ECAD approach relevance in electronic-engineering problem solving

Often the engineering problem's formulations are very complex and the formalism very strict, in this case the use of the computer, for calcolus, is required. Neverless the engineer work is solve enquations!!!

The use of ODE or PDE solution's tool as well as non-linear systems solution's methods also require a study of the program that we want use, a good problem formulation (often different from original one) and many attention to the numeric relevant questions as precision or accuracy. Mathematical programs that we can find in the scientific landscape can use significatly different approach to solve problems. For example Matlab prefer work with a numeric approach, Mathematica (Maple's son) with a symbolic one. Strictly Matlab can define symbolic objects but develop a script (or a script-architecture) with this entities is nasty idea, imho, the syntax is very weigthy and induce the programmer to errors. Mathematica instead use infinity precision algorithms therefore simply can manipulate symbolically espressions, equations or data lists. Obviously I don't say that Mathematica is better of Matlab/Maple but just that for modellistic or problem-solving approach The first is more suitable. Matlab instead have a lot of toolbox designed for may applications, that totally miss in Mathematica, whose the designers can't deprive (as Symulink). We must be able to discern when it's convenient to use either.

Chosen our program we can solve a problem, however, first, we must formulate the problem that include the models of system (typically active device for electronic-engineering problems).

For make a model often a measurement's campain is required and there are several of techniques that aid the engineer to find the model like as describing functions approach or empirical ways.

Identified the model we can formulate the problem whose formalism contains our equations for solve it.

The choise of solving algorithm is another important question's aspect in fact dependently on nature of system or of the problem can be more suitable one approach instead another one. For example if we want study a stiff problem we can use trapeziodal method for integrate the system because this method is stable (for stiff system) and reduce error and so on...

Another aspect that deserve to be remembered is the relevance of termination's criteria. They must to be suitable for the resolution of the problem and can require a separate study.

Obviously we can redo the calculus-flow how many times we want (changing, for example, a lot of parameters or somenthing else...).

The advandages of numerical solving are that we can transfer the algorithms in an embedded machine or in an automatic control systems (pc and process-systems), in actual hardware landscape the choise can fall on microprocessor/microcontroller, FPGA, ASIC or SoC, Fusion technologies.

Thank you for reading.