Flawed Logic

My results yesterday was actually premised on the knowledge of 2 parameters in the system. While this is significant project given the amount of parameters in the model; I must still work on this problem to element the need for this knowledge.

My model is so complex now that I have to decide between:

1. the time it's takes to get results
2. the accuracy of my results

If I increase my data observation I get better results, however it could take hours longer to finish. When you are certain that what you are doing is correct this is fine. However when you are just testing out models it is not cool to wait 2 hours only to find out you accidentally added a line of code that wasn't supposed to be there, which mean you must run the whole thing again!

Write now I am running a very computational intensive step known as a forward backward smoother. Everything else in my program runs quickly, however it takes 6-7 minutes to run this step, and I need to run it at least 10 times. So I try to time my running of the program around lunch time and when I teach. I really should get in the habit of running the really long simulations over night when I am sleeping.

I am sure other statisticians in the department would scoff at my simulations only taking hours as I know some of them are running simulations that take months and in some cases years running on 4-5 different computers! In our computer lab it is not uncommon to see multiple computers with a "please don't touch" paper over it.

Once again, I am giving a brief 2-3 minute talk about my research to a visiting professor at 4. Other than that I just plan on playing around with my results until next week when I can meet with my adviser. I feel as I am making progress at a reasonable pace right now so I am not anxious or anything. This is a very rare feeling in my life.