re-write in python!

[karllark]

Why?

[davidwhogg]

Because right now no-one else can run the code, apparently. Making a new issue...

[dweisz]

I also need to get a handle for how to test the ClusterSED code using mock data. I will need to do some end to end tests.

On a related note, I think the mock cluster data might have a completeness function applied to it; the MF slope is coming back well-constrained by flat. I'd like to test a mock cluster where I'm absolutely certain about the completeness.

[karllark]

As for no one else being able to run the code, this is not surprising. I am in the process of going through the code and cleaning it up so it is easy for others to use it. This step is necessary given I hadn't written the code for others to use it. I was too busy working on the code for capability. I'm most of the way through this step (checked in a bunch of code yesterday), but the code is currently very likely broken as I'm cleaning up the code in the process. Will check it soon to make sure.

[karllark]

I'm actually not excited about having two versions of the code - one in python and one in idl. I can handle this for the cluster fitting as I'm happy to cede the main development of that code. I'm not happy at this point to cede the development of the SED fitting code. I have enhancements that I want to make to the code and don't want to be diverted to learning the necessary python myself. Leaving this code in idl will allow me to worry about the enhancements instead of code learning. At this point, I would prefer to concentrate on getting a paper or two written. Is there an intrinsic benefit to writing this code in python? Or is this a distraction? (assume that soon the code will be straightforward to run in IDL soon when responding).

[karllark]

Dan: the mock cluster did have a completeness function applied - you applied it. It was on F814w and a step function at 24.

[dweisz]

Thanks for clarifying, Karl. I created a lot of mock data, so I couldn't remember which one this set was.

[davidwhogg]

Think of the Python version as a fork -- it won't be your responsibility; it will be your competitor. Competition is good.

[davidwhogg]

Maria -- want to give it a shot?

[davidwhogg]

And by "Maria" I mean @kapala

[karllark]

While I understand the interest in having the SED fitting code in python, I am not at the point where I think this is the best use of our combined time. There are other codes needing writing (e.g. dust fitting from the SED fitting, dust fitting from the SED fitting & and the Spitzer/Herschel data, and I'm sure others). In addition I am actively developing the SED fitting code. My interest is in getting a couple of papers out, and having a python version isn't yet the priority for me. I am not yet enough of a python coder that I want to be distracted right now with trying to do it in python (hence developing it in IDL). Once the SED fitting code is fully developed (at least 6D), that seems like the right time to make a python version. I would estimate this is on the order of a few months. In fact, I would propose that we spend a day at the Nov PHAT meeting doing the python version. I would like to be part of the python effort so I can benefit from the learning experience.

But that is just my 2 cents. Others can do as they wish of course. I understand the statement that competition is good, but this is starting to feel like more competition than I would desire as a member of the PHAT team.

[karllark]

Or maybe I'm just feeling old and cranky this morning. Meh.

[dweisz]

After spending some time looking through the SED code, I generally agree with Karl. It's a little bit early to tackle the python transition. Morgan and I planning to take some time to map out the IDL version of the code and put together a flow chart for what the python coding would take. We'll probably take a first pass at this next week, even if the full 6D code isn't completely done. I think once we have it mapped out, it shouldn't be too hard to have various people code up different modules.

[davidwhogg]

cool with me. Maria just needs some parts of the code -- not the whole enchilada -- to be callable from Python. She can code that part up and then close this issue.

[kapala]

Just want to comment, that the only thing I would like to do is to rewrite the part to retrieve SEDs in python. Mostly for myself, because I don't know IDL...

[karllark]

Or maybe leave the issue open as I think we should be thinking about this. But it sounds like if we want to track a smaller more near-term python issue - retrieving SEDs in python - then a new issue should be started for that.

[dweisz]

@kapala -- by retrieving SEDs, do you mean reading in the 4D (or 6D) PDF for an individual star?

[kapala]

I meant 4D. I really want to write it for myself and if it will be good enough, then I can think about sharing it.

[dweisz]

Sounds good to me.