I wonder if (https://www.dynare.org/ is a good choice for being included in Fedora, or on the contrary, the recent developments with Python (with the Dolo package) make this choice unnecesary in the treatment of the class of dynamic general equilibrium macroeconomic models. In my experience, which is not much (10 months with Silverblue) it’s very difficult to plug Dynare into the system. I attempted at it because of the amazing ‘academic’ library embedded Dynare.
However, the idea of importing systems form Matlab to Octave and then compile in fortran, C, C++, with the intervention of Intel at some point, which in turn connects to Eclipse, when (enough with Jupyter by now) seems a bit twisted and forced. The Intel intervention is to run the variant Dynare++, which seemingly can significantly improve the calculations.
As far as I know, the only real contribution of Dynare with respect to the Python initiative comes from the academic libraries, as pointed above. That means in practice a series of mod files that can seemingly been replicated for any advanced student with a yaml file. These can be run in the dolo library. Dynare users seem to have either more expertise in computing (mine is limited, as you can see) or more infrastructure than I have.
@mosiu: sorry, how is VAT involved here? If that is something specific to dynare, it’s probably best to discuss that on the dynare channels. This thread is merely limited to including dynare in Fedora (if possible).
Thanks for your reply. I am just learning to install Dynare. There is an alternative in Python with the dolo package. Both use some specific graphical libraries and parallel computing, so it is not possible to install dolo with pip from Fedora, at least in Silverblue (33). The documentation of dolo says that a scientific package, like Anaconda, is required to proceed with the installation, so this info is not a particular issue of any kind.
I think it’s not a good idea to mix statistics projects involving say R, Python an Jupyter with others engaged in Dynare or dolo even though both can be run under Jupyter. I have just decided to create two different toolboxes for these two classes of projects. Two days ago I was been trying within a single toolbox. Yesterday I installed dolo within conda successfully, and I am ready to begin the installation of Dynare inside the conda toolbox.
Another toolbox was just crated for LaTeX and so on. It should be easy to create a Dockerfile to automatize this software. In a classroom it can be an asset, because separate installation seems very robust (save for Dynare, which is to be proven), have no further complications and the dnf and pip history is easy.
For the reasons stated above, a simpler solution than docker or even podman alone is that partners (firms, customers[?], professors, students) explore with OS similar to Fedora Silverblue, not even Fedora. If I know in the future of ways of installing dolo and Dynare with a minimal architecture, I’ll tell, but I am not even sure whether I am much interested in that if the conda toolbox happens to run well. Indeed there is an issue because the navigator is not launched, but I can’t remove it, as well as other few useless apps.