I work on my AER.
Every great statue was made by chipping away hard stone bit by bit. And that's also how great papers are made.
What do you all do while your code runs?
-
I work on my AER.
Every great statue was made by chipping away hard stone bit by bit. And that's also how great papers are made.Of course, that's what I'd like to do as well. But what if this is the AER you're working on and the code is running for 3-5 mins? What do you do in that time?
-
What do you all do while your code runs? Sometimes it runs for 3-5 minutes. That's too short to work on something else and too long to wait patiently. I often end up distracting myself with nonsense. What's a more productive use of my time?
Related question - sometimes I’m not sure how long the code will run for. What to do while I’m waiting - think 10-30 min range.
-
I figure out how to make the code run faster. Can you profile it and identify and eliminate hotspots? Vectorize better? Parallelize better? Pre-compute results from an especially slow part that doesn't need to run as often? I also run things on small subsets of the data first, so I identify any potential bugs early. That way, I rarely need to run my code on the full data set more than once. (Of course, that strategy doesn't work if you're p-hacking.) If waiting truly is unavoidable, I either work on a different project or start writing up results and producing figures and tables based on output from a sub-sample.