![]() ![]() Can anyone comment on that? Correct me if I'm wrong but the SciPy family includes different libraries and packages for different purposes including numerics, graphics and symbolic math. My recollection is that they have better symbolic support than Matlab and Octave. I haven't used the SciPy family of tools nearly as much but only because I have to stop learning new tools at some point and actually get work done. At least a few bars from the train of gold we send them every year goes to that. I say "commercial tech support" because I'm assuming the home use license has limited support but that's just an assumption. I'm not saying those companies have bad support necessarily - just that Mathwork's support is first class. I should say in partial defense of Matlab's cost that their commercial tech support is the best of any I know (Altera, Xilinx, Wolfram, Mentor, JetBrains, off the top of my head). Octave is an unbelievably powerful tool to be available without the cost and chains of proprietary-licensed software. ![]() I still find myself launching Octave at the command line for quick calculations. We use Matlab extensively at work with many add-ons including its VHDL generation tools. As it is I guiltily use it as a mere interactive calculator or a toy. It's such a fantastic software package but I think I'd need a four month retreat to really unleash its potential. The symbolic math was very useful in taming EM fields equations in college but I've never been able to tame the full power of Mathemetica. I've been playing around with Mathematica for the last twenty years (no exaggeration). I know Mathematica has had a home license for a while. You write functions and blocks that work and keep adding them to your library. Leo said that he should be writing the code for Matlab and he is right a custom block-that accurately models hardware is needed. He wrote most of the custom blocks and designed the majority of the DSP receivers for years. The guru in our company was Leo Montreuil. The solution is to write custom blocks either in simple Simulink or with C MEX functions. Also, some of the blocks are just not accurate enough to bit accurately model a comm system. Blocks that were in one location can wind up in a different library in another. Next, Mathworks has an irritating habit of moving and deprecating things esp. It has the faster software FFT version and the remez function will converge much better than the modern PM function. I found that for designing very tight filters Matlab version 6.5 works better than current versions. Neil doesn't agree with me, he always just used the latest version. I do most of my development with this old version and occasionally check the model with the current version so I don't introduce incompatibilities. The newer versions of Matlab seem sluggish by comparison. It runs very, very fast even on a machine choked by a company's IT department's safe-ware. ![]() The installer won't run on a 32-bit machine but through a fairly convoluted method I can install it on Windows 7. My favorite one for Simulink is R11.1 Version 5.3 of Matlab and Version 3 of Simulink. That being said, there are caveats.ġst thing, not all releases of Matlab are created equal. It's a great tool for prototyping digital systems with different clock rates and feedback paths. Simulink was been used in my old group (Scientific Atlanta/Cisco/ST Micro) for years and years. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |