While the code is written for generic upwind and downwind processes, the baseline has the same sign for all $x$.
The question is:
[x] what exactly are the boundary values and how do they manifest into A and how are they interpreted in the case of $\mu > 0$ vs. the one we have been focusing on with $\mu < 0$. What changes and why?
For example, in the $\mu < 0$ the complication at the right hand side was that for a maximum of $N$ points, it was trying to reference the $N+1$ due to the 2nd derivative. The first derivative was upwind and using backward differences so didn't reverence the $N+1$ directly.
[x] Is it the same story with a $\mu > 0$? Does it use EXACTLY the same assumptions (but now working forward? We need to write this down explicitly.
[x] Similar complications occur on the left hand boundary when done carefully.
[x] Another part of this is what the simpest form of the $A$ matrix construction looks like if the diffusion is always backwards. We might have this in one of the forms of our model, and removing the max/min functions for the upwind could help a great deal.
[x] Write the construction up in latex and write up the code that just verifies the construction.
While the code is written for generic upwind and downwind processes, the baseline has the same sign for all $x$.
The question is:
For example, in the $\mu < 0$ the complication at the right hand side was that for a maximum of $N$ points, it was trying to reference the $N+1$ due to the 2nd derivative. The first derivative was upwind and using backward differences so didn't reverence the $N+1$ directly.
[x] Is it the same story with a $\mu > 0$? Does it use EXACTLY the same assumptions (but now working forward? We need to write this down explicitly.
[x] Similar complications occur on the left hand boundary when done carefully.
[x] For all cases, write things up very explicitly in optimal_stopping_notes