

3: The advection-diffusion equation is studied via a global Lagrangian

4: coordinate transformation. The metric tensor of the Lagrangian coordinates

5: couples the dynamical system theory rigorously into the solution of this

6: class of partial differential equations. 

7: If the flow has chaotic streamlines, the diffusion will dominate the 

8: solution at a critical time, which scales logarithmically with the 

9: diffusivity. The subsequent rapid diffusive relaxation is completed

10: on the order of a few Lyapunov times, 

11: and it becomes more anisotropic the smaller the diffusivity. 

12: The local Lyapunov time of 

13: the flow is the inverse of the finite time Lyapunov exponent.

14: A finite time Lyapunov exponent can be expressed in terms of two

15: convergence functions which

16: are responsible for the spatio-temporal complexity of both the advective

17: and diffusive transports.

18: This complexity gives a new class of diffusion barrier in the

19: chaotic region and a fractal-like behavior in both space and time.

20: In an integrable flow with shear, there also exist fast and slow diffusion.

21: But unlike that in a chaotic flow, a large gradient of the scalar field

22: across the KAM surfaces can be maintained since the fast diffusion 

23: in an integrable flow

24: is strictly confined within the KAM surfaces.   

25:  

26: \end{abstract}

27: