Port to Julia?

[azev77]

Hi and thank you for this package!
Have you considered possibly porting it to Julia?
The code would be much easier to maintain & use.

Check this out for POC: https://github.com/JuliaMPC/NLOptControl.jl

[vmbecerra]

Dear azev77 Many thanks for your email and for your interest in PSOPT. We have no plans at present to port PSOPT to Julia. PSOPT is a C++ library and although it can be interfaced to work as part of other platforms such as Matlab or Maple, the PSOPT code will remain for the time being in C++. With best regards Victor Becerra

…

On Wed, 14 Oct 2020 at 09:42, azev77 ***@***.***> wrote: Hi and thank you for this package! Have you considered possibly porting it to Julia? The code would be much easier to maintain & use. Check this out for POC: https://github.com/JuliaMPC/NLOptControl.jl — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#35>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADGCMBNWCHKVQHVON4XHCZDSKVP7DANCNFSM4SQJTDFQ> .

[schulz0r]

It is possible to generate a wrapper for Julia with SWIG. However, I think that wrapping this code is going to be difficult, because PSOPT uses function pointers and other stuff which could be a pain in the neck to wrap to modern languages.
This idea should be kept in the back of the head though, as wrappers can boost the popularity of a project.

[azev77]

C++ programs are often called in Julia. See:
https://github.com/JuliaInterop/Cxx.jl
https://github.com/JuliaInterop/CxxWrap.jl

[vmbecerra]

Hello azev77 When you said 'porting' I thought you meant translating the C++ code into Julia code. One of the difficulties that would emerge with an interface between Julia and PSOPT is that PSOPT requires the problem code (e.g. the differential equations, objective function, and constraints) to be implemented in C++, so that ADOL-C can evaluate them symbolically to compute the required derivatives automatically. If this is not possible, then an interface would be restricted to numerical differentiation, which would be slower. It may be possible to rewrite parts of the code so that the symbolic differentiation occurs in Julia's environment (using symbolic differentiation code that works in Julia, assuming that it exists), and then the required derivatives can be evaluated through calls to Julia from the C++ code. It is not impossible, but it would require a fair amount of work. An interface like that was developed between MAPLE and a previous version of PSOPT some years ago, but it does not seem to be online anymore. Best Victor.

…

On Thu, 15 Oct 2020 at 19:58, azev77 ***@***.***> wrote: C++ programs are often called in Julia. See: https://github.com/JuliaInterop/Cxx.jl https://github.com/JuliaInterop/CxxWrap.jl — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#35 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADGCMBIRTALXEBIA7HOOWNDSK5A5BANCNFSM4SQJTDFQ> .