You think it is the man, but it is the destiny the two of you shared for a time, the destiny which is past, that stirs your mind, and you call it love
- p.171 Lord of Light
Follow up about applying to Aurora.I applied for the Oxford role. Going to see if I can apply for the others.- Email Anthony Quinn?
I generally feel tired these days, or just apathetic during the day. I can’t really say if this is the workouts or intellectually somethings gone wrong.
This anxiety about getting probability study done is affecting my research focus. Going to try and nip it in the bud. Really need to be conscious that research is my main priority. Feels a bit easier to think with that in mind.
If I really do feel its necessary I will be able get it in at the weekend. But what I want to be doing is right in front me and is plenty interesting. There’s no need to be looking at some other ‘shiny object’.
Research
Ultimately, what can we do in an uncertainty analysis frame at the moment?
- We can run a process with a given set of inputs any number of times.
- How does MC uncertainty propagating work? The way I see it is to draw random samples of input paramters from a known distribution and then statistically classify the output distribution you get.
- Each unit process would then have an output uncertainty representation based on an input. This is assuming that all input parameters are drawn randomly on each run, not just one with the others constant.
- Take the output distribution and characterise it for each unit process.
- Connect unit processes together. Do they need to be ‘connected’? The uncertainty propagation would just be the addition of a load of random variables (unit process outputs).
There seems to be some material also on that ‘value of inforamtion’ area I’ve been trying to think about. It’s really about maximising what the core of the paper will be though, which is an uncertainty analysis. I mean, all this extra layer stuff would just be nice to have.
Should the Python code become more Uncertainty orientated?