statistical monte carlo

Hey,

  I want to do a statistical corner analysis using  'statcrolles', which I can see in my corner file i.e. corners.scs. I am aware that this file is loaded in the model libraries or in the corners-section in ADE XL.

 for this,

1. Should I enable both process data and mismatch data or only process data?

2. What sampling method should i use- random or Latin hypercube or Low-discrepancy sequence ?

3. Should I Specify instance/devices ? I am using a sub-circuit and I want the inter-transistor mismatch for the corners statistically.

4. Also I would like to know, that in ELDO there are step to do monte carlo- statistical from .pcsf (process corner setup file) and .dcsf (design corner setup file). This allows the user to do statistical monte carlo in the ADE L itself. Will these results be same when done in ADE XL ? The kit came with the feasibility of Monte carlo in Spectre (ADE XL) and in ELDO (ADE L).

5. The standard deviation of my Monte Carlo results is low. I am afraid, if I am doing something wrong. Any guess on the mistake I am doing ?

thanks in advance.