bananaphone

Haha jesus, the fellow-to-applicant ratio in this thread is like 4:1. But seriously, it's an awesome program.

First off, congrats! Speaking as a grad student in atmospheric science, those are both phenomenal programs. I don't think you can really go wrong with either choice. So I grew up in Boulder and did my undergrad at CU. I love Boulder. It's amazing. There's world-class climbing, skiing, and biking. Boulder has an awesome microbrewery scene. Plus, it's god-damn beautiful there. I didn't really appreciate just how amazing Boulder was when I was growing up, I took it for granted. I'm blown away by how beautiful Boulder is whenever I go home. That said, I've spent some time in Seattle (during the summer when it wasn't raining) and I loved it. It's easily one of my top-5 favorite cities. I would be happy to end up in either of those cities. I'd mostly agree with your academic assessment, UW is probably a better program overall. Anecdotally, I've recently seen a lot of UW alumni getting TT faculty positions at some solid programs. Plus, having a funded project is a HUGE plus. (However, take my "academic advice" with a grain of salt because I'm in a different subfield.) Other considerations: advising style, group size, prestige, location, etc. As you've already figured out, there's no right answer on where to go. It really depends on what you're looking for. Everyone has different criteria. It seems like you've identified most of the big factors people use to make their decision but there are a couple others that might (or might not!) be relevant to you. In particular, the advising style, group size, and seniority can have a huge influence on your overall happiness, productivity, and success in graduate school. Some questions you might want to ask yourself are: "What kind of advising style do YOU need? Do you want a hands-on advisor that you meet with every week or do you want independence?" Personally, I like independence and would not do well working for a hands-on advisor. I meet with my advisor a couple times a year and just email him when I've got questions. This would not (and does not) work well for some people. "Do you want want to be part of a big group or a small group?" A big group would probably mean interacting more with students/postdocs in the group and a small group would probably be working with your advisor more. Depending on your preference, either of these could be pluses or minuses for you. A big group would mean more people to hang out with/complain to/talk to but less face time with your advisor. Again, what do you think would best suit you? "What stage is your advisor at?" Are they a young hotshot with a lot of ideas? Are they a famous old professor? Do you want someone with an established network of former students or someone who is currently revolutionizing the field? A young professor would probably be very invested in your success (e.g., they may not get tenure if you don't do well) which could mean a lot of pressure on you. It could also mean they'll help promote your work and be an advocate for you. That last sentence is an important one, find someone who will be an advocate for you! Look at their past students and see if their career paths are similar to what you'd like to do. Do they have a history of students not graduating? That would be a red-flag. I'd be happy to answer any questions about life in Boulder or other grad-school related things. Otherwise, congrats again and good luck with the decision!

I think I had the lowest GPA in my grad school cohort. So it's probably on the low-to-mediocre end of GPAs for grad school, although I could be mistaken. Haha, touche. Sorry about temporarily sidetracking the discussion.

No. Research experience helps because that's what graduate school is about. Research. Your application should be about demonstrating that you are a competent researcher and that you'll (eventually) be able to make a novel contribution to the existing body of knowledge. That's pretty much the definintion of a PhD. I disagree. I would say the top students are the best because they have extensive research experience. When I was applying to graduate school I expressed concerns about my mediocre GPA (3.6) to my undergrad PI. He said that no one will even look at my GPA because I had 3+ years of research experience and 3 first-author papers. I ended up getting accepted to all the schools I applied to. My current PI later confirmed that he didn't have a clue what any of his students GPAs or GREs were, he doesn't even look at them. All he cared about was research experience. Furthermore, I don't know of anyone in my program (~50 students) that didn't have prior research experience. I was not making a statistical claim. As I mentioned, I skimmed the recent results lists and quickly found evidence contrary to your implied claim that good GPA and GREs will get you in: "You get a 4.0 GPA and a 900 on the physics GRE and come from an american university and see who rejects you?". Top schools may still reject those students. You said it's easier to get into grad school in physics/chemistry/math without prior research than it is to get into earth science without prior research experience. Here's the quote: "I think thats really the difference between heavy description sciences...and quantitative sciences...you see math, chemistry, physics and computer science students get into graduate schools without much research experience. Even a lot of lab based research is somewhat lenient.". I was saying that research experience is equally important in those fields.

I disagree with your generalization that it's easier to get into grad school in physics/chemistry without prior research experience. Good grades and test scores alone are not enough to get into those programs at my university. Just have a look at some of the MIT/Harvard/Stanford rejections from Physics/Chemistry. Here's a few examples from skimming this year's results page: (1) Harvard Physics rejected a student with a 4.0 GPA and a 980 on the physics GRE, (2) MIT Physics rejected a student with a 4.0 GPA and a 900 on the physics GRE, and (3) Stanford Physics rejected a student with a 3.92 GPA and a 900 on the physics GRE. Good GPAs and GREs alone are not sufficient to get into the top programs in physics/chemistry/math just like good GPAs and GREs alone are not sufficient to get into the top programs in earth science.

Really? I disagree with that statement. My friends in math/chemistry/physics departments all had extensive research experience. Most had 3+ years of research experience and had their name on at least one paper in undergrad. Those fields are super competitive.

I also applied during my first year of graduate school. I just wrote about the project that I thought I was going to focus on even though I hadn't been formally proposed it at the time. Personally, I think it's important to be able to identify specific issues/problems that computational science could advance. You don't have to get into the nitty gritty details but do go into some specifics beyond your general subfield. An example could be in machine learning (not my field), the algorithms for deep learning/neural nets have been around since the late 80s but it wasn't really that useful because they were too computationally expensive to be practical. However, deep learning is all the rage nowadays because GPUs, among other things, have made the problem computationally tractable. So the essays could discuss how these computing advances would allow you to study problems that were previously deemed too expensive. This is essentially the approach I took in my essays, just applied to my field/research problem. As an aside, an issue in computational science is that we've got a ton of computing resources but there are a lot of fields that don't utilize it. The DOE would love to see new applications or questions tackled with their vast computing resources. Does it? You should probably ask Krell about that one. Personally, I completed all my Sci/Eng requirements before I applied and completed my Math requirements before starting the fellowship - I applied during my first year.

Hey All! The DOE Computational Science Graduate Fellowship (CSGF) application is finally online here. Anyone here applying (or thinking about applying)? I'm a fellow of this awesome program and would be happy to offer advice or answer questions you may have. There's also some advice on this fellowship in the

1. No, most fellows take at least a few courses not in the POS, list those courses in the "Other Planned Courses". 2. You need to take two courses in each category.

I wrote one short sentence for each project, similar to the example they gave: Lab X, 01/1776-Present, Awesome science with super high resolution models and big computers.

Adding to intirb's answer for #2, I would definitely recommend putting a parallel computing course (or something along those lines) on your POS for one of the CS courses. As for your other question, I don't think they're too concerned with the "Other Planned Courses" and haven't needed approval to change my "Other Planned Courses". They're investing a lot of money in the program with the goal of training scientists to do HPC. The courses on the POS are the ones that will teach you the fundamentals of HPC, so those are the ones they care the most about.

Olga Botvinnik posted a link to her blog on here a while back. She was awarded the NDSEG and has a lot of other example essays: Link to her essays. Philip Guo also has some excellent advice on fellowships (Hertz, NSF, and NDSEG): Link to his advice.

The people in fluids/CFD were generally developing CFD models to study specific phenomena. So it depends on what you'd classify as "applications".

Here's some good advice from one of last years finalists.

I can't say for certain what they are looking for, but I can expand a little more on my previous comment about advancing/utilizing HPC. Some of the fellows I've met were doing HPC long before getting the fellowship and were at the bleeding edge of HPC (e.g., they'd done simulations using thousands or hundreds of thousands of processors), their research was advancing the state of PHC. Other fellows had much less HPC experience but their field was be more experimental and they'd proposed to study something novel using HPC. I would classify the former as "advancing HPC" and the latter as "utilizing HPC". Personally, I highlighted some of the current computational limitations in my field and talked about how advancements in HPC would allow me to study some interesting questions. Quite a few of the fellows do CFD, so they're definitely interested in your field. You should put this on a door in your lab/office. They don't constrain you to just compuational work if you're awarded the fellowship (although I'm sure they'd prefer you continued doing computational work).

It's no more than 300 words for each essay, so very brief. I'm happy to answer questions about my approach in the essays but I don't think I want to distribute the actual essays. However, you can probably find exerpts from a lot of winning essays on the CSGF website by looking at the research statements of the current fellows. When we're awarded the fellowship we're supposed to write a 300 word research statement (which happens to be the same length as the essays). I was pretty busy at the time so I just cut and pasted from my applications and I'd bet that other fellows did that too. I wouldn't put the outreach/teaching in the DOE CSGF, focus on the science for the CSGF. They really want to see that you understand what it means to be a computational scientist and what questions can HPC help you answer. So, how can you advance HPC or utilize HPC to advance your field? Their vision of the fellowship is to train scientists how to do this. I tried to emphasize that my research would benefit from advances in HPC, coursework/training in HPC would benefit me, and that these are things that my field typically wouldn't teach me so I'd gain a lot from the fellowship. Definitely put some thought into the courses you propose. They take the courses very seriously and want to see that you'll learn the math/CS fundamentals.

There were some funding issues last year, which is why there were only 10 awards instead of 20, but they told us those were resolved. They're pretty candid with us about the state of the program and the last I'd heard was that it was continuing with the hope of 20 new fellows. I'd give it a little time (I'd bet the shutdown just delayed things a bit). Also, here's the essay prompts from previous years (since the application's not up yet): 1. Field of Interest and the Role of Computational Science Please describe your chosen research area and what contributing role computational science will play. Computational science involves the innovative and essential use of high-performance computation, and/or the development of high-performance computational technologies, to advance knowledge or capabilities in a scientific or engineering discipline. 2. Research Using High-Performance Computing and/or Large Data Analysis What new science or engineering would high performance computing or large data analysis and management enable in your area of interest and why do you think this is the case? In particular, what are the challenges that need to be addressed to make this advancement? 3. Program of Study The fellowship program of study requirement is designed to give you a breadth of competency in fields outside your own that will enhance your ability to perform computational science research. Please describe (in no more than 300 words) how you expect that the courses listed in your planned program of study outside your chosen discipline will contribute to your own research in the future. Describe why you chose these courses and how they will impact your research plans.

Hey Guys! Anyone here applying (or thinking about applying) for the DOE Computational Science Graduate Fellowship (CSGF)? I'm a fellow of this awesome program and would be happy to offer advice or answer questions you may have.

Hertz, NSF GFRP, DOD NDSEG, and the DOE SCGF (wasn't offered last year) are all appropriate for earth science PhD students. Also, if your work is computational you should consider the Department of Energy's Computational Science Graduate Fellowship (DOE CSGF), it's got some really nice perks. First year students definitely have a shot at it. A first year in my department was awarded it. That said, it has a longer research proposal than the others (~5 pages) so you need to have a pretty well developed idea of what you're going to work on.