I have produced a plan for the "American Option Project" with
all the initial steps. I have found several steps that can be done before
we get any legal action from Ann and suggest we start these now.
They fall into two areas
1)Prepare test data and results to compare with.
2)Make Makivic program more effient. This is very relevant as one can
view American Option calculation as a particular and very complex
"measurement". So now measurement totally swamps Monte Carlo path generation
and so one needs totally uncorrelated paths -- which Miloje does not
calculate now but can be straightforwardly done!
I will give Scott these plans friday and we can get started with or
without Ann

As an aside, I note that American Options will be a totally separate
program from Miloje's code. It could be put in as a new measurement
routine but I see many reasons against this. We will generate paths
from Miloje -- certify them as "perfect", store on disk and read back
In particular "American Option" can use a different language such as
Fortran (for high performance) or Java (for the future)

As another aside, I read chapter 15 of Hull. These derivatives fall into 2
classes
a) Those which are like European options but more complex. We can do now
 as can I think any Monte Carlo
b) Those like mortgage backed securities that can have early exercise
  These can be treated using "American Option" technique