I concur. I recently finished my Ph.D. in applied (computational) physics, focusing on Materials Science. I'm now a postdoc in a very well regarded engineering program, doing an entirely different kind of computational work (NEGF quantum transport). The most important trend I have seen across several disciplines is the practice of writing highly efficient, modular, numerical libraries (e.g. molecular dynamics, kinetic Monte Carlo, ab initio first principles, linear solvers) in C or Fortran, then creating wrappers for those libraries in Ruby, Perl, Python, Tcl, Matlab, etc. One can quickly write very flexible and adaptable tools without delving into the complex guts of a code. Ruby makes this process so easy, and the language itself is beautiful. I write all my top level simulation tools in Ruby, calling C libraries, and the data analysis is easily done in Ruby as well.
I agree. For hardcore numerical work, most people
write custom code. In our group, it's C/Fortran 90 with LAPACK, FFTPACK, mpi,... libraries. For data analysis and visualization, Mathematica and IDL are our tools of choice.
Slashdot Mirror
I concur. I recently finished my Ph.D. in applied (computational) physics, focusing on Materials Science. I'm now a postdoc in a very well regarded engineering program, doing an entirely different kind of computational work (NEGF quantum transport). The most important trend I have seen across several disciplines is the practice of writing highly efficient, modular, numerical libraries (e.g. molecular dynamics, kinetic Monte Carlo, ab initio first principles, linear solvers) in C or Fortran, then creating wrappers for those libraries in Ruby, Perl, Python, Tcl, Matlab, etc. One can quickly write very flexible and adaptable tools without delving into the complex guts of a code. Ruby makes this process so easy, and the language itself is beautiful. I write all my top level simulation tools in Ruby, calling C libraries, and the data analysis is easily done in Ruby as well.
I'm waiting for the one called Balrog.
Hide the dog well.
You'll never find him now! HAHAHA!
I agree. For hardcore numerical work, most people write custom code. In our group, it's C/Fortran 90 with LAPACK, FFTPACK, mpi, ... libraries. For data analysis and visualization, Mathematica and IDL are our tools of choice.
rm -rf /