Plotting even moderately complex 3D Regions (implicit or discretized) is very unstable and it is really a nuisance. It took us eternities to grasp the stuff. Neural Networks work fine but the description of the details is dismal! Please improve with some example networks. CONSĭocumentation is a bit thin, especially newly added functions tend to be poorly described. In addition, we have own packages to generate Gantt charts, project plans, 3D printer code, PCB circuit layouts. Generally, one can get algorithmic packages to run in one working day that take a man month to be implemented in c++. signal processing versus mathematical parameter sets for Fourier transform) and we get exactly the envisaged behavior. In contrast to other commercial available packages the output of integral transforms can be adapted to the culture (e.g. The general strategy is to implement an algorithm in Mathematica and use this as a benchmark when the same algorithm will be transferred to c++. We do lots of RADAR data evaluations, FEM simulations, recently also NN modelling. It is a really sturdy workhorse for any purposes.

I am personally using Mathematica since 1990 using the 'good old' Version 1.2 on a MacIntosh. With Mathematica we have a tool to see almost instantaneously if any new idea has a perspective to work or not. Of course it cannot do real time but that is not the requirement. Mathematica is used as a 'canned data' version to process RADAR data.

Mathematica software student version