Selectivity of computational physics vs. theoretical vs. experimental?

In summary, most professors in my physics department have said that it is harder to get in as a theory student than as an experimentalist.
  • #1
tiyusufaly
29
0
Pretty much every professor in my physics department has said that it when it comes to American physics graduate school, it is harder to get in as a theory student than as an experimentalist.

Do graduate programs consider computational physics, numerical material sciences, etc... essentially theory or do they consider it separate from theory and experiment. Specifically, if one were interested in computational work, would that make it as hard to get into graduate programs as if one were more interested in analytical, "pure" theory?
 
Physics news on Phys.org
  • #2
From my observation, some computational physics (science in general) people need not much deep theoretical mastery and powerful analytical, mathematical skill to the point that once you know how to program something, you can get the (computational) jobs done
Some do not even understand the very basic principles of quantum mechanics, but with skills in C/C++, Fortran, etc, in some colleges, one can easily get a Master's or PhD degree doing very applied physics projects computationally or those fields which hardly need any other things than computers, like nonlinear dynamics; chaos theory, etc
(some claim they are doing "numerical/computational EXPERIMENT", very funny)
On the other hand, IMO, pure theoretical physicists have to deal with the worst mathematical beasts most of the time to deserve being considered theoretical physicists, second only (in terms of difficulty of the research) to pure mathematicians and resort to computational solutions as the very very last and ugliest, least elegant option
I don't know how one can compare the two, they don't seem alike to me, not even close
And for the two to be considered equals in terms of selectivity for a grad school admission process, especially in top American universities, doesn't make any good sense at all to me, IMHO
If the two were equals, perhaps IAS at Princeton would be a computer lab and Einstein or Ed Witten discovered their theories using commercial softwares there, so silly
 
Last edited:
  • #3
You seem to look down on computational physicists, as if they somehow know less than you do, why is that Urkel? Most of the theoretical physicists I've known spend a good deal of time using computers to compute numerical solutions to equations, and often there is a lot of overlap between theoretical and computational physics.

On the other hand, experimentalists often use computer simulations (yes computational experiments) to test the effects of various parameters, since sometimes certain things are hard to control in a lab and/or the experiment is expensive to repeat for many different sets of parameters, so there can be quite a bit of overlap between computational physics and experimental physics in that way.

Someone who specializes in computational physics is (should be) more familiar with the algorithms known/available, how to optimize calculations/programs, how to structure programs. An experimental or theoretical physicist, when doing computational problems will often put a program together which `works', but which could be optimized a lot through the use of different algorithms/program structure.

Urkel said:
(some claim they are doing "numerical/computational EXPERIMENT", very funny)

Why do you find this so funny? You input a set of parameters into your simulation, and see what sorts of results you get. How is this any different than setting up an experiment in the lab, and seeing what sorts of results you get?
 
  • #4
I don't look down or anything like that; but intended to emphasize the clear distinction between a purely theoretical-oriented research and computational research
To regard them as not different from each other does not sound that right
About computational/numerical "experiment", I was being very conservative about definition of experiment; simplest definition; hands-on, practical, handling the real (physical) stuff (instruments, equipments, etc); put simply, if theoreticians predict how things work, experimentalists try in lab themselves whether the things work indeed, also test if something interesting happens in the real system
Computational physicists try to predict, but since they just want to do little math other than Euler/leap-frog method or trapezoid rule or Runge-Kutta for integration, for example, they ask the computer to solve sometimes the supposedly analytically-solvable equations for them; working on model is not working on the real thing (physically, in lab), if that's what one thinks, the Prizes for discoveries of quantum Hall effect or superfluidity in Helium or superconductivity would have gone to computational physicists; I wish that ever happened!
Also, IMO, experimental skill is talent, lucky those who have that, in fact many great theoreticians failed to do experiment; like Heisenberg who failed his astronomy exam in his PhD defense, also C N Yang who failed to make his particle physics experiment work, in his PhD with Fermi,
Luis Alvarez was great experimentalist without whom Glashow-Salam-Weinberg would never come up with their standard model! Particle physics revolution of 70's might have been fairy tale then! See, I don't look down to any field or people.
 
Last edited:
  • #5
tiyusufaly said:
Pretty much every professor in my physics department has said that it when it comes to American physics graduate school, it is harder to get in as a theory student than as an experimentalist.

In my experience this is correct, though it can depend. I left graduate school with a high opinion of theoreticians as people, but a low opinion of their usefulness in physics. I believe there are others - though obviously not everyone - who share this view.

Often experimentalism brings in more money and therefore is more sought after. They also have higher starting costs, so some of this depends on the school. If you're a big school looking to draw in big money, more experimentalists is typically better, because the grants are larger and the Uni gets a percentage of the grant. On the other hand, if you're a small school with limited resources, theorists have small or no start up costs, and so you can get a professor on the cheap.

Do graduate programs consider computational physics, numerical material sciences, etc... essentially theory or do they consider it separate from theory and experiment.

In the condensed matter department I worked in the term "theory" was used to include computational; in fact, essentially no one was just a theorist, they were all doing computational work. In our HEP department the theorists did dramatically different daily work than physicists doing numerical/computational work. I suspect many other places are similar.
 
  • #6
You get into graduate school based on your performance in undergrad. Once you are accepted by the faculty of graduate studies and the physics department itself, you are then free to look for a project and supervisor. Many students do not make this decision until after a semmester or two of graduate courses. Under certain circumstances having a potential supervisor who wants you as a student can help in the admissions process, but it generally won't override anything if your marks don't meet the cutoffs.

If you come into grad school with a full scholarship, you can pretty much choose whatever project you want - provided you can find a supervisor willing to take on that project as well. If however, you need financial support, you'll gravitate towards where the money is. And the bottom line is that experimentalists often have more money for students than theorists.

Computational projects can be either theoretical or experimental in nature. Numerical methods are a great tool to learn on either end of the spectrum and they will give you a skill that will make you more marketable when you graduate.

With regards to Urkel's statement:
Computational physicists try to predict, but since they just want to do little math other than Euler/leap-frog method or trapezoid rule or Runge-Kutta for integration, for example, they ask the computer to solve sometimes the supposedly analytically-solvable equations for them; working on model is not working on the real thing (physically, in lab), if that's what one thinks
I would argue that it's not a question of wanting or not wanting to approach the particular problem analytically. Once the complexity of a problem reaches a certain point, it becomes more efficient to solve it computationally. Although, I think in context his point is that you can't rely on the computer to do the creative thinking - with which I completely agree.
 
Last edited:
  • #7
Urkel said:
I don't look down or anything like that; but intended to emphasize the clear distinction between a purely theoretical-oriented research and computational research
...
Computational physicists try to predict, but since they just want to do little math other than Euler/leap-frog method or trapezoid rule or Runge-Kutta for integration, for example, they ask the computer to solve sometimes the supposedly analytically-solvable equations for them

The second statement clearly shows your bias, therefore contradicting the first. If you think the only techniques useable are the few that you learned in your undergrad course (I learned all the techniques you mention in 2nd year) you're sadly mistaken.

Urkel said:
About computational/numerical "experiment", I was being very conservative about definition

If by "being very conservative" you mean "completely making up your own"

dictionary.com said:
ex·per·i·ment–noun
1. a test, trial, or tentative procedure; an act or operation for the purpose of discovering something unknown or of testing a principle, supposition, etc.: a chemical experiment; a teaching experiment; an experiment in living.
2. the conducting of such operations; experimentation: a product that is the result of long experiment.
3. Obsolete. experience.
–verb (used without object)
4. to try or test, esp. in order to discover or prove something: to experiment with a new procedure.

By any of the definitions, a computational/numerical trial fits.
 
  • Like
Likes math_denial
  • #8
Being a person who does Computational Physics I'd say that being a student who has a strong computational background and indicates that they wish to continue working computationally will be put in a different pool for consideration then pure theorists. And often this pool is smaller so having a strong computational background can be marketed to an application committee such that it looks very attractive.

As for the digging on computational physics let me just say that in my undergrad I took every single course that the straight physics majors tooks, the difference? I didn't get electives (I had to use them for CS courses) and good luck finding an analytic solution to quantum many-body systems (many being 1000's of particles) and thus good luck getting a handle on emergent phenomena without any computational
 
  • #9
NeoDevin said:
You seem to look down on computational physicists, as if they somehow know less than you do, why is that Urkel? Most of the theoretical physicists I've known spend a good deal of time using computers to compute numerical solutions to equations, and often there is a lot of overlap between theoretical and computational physics.

...

Someone who specializes in computational physics is (should be) more familiar with the algorithms known/available, how to optimize calculations/programs, how to structure programs. An experimental or theoretical physicist, when doing computational problems will often put a program together which `works', but which could be optimized a lot through the use of different algorithms/program structure.
/QUOTE]

I consider myself a Computational Physicist, the theory for my entire dissertation took about a month to work out from start to finish. The computational modeling took over 4 years to get it worked out, written and coded.

I agree with Maverick_... try finding an analytical solution to just about any problem of consequence today, chances are that you won't.
 
  • #10
could some experienced person tell whether this idea is correct? Computational Science deals with the aspects of computing such as designing efficient algorithms to solve an analytical model, the issues that might arise from the numerical methods to solve them etc., and that, it's not the actual science(physics, neuroscience etc) per se.
As such, a course in computational physics/neuroscience, deals with the mathematical ways to solve a problem and isn't concerned with the discipline, the problem arises from?
that is to say, a person with undergrad in math can switch to computatonal neuroscience since he doesn't have to know neuroscience anyway?
i hope someone throws light on this.. thanks!
 
  • #11
Computational science is indeed science!

Maverick's example is very good - Emergent phenomena of many body systems; these phenomena are the science of what's going on. The models that we design to try and explain those phenomena are also the science. Having solutions, analytic and numerical, to the models we invent is great, but hell, that's just the math.

Granted, there are a lot of people who do computational science who could afford to know more about computers, and you can be a very handy research assistant by being the one who does... So yeah, you could probably find a job in computational neuroscience, but if you want to make a career of it, you'll need to learn neuroscience. No one wants to be an RA forever.
 
  • #12
"could some experienced person tell whether this idea is correct? Computational Science deals with the aspects of computing such as designing efficient algorithms to solve an analytical model, the issues that might arise from the numerical methods to solve them etc., and that, it's not the actual science(physics, neuroscience etc) per se.
As such, a course in computational physics/neuroscience, deals with the mathematical ways to solve a problem and isn't concerned with the discipline, the problem arises from?
that is to say, a person with undergrad in math can switch to computatonal neuroscience since he doesn't have to know neuroscience anyway?
i hope someone throws light on this.. thanks!"


Ya that's absolutely not true. As I said I did my undergrad in Computational Physics and I took every physics course that straight physics majors took and I am now doing my masters in theoretical (albeit I still do a lot of computational but my point is that I take the same physics exams and courses that physics majors do). It's not like there's a different GRE for computational.
 

1. What is the difference between computational physics, theoretical physics, and experimental physics?

Computational physics involves using computers and numerical methods to solve physical problems, while theoretical physics involves using mathematical models and equations to understand and predict physical phenomena. Experimental physics involves conducting experiments and collecting data to test theoretical predictions and understand the behavior of physical systems.

2. Which field is more accurate and reliable: computational physics, theoretical physics, or experimental physics?

All three fields have their own strengths and limitations. Computational physics can provide precise numerical solutions to complex problems, but relies on accurate input data and assumptions. Theoretical physics can provide fundamental understanding and predictions, but may not always accurately describe real-world systems. Experimental physics can provide direct observations and measurements, but may be limited by equipment and human error. Therefore, a combination of all three approaches is often used for a more comprehensive understanding of a physical phenomenon.

3. How does the selectivity of each field impact the research process?

Selectivity refers to the ability of a field to choose and focus on specific aspects of a problem. Computational physics allows for precise control and manipulation of variables, making it useful for studying complex systems. Theoretical physics allows for the exploration of fundamental principles and generalizations, while experimental physics allows for direct observation and measurement. The selectivity of each field impacts the research process by guiding the approach and methods used to study a problem.

4. Can these fields be combined in research projects?

Yes, these fields are often combined in research projects for a more comprehensive understanding of a physical phenomenon. For example, a computational physicist may use numerical simulations to test theoretical predictions, while an experimental physicist may use data to validate or refine theoretical models. Collaboration between these fields can lead to a more complete and accurate understanding of a problem.

5. How do advancements in technology impact the selectivity of computational physics, theoretical physics, and experimental physics?

Advancements in technology have greatly expanded the capabilities of all three fields. Computational physics can now handle more complex and realistic simulations, theoretical physics can explore more abstract and advanced concepts, and experimental physics can make more precise and detailed measurements. These advancements have increased the selectivity of each field, allowing for more specialized and focused research in specific areas of physics.

Similar threads

  • STEM Academic Advising
Replies
2
Views
937
  • STEM Academic Advising
Replies
6
Views
1K
Replies
2
Views
892
Replies
5
Views
2K
  • STEM Academic Advising
Replies
10
Views
1K
  • STEM Academic Advising
Replies
14
Views
1K
Replies
8
Views
1K
  • STEM Academic Advising
Replies
4
Views
2K
  • STEM Academic Advising
Replies
11
Views
528
  • STEM Academic Advising
Replies
4
Views
673
Back
Top