For a last couple weeks I’ve been working on a marine ecosystem model using a technique called Inverse Modeling[ref]I’ll be sure to do a writeup on what Inverse Modeling is and what makes it interesting in the future.[/ref]; and while there’s been lots of progress, I’m starting to get to the point where the model takes some time to run. As you can imagine, there’s a lot of waiting-around time associated with having a model run for an hour or more at a time. As fate would have it I thought up a pretty fun way to keep updated on the progress of my model during a run cycle today.
I decided that it would be nice if my model would send me a tweet when it finished, and that seemed pretty doable since I know that Twitter has a pretty decent API. So I searched around for a few minutes, setup a new Twitter account, and the system was good to go after no more than ten minutes. And it was awesome! But since it didn’t take long to implement I still had about an hour left.
Thats when I got the idea that it should also send me the compiled results from the last data run. Now this step was a bit more tricky since I needed to upload the results to my misc directory and then to tweet out a link to said file. While I wont go into details here (I’ll save it for a dedicated post), I used FTP and a 32 bit hash function to avoid collisions along side the Twitter API for this system. I believe that the results speak for themselves.
@TBryceKelly Starting cycle 2 run for 1e+06 iterations. Jump size is 1.5 and burnin period is 1e+06
— TKelly’s Update Net (@TKellysNet) February 26, 2015
Now I’m on the lookout for more ways to use this Twitter interface for everyday things.