DOE CSGF

[intirb]

Hello all! Anyone here applying or thinking of applying for this fellowship? I'm a new fellow of this wonderful program and would be very happy to offer advice or answer questions if it's needed.

[nl247]

I am planning to apply for this fellowship for my first time this year. I am wondering in general what all the application consists of in terms of essays required. I know the official application won't be available until October but I was hoping to start gathering my thoughts together early. Is it similar to the NSF GRFP with three required essays? I know this fellowship is incredibly competitive. Do you think there is something in particular that they look for in applicants or do you just have to be an all around super stud to get this fellowship?

I really appreciate you coming here to lend some advice as a successful applicant!

[intirb]

Hey there! So glad you're interested in applying! I really do hope there are others out there, and that it's just a little too early in the application process for people.

The application requires a few essays. I don't know if they're going to change the format from last year, but we had one essay describing our intended research project within computational science, one essay describing our need for high performance computing, and one essay defending our program of study. These essays will have to be very short (probably just a paragraph), so you'll need to be very concise. They'll ask you to talk about specific things, so follow the rules and talk about those specific things. Reading very carefully seemed to be a pretty big part of this application.

Talking to a few people at the conference last year, it seemed like they looked for the following:

1. Interest in computational science. Note that computational science IS NOT computer science (I think a lot of people get tripped up here). I'm not sure I can give a perfect definition of what computational science is, but in general it involves using highly-intensive computations to solve some science/engineering problem and sometimes can also be developing new, faster ways to perform highly-intensive computations. If you think you still need a better idea, browse this page. That publication should give you a good idea of what kinds of things people are currently working on in computational science, including fellows.

However, don't feel like you need to stick to traditional DOE science domains (simulations, energy, etc.). I wanted to apply computational science to a new science domain - as long as it's within the fields of study listed, it's ok. Your essays should definitely reflect that you know what computational science is and you have a good need for it in your work. Also, don't feel like you need to be a computer scientist. One of the things they like to do is get scientists and engineers from other fields into the field of computational science, and actually I think your chances of getting in as a CS major are a lot smaller (perhaps because CS majors are more likely to confuse computer science work and appropriate computational science projects).

Anyway, in your essays, you really need to convincingly justify why a supercomputer would be absolutely necessary in this really cool science/engineering project you have.

2. Great letters of recommendation. This one seems to be really key. Your recommenders should be able to comment on your research ability and rank you very high among the other students they've had. Ideally, all of your recommenders would have seen you in a research setting. Even more ideal, they should be familiar with the CSGF program and comment on how well your proposed research fits in with their mission (I didn't have this, but it can definitely be a huge plus).

3. Program of Study. Your program of study should meet all of the rules and criteria set out for you (when the application opens up, they'll have a bunch of places with more information about the POS, so read carefully!). Those course choices should also make a ton of sense regarding your research project. Unless you have a ton of experience with parallel programming already, one of your computer science classes should probably be that. If you're struggling to find classes that meet all the rules and would help your research, you should think carefully about the school you've chosen and possibly also whether your research project truly makes sense for this program. Ideally, you should develop your program of study together with the advice of some professor and graduate students at your top-choice school, and that professor should be willing for you to tentatively write him/her in as your advisor (but this is non-binding).

Grades, awards, publications and all that stuff are of course important, but I think those above things are considered more carefully.

Of course, all the usual disclaimers about how I'm not actually involved with the admissions process, don't speak for Krell and am not claiming to do so, and I'm just a very new first-year fellow.

I will say, with the authority of a fellow, that this fellowship is wonderful, all of the 'obligations' seem a lot more like opportunities to me, and you'll meet some crazy smart and interesting people if you get it!

If you have more questions, feel free to ask!

Edited September 10, 2012 by intirb

[anacron]

I applied last year and will be applying again this year. I ended up with one of those pseudo-honorable mention thing, so hopefully things will work out this time around. Just out of curiosity, did they select first-year graduate students during last year's competition? I feel like the bar would be slightly higher for students who are already in grad school - even if it is just a semester into it.

[intirb]

Yes, there were at least a few first year grad students. I can't really remember the exact breakdown. I don't really know how they compare first year graduate students to undergraduate seniors.. I'd also imagine the bar is slightly higher, but you also have a much better shot of putting together a plausible POS with an advisor at your school.

[InquilineKea]

I'm thinking of applying - I'm in climate science - so there is definitely a lot of good reason for me to apply..

I wonder about the "Anyway, in your essays, you really need to convincingly justify why a supercomputer would be absolutely necessary in this really cool science/engineering project you have. " thing though. Could this mean that students from universities without powerful supercomputers could actually be given preference?

Do they put reviewer comments up too?

Edited September 30, 2012 by InquilineKea

[anacron]

Could this mean that students from universities without powerful supercomputers could actually be given preference?

I don't think this plays any part during the review because they're not providing access to supercomputers... though I guess you could interpret the practicum as a way of giving you access. By definition computational science uses supercomputers to run large scale simulations, so the real question is: Does your research project come under computational science and can you provide a convincing argument for it?

[intirb]

To InquilineKea - nope, they don't consider that at all. They just want to know that your project is beyond the scale of a typical cluster (say ~32 processors in parallel). If you have access to a supercomputer already, then great! Krell will give you a ton of resources on how to take advantage of that access. And no, sadly there are no reviewer comments on your application (though some people who just missed it are strongly encouraged to apply again). It's not like NSF where you get feedback on your essays or anything.

To Anacron - you do actually get a small allocation outside of your practicum, and I think if you have a large-scale project in mind and you're serious about pursuing it, there are people that will help you apply for more time. Also, computational science is definitely not limited to simulations! If you're doing computational or analytical work on large datasets (classically, astronomy), for example, that counts too. Here's another example.

[anacron]

Good point, I shouldn't have limited it to simply simulations but I assumed it was much more likely that InquilineKea's research was related to climate modeling and simulations. In fact, my proposal is about large data analytics. Either way showing that your research needs massively parallel machines and that you have already done research on these classes of machines would be a huge bonus.

[intirb]

Yep, agreed! Best of luck to all of you, and feel free to ask more questions as they come up during the application process!

[Chasely]

I'm in the process of applying.

Intirb, you noted about the "obligations" in the program. If I read the materials correctly the only "obligation" is the 12-week research practicum, correct? I saw that fellows should "seriously consider" job offers from the DOE and their partners after completion of the fellowship, but you're not required to work somewhere specific upon completion of your PhD.

Oh, and there's also the conferences each year. But that's it, right?

I couldn't tell if you're then tied to your proposed research topic, or if you can change your research and still receive the fellowship (contingent upon you still doing computational science relevant work).

Let me know if there is anything else.

Edited November 16, 2012 by Chasely

[intirb]

The program of study would be the other big obligation.

As with all of these fellowships, they're ultimately funding the fellow and not the research, though they'd probably seriously reconsider if you switched fields. I don't think every fellow ultimately uses high performance computing in their thesis research, but they all use computers and computation in one way or another.

[Chasely]

Thanks, the twelve week practicum is a one-time event, correct? You don't have to do that every year you have the fellowship, I hope.

EDIT: Found the answer. It's a one-time requirement, to be completed in the first two years of the fellowship.

Edited November 20, 2012 by Chasely

[InquilineKea]

So the "Additional Comments" section has space for 2400 characters... Would it be a good idea to fill that all out, given that all the other boxes have stringent character limits?

Also, what is the difference between "Department" and "Academic Discipline" on Higher Educational History?

Regarding the program of study - are you really expected to follow what you've planned? There are so many changes in course plans that can happen (new courses being announced, a course not being taught anymore) that I'm not sure if it's all that realistic, especially given that grad-lvl courses tend to be more "volatile" in their availability than UG-lvl ones..

==

Also - do they expect "Mathematics & Statistics" courses or "Applied Math" courses? The "Program of Study Requirements" says "Applied Math", while the main form says "Mathematics & Statistics"... Does "Fluid Dynamics" count as "Applied Math" if it's offered under an Applied Math dept?

==

Is it possible to transfer schools while under the CSGF?

Edited December 19, 2012 by InquilineKea

[intirb]

So the "Additional Comments" section has space for 2400 characters... Would it be a good idea to fill that all out, given that all the other boxes have stringent character limits?

I would only use this space if you really have something additional that's important to add. I wrote something small because I was having technically difficulties filling out one of the boxes with the department name, so I just wanted to make sure it was correct. Don't feel the need to fill space just because it's there

Also, what is the difference between "Department" and "Academic Discipline" on Higher Educational History?

Uh... I forget. Do you have a screen shot or something (you can blur out the details) to jog my memory?

Regarding the program of study - are you really expected to follow what you've planned? There are so many changes in course plans that can happen (new courses being announced, a course not being taught anymore) that I'm not sure if it's all that realistic, especially given that grad-lvl courses tend to be more "volatile" in their availability than UG-lvl ones..

Yes. They take this VERY seriously. Obviously, if a class gets dropped or moved around a semester, you will have to change, but that's pretty much the only allowable excuse for changing courses once you've accepted the fellowship and signed the POS.

Also - do they expect "Mathematics & Statistics" courses or "Applied Math" courses? The "Program of Study Requirements" says "Applied Math", while the main form says "Mathematics & Statistics"... Does "Fluid Dynamics" count as "Applied Math" if it's offered under an Applied Math dept?

Your math courses can be Mathematics, Statistics, or Applied math, as long as they are in one of those departments. Different schools have different department names. I'm not sure, but I doubt they'd accept a fluid dynamics course just because it's cross-listed in a math department. You should really try to follow the spirit of the POS and find two relevant math courses.

Is it possible to transfer schools while under the CSGF?

Yes. Are you an incoming student or a first-year graduate student? If you're an incoming student, you can take the fellowship to any school you get accepted to - no worries about that. If you're a grad student, you can transfer, but they'll probably want to know that your research will stay related to computational science.

[frogism]

I come from a very non-mathy biological field that is becoming very mathy (and there is an increasing need to be more mathy and to utilize computational resources). As such, my training in mathematics and computational science is lackluster. Does this matter? As in, all the graduate courses I am planning to take all require a large amount of prior information [from upper division undergraduate courses] for which I have no foundation to rest upon. Do they care about this?

[intirb]

Well, first: I also had very little formal training in computer science, though I had taken a decent amount of math courses and programmed a lot for lab.  I'm not sure how much background they expect.  They do like to bring in people who are primarily domain scientists or engineers and introduce them to computational science.  Just tell them about the training you do have, even if it was informal.

 

Second: I think the more important question is - do you care about your lack of foundation for the courses you're signing up to take?  If you get the fellowship, you will have to take them and submit your grades back as part of the renewal process.  Are you confident you can take the courses in your POS as you've scheduled them and do well?  If you think you can handle it, I'm sure they'll be fine with it. 

[InquilineKea]

Thanks so much for the excellent responses, intirb!

 

Regarding the question that you requested an image of - here is the image (that's where I'm getting the issue between department and academic discipline):

 

[intirb]

So I'm not really sure what I put, but I THINK that department is the official department, and academic discipline would be your concentration or area within that department.  For example, Department: Neuroscience and Academic Discipline: Computational Neuroscience.  I'm sorry I can't be more helpful.

[InquilineKea]

Ah okay I see. Yeah that's what I'm thinking too. Thanks again!

[InquilineKea]

Could anyone be willing to share their personal copies of the short essays that they wrote for the DOE CSGF? I've never seen any sample essays yet, and all 4 DOE CSGF people I emailed didn't reply to my emails.

[InquilineKea]

Also - regarding the course plans - is it okay if I take most of the CS/Applied Math courses in, say, my 3rd year of grad school? Or my 4th year if I'm a current 1st-year student who is applying?

 

As a side question - what would happen if I actually dropped or failed one of the courses in my POS?

Edited January 7, 2013 by InquilineKea

[InquilineKea]

Also - when I'm trying to describe my courses, as shown below...

 

 

Course description Required Please do not use URLs. (Max characters is 800,  ) 

 

Is it okay if I simply copy the course description from the course catalog?

 

==

 

Also, how important is it when I list "other planned courses" in my list of courses? The one thing with my school is that it doesn't offer that much coursework in my specialty, so I would have to go over to Harvard/MIT to take more courses in atmospheric science. Would they be okay with my taking courses at other institutions? (there is somewhat more friction with that though, and the course plan may not be believable since I would be even less certain about the course availability)

Edited January 8, 2013 by InquilineKea

[intirb]

If you are an incoming student (meaning, you'll start the fellowship when you start grad school), you can take classes from your POS in the 3rd year of grad school.  If you are a first year grad student, you need to finish your courses in the next two years (so your last class for your POS should be completed BEFORE Fall 2015).  I think this smaller timeline is made up for by the fact that you can list some classes you've already taken or will take this year.

 

Edit: but within those parameters, just take the classes whenever they make the most sense for your own program requirements and research goals.

 

If you dropped or failed a class, I think you'd probably seriously risk your renewal for the fellowship.  Renewal is not automatic, and it depends on how well you're progressing in grad school, including how well you're doing your POS.  Of course, if there were an unexpected life event (illness etc.), I'm sure you could work it out with Krell. 

 

I definitely copied the description from the course catalog .

Edited January 8, 2013 by intirb

[InquilineKea]

Thanks for the reply again. How many courses did you list in your "Other Planned Courses", by the way?

 

Also - do you think this course below could count under Applied Math? (from http://www.dam.brown.edu/courses/ )

 

 

APMA 2580. Computational Fluid Dynamics
An introduction to computational fluid dynamics with emphasis on incompressible flows. Reviews the basic discretization methods (finite differences and finite volumes) following a pedagogical approach from basic operators to the Navier-Stokes equations. Suitable for first-year graduate students, more advanced students, and senior undergraduates. Requirements include three to four computer projects. Material from APMA 1170 and APMA 1180 is appropriate as prerequisite, but no prior knowledge of fluid dynamics is necessary. 

 

I'm not counting Fluid Dynamics as an Applied Math course per your advice, but one of my advisers is suggesting that I list Computational Fluid Dynamics as an Applied Math one (even after I told him about the standards they expect of the Applied Math courses). Maybe the "basic discretization" methods count as Applied Math?

 

==

 

Also - if you have past research that did not lead to a publication, then do you pretty much have to refrain from mentioning it? Or can I actually list them as extracurricular activities?

 

Of course, some professors might write LORs about your research experience but you can't be certain about that.

 

==

 

Here are my essays, for anyone to proofread/comment on: https://docs.google.com/document/d/1NtwDK2B8kFYfN6beoDSDIT5poUqr5NMdM6KXHgqWLB8/edit . Due tomorrow at 7 PM EST/4 PM PST.

Edited January 8, 2013 by InquilineKea