Hacker News Reader

The secret to Modelica's results here isn't the numerics but the symbolics: they always preprocess the equations to something much simpler to simulate. With Julia we have the https://github.com/SciML/ModelingToolkit.jl to bring the same kind of symbolic-numerics approaches to practice.

View on HN

Chris, I agree that the symbolic processing is quite important. I've often made the argument that a purely numerical approach just can't compete because of the optimizations and insights that come out of the symbolic analysis will always put it on a better footing (curious to know if you agree). But I'd say another factor in the success of Modelica is the ability to make it approachable by non-programmers. All the graphical aspects, acausal semantics, etc. make it easy to understand for non-programmer engineers as well. This has nothing to do with the results, per se, but is I think a factor in its success.

Reply on news.ycombinator.com