I have been using MATLAB to code, but I seldom use the MATLAB tools and I always code everything by myself. I am really used to MATLAB and I am not sure whether it is enough to code only on MATLAB. I know MATLAB is not as fast as C++ or Fortran now. But is it possible that MATLAB will be updated and the running speed is increased to C++?
It is highly unlikely MATLAB will ever be used as widely as fortran or c++, since it is proprietary and buying a license for the MATLAB interpreter is quite expensive. There is a free clone, called Octave, that is very similar to MATLAB but it is not commonly used. MATLAB is also not a general purpose language, it was developed specifically to do linear algebra, scientific computing and engineering.
Python is probably the fastest growing language right now.
I agree with Jonas. Python is open source MATLAB is not and requires a license. This makes a big difference in sharing code. The popularity and growth of Python in the science community makes it a safe choice for the future.
Ignore for a moment that Matlab has an open version, octave. Matlab is designed for matrix and mathematical solutions. C++ was designed to be an improvement on C, and such is designed with comp sci and general computing in mind. Two different tools that overlap in some functionality. Like a drill and electric screwdriver.
Now between python, Matlab and R, that's a different question...
Matlab/octave are wonderful for general scientific computing AND data visualization and analysis. They are also great to develop strategies to deal with the data before an implementation in a compiled program. The manipulation an visualization of small and medium amount of data (
Answering your question: because MATLAB is an interpreted laguange it will never be as fast as C, considering both programs are written and optimized to their full extent.
It's not about a programming language replacing another. It is about finding the most appropriate language for whatever it is you are doing. Speed is not the only thing to consider when writing code, there are other things that may be more important, design, efficiency, collaboration, deployment, code maintenance, longevity, etc...
You are right to be worried about only using MATLAB to code. As a programmer, you should not be fixed on one language... you should know how to code (period), not how to code in MATLAB only. You should be able to learn different languages for different purposed and use the most optimal one for each task. Also, coding everything yourself could also be a bad practice. Time is the most important thing you have, *your time* (not computing time)... When working with open source languages you will often find packages or scripts from other people around the world that may have developed the exact tool you are looking for, in a very efficient and abstract way. So you don't really need to write everything yourself, you can use a package someone else developed and focus your time on the real important thing you are tackling or trying to solve.
I would recommend that you take a look a the Julia language (www.julialang.com). It's an open source language, specifically designed for the 2 language problem you are referring to. I moved from MATLAB to Julia, a few years ago and I did it because I needed a faster version of my program but didn't want to rewrite the whole program. Julia tries to bring the best of both world, the easy syntax of MATLAB and the speed of C, among MANY other things.
There may be two use-case spaces where performance criterion becomes critical.
1. The analysis code handles large data-sets. [ ~ O(petaBytes) ]
2. Numerical experiments those need to be re-run several times [ ~O(10^6) runs]
In both cases, it's the computing time that prevails the choice of tool/language. Matlab offers parallel implementations which could serve such purposes provided it is setup with a cluster.
Another alternative is ROOT [https://root.cern.ch]
It also has python bindings [https://root.cern.ch/pyroot] which combines best of both worlds. Python front-end conserves developer's time during prototyping, while C++ back-end does the heavy lifting under the hood.
- Badges
- Science topic
- Similar topics
- Comment