computer language?

  • This has become our largest and most active forum because the physics GRE is just one aspect of getting accepted into a graduate physics program.
  • There are applications, personal statements, letters of recommendation, visiting schools, anxiety of waiting for acceptances, deciding between schools, finding out where others are going, etc.

Post Reply
tengao
Posts: 13
Joined: Sun Nov 04, 2007 4:38 am

computer language?

Post by tengao » Wed Dec 12, 2007 9:48 am

does anyone know what is the most popular computer language(as a graduate majoring in cmp)?
your answer is appreciated :wink:

User avatar
butsurigakusha
Posts: 293
Joined: Sun Oct 07, 2007 8:05 pm

Post by butsurigakusha » Wed Dec 12, 2007 10:44 am

I am pretty sure it is fortran

User avatar
twistor
Posts: 1529
Joined: Thu Apr 13, 2006 2:47 pm

Post by twistor » Wed Dec 12, 2007 11:33 am

I don't know anyone who uses Fortran, but many who use C/C++. Java is important for engineering, and similiar enough to C++ that it may become important in the near future.

User avatar
butsurigakusha
Posts: 293
Joined: Sun Oct 07, 2007 8:05 pm

Post by butsurigakusha » Wed Dec 12, 2007 12:18 pm

Really, you don't know anyone who uses Fortran? I use Fortran, and I was under the impression that it was still the primary language for computationally intensive simulations and modeling. Outside of the science community, it's definitely not used, and C++ and Java are much more popular.

User avatar
twistor
Posts: 1529
Joined: Thu Apr 13, 2006 2:47 pm

Post by twistor » Wed Dec 12, 2007 12:48 pm

Nope. I remember one the professors I work with specifically saying something like "I remember having to learn Fortran to do these things..."

User avatar
butsurigakusha
Posts: 293
Joined: Sun Oct 07, 2007 8:05 pm

Post by butsurigakusha » Wed Dec 12, 2007 1:17 pm

Perhaps it is true that Fortran is no longer used in experimental particle physics. But I know that it is still commonly used in many areas of science, such as molecular modeling, quantum chemistry, climate modeling, etc. because Fortran programs run faster than those written in C or C++, and in these types of simulations, speed is more important than whatever advantages C or C++ have. Plus, Fortran has historically been the primary language for these purposes, and there is a lot of old programming still in use.

User avatar
twistor
Posts: 1529
Joined: Thu Apr 13, 2006 2:47 pm

Post by twistor » Wed Dec 12, 2007 2:04 pm

Projects like Open Computing Grid will soon obviate many concerns with speed....

User avatar
butsurigakusha
Posts: 293
Joined: Sun Oct 07, 2007 8:05 pm

Post by butsurigakusha » Wed Dec 12, 2007 7:38 pm

Yeah, Fortran may eventually die out, but I think there will be areas where it remains strong for quite awhile.

User avatar
twistor
Posts: 1529
Joined: Thu Apr 13, 2006 2:47 pm

Post by twistor » Wed Dec 12, 2007 7:53 pm

If that's true then I think it has more to do with physicists' reluctance to learn a new languate rather than anything special about Fortran.

User avatar
fermiboy
Posts: 437
Joined: Fri Nov 16, 2007 7:41 pm

Post by fermiboy » Wed Dec 12, 2007 10:11 pm

I'll have to disagree with you twistor. I use FORTRAN right now in my research, and I also have a lot of experience with Java, C, and C++. If you want to do numerical computing, FORTRAN beats these languages hands down. Newer versions, like 2003, have objects, pointers, and all that other stuff if you need it. I read somewhere that AMD and Intel still use FORTRAN code to benchmark the floating point performance of their new processors.

The bottom line is there is no one language that's suited to all tasks. I would never write an operating system in FORTRAN, and I would never do numerical computing in Java, etc. Different tools for different tasks, and FORTRAN is still the best tool around for some of these tasks.
Last edited by fermiboy on Wed Dec 12, 2007 11:09 pm, edited 1 time in total.

User avatar
twistor
Posts: 1529
Joined: Thu Apr 13, 2006 2:47 pm

Post by twistor » Wed Dec 12, 2007 10:15 pm

I didn't think it was still widely used. Maybe it's worth learning then. I have an ancient (1986) book on scientific computing where most of the examples are written in Fortran. I figured it was hopelessly outdated. It also has examples in Pascal, but I think we can all agree that Pascal's time has come and gone....

User avatar
fermiboy
Posts: 437
Joined: Fri Nov 16, 2007 7:41 pm

Post by fermiboy » Wed Dec 12, 2007 10:49 pm

I'll agree with you about Pascal. The Pascal of today is Java. It is a common misconception that FORTRAN is no longer in use. I thought so until my adviser told me to learn it. I discovered that not only is it still in use, they are still making new versions, 2003 was the last, and 2008 is in the works. The newer versions don't have all the archaic syntax like FORTRAN 77 and before. I have grown to really appreciate it's use for numerical stuff, especially if you want a small program to do something like a finite difference method. Plus in addition to integers and floats, it has an intrinsic complex number data type, how cool is that?

tchotchke
Posts: 6
Joined: Sun Oct 07, 2007 6:37 pm

Post by tchotchke » Thu Dec 13, 2007 1:08 am

I am currently learning lisp even though I am pretty sure no one outside of computer science (AI, specifically) uses it. Its a really interesting language and it changes the way you think about programming. Its about half as fast as C++, which is not that bad, but for numerics you might want to stick to C++ or fortran, even though personally I think learning lisp is worth it for the cool factor.

Oh, and talking about old, I believe it has been 50+ years since lisp was invented.

User avatar
butsurigakusha
Posts: 293
Joined: Sun Oct 07, 2007 8:05 pm

Post by butsurigakusha » Thu Dec 13, 2007 1:23 am

By cool factor, do you mean lisp is cool, or that knowing lisp will make you cool?

tchotchke
Posts: 6
Joined: Sun Oct 07, 2007 6:37 pm

Post by tchotchke » Thu Dec 13, 2007 1:51 am

that lisp is cool

User avatar
twistor
Posts: 1529
Joined: Thu Apr 13, 2006 2:47 pm

Post by twistor » Thu Dec 13, 2007 10:38 am

LISP will definately help when you try to use emacs.

User avatar
fermiboy
Posts: 437
Joined: Fri Nov 16, 2007 7:41 pm

Post by fermiboy » Thu Dec 13, 2007 6:22 pm

Knowing LISP will also help you pick up chicks.

schandre
Posts: 55
Joined: Fri Nov 09, 2007 3:43 pm

Post by schandre » Thu Dec 13, 2007 6:48 pm

Knowing LISP will also help you pick up chicks.
So LISP is a language definitely worth learning. I wish I had been told this earlier.

User avatar
twistor
Posts: 1529
Joined: Thu Apr 13, 2006 2:47 pm

Post by twistor » Thu Dec 13, 2007 7:35 pm

Didn't you know: chicks love a man with a LISP!

User avatar
will
Posts: 399
Joined: Sun Nov 04, 2007 3:26 pm

Post by will » Fri Dec 14, 2007 7:47 pm

Python with NumPy is also getting to be pretty widely used by younger professors who do numerics. Yes fortran comes out with new versions, but the people that really use fortran use fortran 77. And you should definitely learn it, it's not so bad.

If you're really interested in programming though, outside of just numerics, Python is awesome.

User avatar
fermiboy
Posts: 437
Joined: Fri Nov 16, 2007 7:41 pm

Post by fermiboy » Fri Dec 14, 2007 8:57 pm

Is that so? Hmmm, my supervisor and I use FORTRAN 90 to solve nasty PDES with finite difference methods all day long. I think we are really using it, and it ain't FORTRAN 77.

User avatar
will
Posts: 399
Joined: Sun Nov 04, 2007 3:26 pm

Post by will » Sat Dec 15, 2007 1:11 pm

fortran 90 is still almost two decades old... And yes, there's people that use the newer versions. You are one of them. Most people don't.

tokamak
Posts: 25
Joined: Sat Nov 17, 2007 4:40 pm

Post by tokamak » Sat Dec 15, 2007 4:32 pm

nobody mentioned anything about IDL....I wonder why...



Post Reply