How important is MATLAB

Forum: PC programming I hate matlab





I don't really know, maybe you should deal with it more intensively, because the people in my vicinity are almost exclusively fans. I haven't seen any software myself that is more practical for an electrical engineer.



I love matlab! As I know this forum, it is now a neverending thread. Incidentally, I also have to mention: "I hate fraud".



Matlab is damn exhausting and inefficient. If you haven't got it right. Once you've got your bearings and come to terms with the few really crazy idiosyncrasies, then it's okay. It is stupid that (mostly) Java is under it. Otherwise ... usable. Matlab is somehow the intermediate form between programming and spreadsheets. Very useful for recreational data jugglers (scientists, ...), quickly annoying with really complex things. So much for my opinion and experience. Edit: Of course, not everything in Matlab is based on Java. It's just the language that the tormented souls of the users mostly also had to learn.

: Edited by user


Why get excited about Matlab when you can also mess with ROOT ...



fraud wrote:> who is still feeling like that? Counter question, with what can it be replaced? All the tools from MathWorks are currently hard to replace for people who want to model, simulate, test, design and program quickly. Very practical, especially for prototypes but also for series development.



Just use Octave for a while. Then you will be happy about Matlab ;-)



Unbeatable in combination with Simulink. Design control loops, simulate, test, ... I think it's great. Unfortunately, never used again after graduation.


First use LabView and then say again that you cannot cope with Matlab! They are different programs, but how you can work with LabView remains a mystery to me in the chaos. Matlab can do everything !! besides cutting bread.



I think Matlab is great too. Had already enjoyed working at the university 10 years ago. Everything under one roof in one IDE. Otherwise check out Scilab or Octave. If you don't want it to be as performant Python with Numpy.



Simon wrote:> Counter question, what can be replaced with? All of the tools from MathWorks> are currently hard to replace for people who want to model, simulate, test, design and program quickly. Very practical, especially for> prototypes, but also for series development. In the industry it is sometimes impossible to imagine life without it, because you can work professionally with it, and you can tell from the list prices that there is currently no alternative.



Recently, I actually met a colleague who wasn't familiar with Matlab. When someone cannot use Matlab, for me it is like when someone says they cannot use the calculator.
1Matlab is the standard tool for engineers.
Alternatively, you can of course use Octave: https://www.gnu.org/software/octave/ That has become pretty good by now. It's a shame that there is no such thing as Simulink. Some colleagues also have a little linguistic confusion: They work with Simulink, which is also from Mathworks, and then refer to Simulink as Matlab. But Matlab is not Simulink. Just as a car tire is not a car.



chris_ wrote:> That has become pretty good in the meantime. It's a shame that there isn't> something like Simulink. Then you take SciLab with Xcos.


MATLAB gets ugly if you want to do something other than Simulink, DSP, or matrix calculation with it. Then you are faced with a language with a standard library that is similarly "well" structured as that of PHP - not at all. Loads of functions in the global namespace. Then there are strange scoping rules (first function in a file global, all others local). It also gets very ugly when people think they are doing bit operations in MATLAB. Where every other language has binary and hexadecimal literals, MATLAB comes with a stall of functions, bitshift, bitand and so on. The not serious object orientation is just as bad. MATLAB can probably do OOP, but nobody uses it, especially if external hardware (measuring devices, etc.) would be appropriate. If you don't need Simulink or special toolboxes, you are better served with python, numpy and matplotlib in jupyter notebook or spyder.



chris_ wrote:> Recently I actually met a colleague who> did not know Matlab. These are then the candidates who are just not allowed to become unemployed ...



annoyed wrote: >> Recently I actually got to know a colleague who >> did not know Matlab. >> Those are the candidates who are just not allowed to become unemployed ... I got to know colleagues who believe that symbolic calculation means searching - Replace in Excel. And that wasn't fun. So top hardware developers. Simon wrote:> Up to now I have not seen any software> that would be more practical for an electrical engineer. For the electrical engineer, the programs would be important in the following order: LTSpice KiCAD (wx) Maxima ... that can calculate symbolically Scilab XCOS And everywhere in between Code :: Blocks and a reasonable editor.



I have to say I have a love-hate relationship with Matlab. Negatives: * The plots are ugly. (matplotlib, pygal ...) * You have to pay extra for the interesting toolbox. * How can indices only start at 1 ... * Column and row vectors exist, and are also processed inconsistently by functions * Functions sometimes implicitly adjust matrices that match (plot ()). ((If necessary, 10x20 becomes 20x10. If the data then happens to be square, everything flies in the air. * Strings by default are vectors from characters, so you have to switch to cell arrays ?! * Why in heaven's will does "1: 1: 5" become a 1x5 and not a 5x1 vector ?! * No break points in sub-functions ?! Positive: * Somehow it works * GUI is usable * There is almost nothing there (for money) not there. * The documentation is not bad at all. * Excel vs. Matlab is like sand shovel vs. wheel loader when it comes to measuring data analysis.

: Edited by user


> * Excel vs. Matlab is like sand shovel vs.> wheel loader when it comes to the evaluation of measurement data. More like hand ax and moon rocket.



Simon wrote:> I don't really know, maybe you should> deal with it more intensively, because the people in my immediate vicinity are almost> exclusively fans. I myself have not seen any software> that would be more practical for an electrical engineer. I'm a chemical engineer, but I've programmed all sorts of stuff in the past ... with JAVA, with Pascal, Kobol, Visual Basic, C, C ++ .... But none of them were as "cumbersome" and not as "logical" as Matlab . The most fun was C ...



PHP * used to be a small tool for a limited purpose and is now a pieced together collection of all the features that anyone has needed over the past few decades. * is used by people who don't want to program, just the results. Often by bloody autodidacts. * is often the only programming language mastered. * seems logical and well thought-out if you don't know anything else. * has annoying hooks and edges in all places. * has non-replaceable killer applications that are based on it. * enables good programming. * makes bad programming easy. * is industry standard. Matlab * used to be a small tool for a limited purpose and is now a pieced together collection of all the features that someone has needed in the past few decades. * is used by people who don't want to program, just the results. Often by bloody autodidacts. * is often the only programming language mastered. * seems logical and well thought-out if you don't know anything else. * has annoying hooks and edges in all places. * has non-replaceable killer applications that are based on it. * enables good programming. * makes bad programming easy. * is industry standard.



Thomas wrote:> has annoying hooks and edges in all places. I can agree with that. I wanted to add up some values, but only 255 came out. It took a while until I found the error at that point. Then I tried to fix that. The variable was automatically cast to eight bits, even though the total was in the four-digit range. I tried to create the variable with a different type, I initially assigned the variable a value of 0.01, which was also saved, but after the first addition it was always an eight bit integer type. That's really stupid. If I explicitly say that floating point numbers should be put in and that Matlab simply deletes the contained value after an addition and turns it into an integer, that's very stupid. The solution was that I cast the complete summand matrix. If you have variables with different value ranges, it is of course logical to assume the smallest value range for the result ...



Simon wrote:> fraud wrote: >> who is still feeling like that? >> counter-question, what can it be replaced with? That's a good question! Is Matlab really that widely used in industry? Martin P. wrote:> First use LabView and then say again that you cannot> cope with Matlab! >> They are different programs, but how you can work with LabView> remains a mystery to me with the chaos. I worked with DiaDem during my bachelor thesis and that was pretty shitty ^^ Matlab lover wrote:> I also have to mention: "I hate fraud". That is granted to you; = D Lukas T. wrote:> Matlab is damn exhausting and inefficient. If you don't> properly master it. >> If you've got your bearings and can deal with the few really crazy> idiosyncrasies, then it's okay. Yes, the crazy idiosyncrasies bother you. In this respect, C is designed to be much more logical and "catchy" ...> Matlab is somehow the intermediate form between programming and> spreadsheets. For recreational data jugglers (scientists, ...)> very useful, quickly annoying with really complex things. >> So much for my opinion and experience. As a matter of fact.



In my old company we did a lot in development with Matlab. Matlab is awesome in the prototype phase, development and research. BUT it is not a programming language. Unfortunately, many people abuse it for this purpose. The problem is with the nature of the matter. I'll start with a simple problem, e.g. have measured a sample with a sensor. That's all for 50 samples. What do I need: an import of the sensor data into a vector. This is quite uncomplicated with Matlab, with CSV, TXT, etc. Files also super easy via GUI or with 2 lines of code. Then I do an FFT for each measurement, remove the background noise and use the software to find the 3 largest frequencies. I output the results both graphically (plot of the sensor data, plot of the FFT, mark the 3 peaks with an arrow or circle) and as a table and save this as a DOC or ODT file. Time required, depending on the level of knowledge, 0.5h - 3h. Matlab is a good documentation, it's a scripting language where I don't have to compile anything, it's own debugger and by default I get the result pretty quickly. If I need the above construct in production or QM, it looks a little different. I need a GUI or something that Hans-Wurs can use. I can do it with Matlab, but ne GUI development, nested functions, etc. take some getting used to, as mentioned in other posts. I also think that the above project is not built in 3 hours in C, C ++ or similar. Reading in the data is still quite easy. It is more problematic to handle faulty files correctly. FFT, yes there are libraries. But I have to learn and understand them first. Vector arithmetic in C / C ++ works, but I also need a good library to not make my life too difficult. Plot, also again with ext. Library, Word Export ... I had no idea how to do it properly. If I do something like that every day, I'll be through it in 3 hours. But then I have to know the tools exactly beforehand.



Andreas R. wrote:> * Excel vs. Matlab is like a sand shovel vs.> wheel loader when it comes to measuring data analysis. So much for D'accord. Yesterday I did PCA with Matlab ... it's quite uncomplicated with princomp from the statistics toolbox. Would that also work with Excel or VBA?