So the result I was working on previously is done—four or five weeks after I finished writing my proposal. Suck it, planning fallacy. Today and tomorrow I’m going to work on writing it up carefully.
The new direction my project is taking is to get some quantitative idea of when exactly the result will hold. It’s one of those things where we know it will hold for small epsilon, but that’s useless to everyone unless we know how small epsilon needs to be. Given my schedule and teaching responsibilities this semester, I imagine I’ll make reasonable amounts of progress before the end of the month.
For summer, I still need to apply to various summer schools and conferences.
I’m still thinking about continuous-time causal models, but I’m no longer convinced that I have the right definitions for things. The phase space treatment I gave in an earlier discussion thread really does only make sense if the paths have well-defined derivatives everywhere, and not just outside a set of measure zero. If they have well-defined derivatives everywhere, then the process is no longer Markov.
I’d like to have a second post written by the end of the month, but this project has to take lower priority to my research.
So the result I was working on previously is done—four or five weeks after I finished writing my proposal. Suck it, planning fallacy. Today and tomorrow I’m going to work on writing it up carefully.
The new direction my project is taking is to get some quantitative idea of when exactly the result will hold. It’s one of those things where we know it will hold for small epsilon, but that’s useless to everyone unless we know how small epsilon needs to be. Given my schedule and teaching responsibilities this semester, I imagine I’ll make reasonable amounts of progress before the end of the month.
For summer, I still need to apply to various summer schools and conferences.
I’m still thinking about continuous-time causal models, but I’m no longer convinced that I have the right definitions for things. The phase space treatment I gave in an earlier discussion thread really does only make sense if the paths have well-defined derivatives everywhere, and not just outside a set of measure zero. If they have well-defined derivatives everywhere, then the process is no longer Markov.
I’d like to have a second post written by the end of the month, but this project has to take lower priority to my research.