Watershednormative- Ecochain (no jobs but could email).
- Greenly (could email).
I was thinking today about that skier, who had the idea that when you do something, rememebering that its voluntary. This is something that I thought about a lot through the winter and part of me thinks it would do best for me to remember it now.
It applies to training but also, if you have a plan and its coherent etc. why let noise disrupt that signal, why whine about the hard parts or second guess it. It can only be evaluated on the agreed upon time scale.
I’m generally being very whiney today, its making me annoyed at myself. Hopefully I can suck it up.
Hamming seems to have a nice book on probability.
I need to sleep.
Research
- It might be a good idea to show what the calculation ‘looks like’ in the Python vs excel.
- I’m really frustrated at the moment with the project, I’m not sure why.
- I think it would be nice to have a completely accurate ‘fuel production’ model to work with. At the moment I’m not too sure that it is completely accurate, nor what it would take.
- I think I have to try and finish the fuel production process.
- Is there a way to mark uncertain things in the process of calculation (like say in the reaction mechanisms).
Trying to pin down the ‘bread and butter’ uncertainty analysis according to the literature.
- MC simulation for error propogation.
- Analytical uncertainty analysis would be more painstaking but less computationally expensive.
- Sensitivity analysis (the benefits of which might be underrated?).
- Something like SALib could be useful here.
- Local.
- Global.
Tangential things would be incorporating expert knowledge.
What does it mean for a model to be data-driven?
It feels like theres a divide of paralysis by analysis, what is the minimum set of tests we can perform that are useful in that we can be highly confident that we’ve captured at least some element of uncertainty. A detailed sensitivity analysis of the three processes that I currently have coded, along with their combination. This followed by a detailed MC sampling of the three processes. This won’t classify the whole fuel production process but it will give us some idea of combining processes and how that propogates uncertainty as well as being super careful on the sensitivity analysis we perform.
This could become a template process then for any new unit process we want to characterise.
Generally, I’m struggling to communicate what I’m doing effectively. Hopefully some of the modelling terminology can help.