Hi there,
As said on my previous blog post, today I’m talking more specifically about my GSoC 2017 project, named “OpenModelica Integration [for Scilab]”.
“So, what is this you plan to integrate ?”
Imagine that you need to simulate a physical phenomenon, or implement any sort of equation, in your preferred C-like computer language (e.g. C++ or Python)… Simply describing it as
a = b * c + d
only allows you to obtain a, given all the other values, right ? In this format, you couldn’t calculate c, for instance, just by having a, b and d. If you need to obtain different variables at different processing steps, the equation must be arranged in all correspondent ways:
| 1) a = b * c + d | 2) b = (a - d) / c | 3) c = (a - d) / b | 4) d = a - b * c |
|---|
That’s called causal modeling: one have to describe every needed different relation between these terms.
When you think about it… It kind sucks for those types of problems, am I right ?
“But what can we do about it ?”
I’m glad you asked that (maybe not, but…). Thankfully, some people have already come up with solutions for acausal modeling, where you just define equations in one way and the other are implicitly derived from that. One of the most well-known tools for that is called Modelica.
Not really a tool, but a standard. Have I ever said I love standards ?
Anyways, Modelica is an object-oriented and declarative language specification for description/modeling of dynamic systems (mechanical, electrical, thermal, mixed, etc.). Being domain neutral, different compiler and simulation engine implementations use the information written in a Modelica file (.mo) to generate source code for multiple programming languages (C, C++, C#, Java, etc.), used for actual processing.
(Modelica text and graphical (blocks) representations of an electromechanical system)
Currently, there are a lot of actual simulation tools (commercial and free) which implement the Modelica text and graphical languages, like CATIA Systems, Dymola, LMS AMESim, JModelica, MapleSim, OpenModelica, SimulationX, Wolfram and, guess what, Scilab.
In Scilab, Modelica language is used to describe more complex Xcos blocks (mainly from Coselica package), taking advantage of existing compiler/translator implementations to generate correspondent C code, in order to integrate those components into complex Xcos simulations.
I have even used that in one of my academical works:
(Can you notice the green blocks ? They are Modelica blocks integrated into a Xcos model)
Nowadays, Modelica in Scilab is managed by an old LMS compiler, released under the GPL license, which is not actively maintained (needing out-of tree patches) and support only a subset of the current language specification. To address that, Scilab developers proposed modifications to allow the usage of OpenModelica’s OMCompiler (omc) as an alternative. As a more up-to-date solution, omc integration would provide Scilab/Xcos the ability to use more advanced Modelica features, and avoid bugs, issues and maintenance burden of the current modelicac compiler.
That represents a pretty important task, but someone needs to step up in order to make it happen… Guess who is willing to try ?
The following months are going to be a tough test for my software development skills. I hope to succeed to some extent.
Next time, I’ll bring you something more detailed about how OMCompiler works.
Thanks for sticking with me. See ya !