Pages

Sunday, January 9, 2011

It's Astro-physics not Astro-recipes

The other day my adviser said something interesting. We were talking about the current state of theoretical astrophysics papers out there and he said, "All these papers just give recipes. They say, do this and do that and you will get this. They don't talk about the physics. They don't try to look into the different physics that could cause what we see. It's supposed to be Astro-physics not Astro-recipes."

This struck me as a particularly good criticism because the vast majority of papers I have been reading all have a lot of recipes for how to recreate their simulations, but very few actually look at the different physics involved. They focus more on the number of grid points in their simulation than on whether or not their physics produces something we see through a telescope. Perhaps this is a problem more for my particular sub-field, but it does seem to pop up everywhere.

3 comments:

  1. It's in all fields of science. It's one of the byproducts of methodology-driven science instead of theory-driven science. I'm not bashing methodology-driven research, it needs to be done, but it's easy to lose sight of the theory behind the methodology.

    Science as a whole has lost much of the theory behind the research. Further, most scientists do not think about the philosophical foundations and assumptions that drive science in general. Many have lost sight that they (we) earn doctorates of philosophy.

    ReplyDelete
  2. Quantumleap42,

    Yeah, it seems like in a lot of science a substantial part of time is devoted to algorithm development as opposed to theory development.

    How often when you have had to TA you would say things like:

    1. First draw a picture of the problem.
    2. Write down each variable, and whether it is a known or unknown.
    3. Draw the forces with arrows....
    4. Etc...

    So much of Physics is learning useful algorithms or as you call them: receipts. You may not understand the theory but if you master these algorithms you can sure get a lot of physics problems right.

    ReplyDelete
  3. Joe,

    Very true. I had to laugh at your list, because I've said (and still say sometimes) those exact things. Also, how many times is the first step in your experiment to "re-do the other guy's experiment, following his experimental conditions, and see if you can reproduce his data." It's useful and important, but we also need to make sure we understand why doing A, B, and C gives you X, Y, and Z.

    It's true, at some level we all need to learn the "recipes" -- how to do the problem. However, at some point, it's really useful to really understand what's going on behind the problem and why the recipe works. Great post.

    ReplyDelete

To add a link to text:
<a href="URL">Text</a>