yade-dev team mailing list archive

[Jerome Duriez <jerome.duriez@xxxxxxxxx>]

Hi,

For your information, I also faced at some point segfaults with 
boost::python stuff during O.run() and not O.step() : 
https://www.mail-archive.com/yade-dev@xxxxxxxxxxxxxxxxxxx/msg13223.html

(I never really solved / understood it.. If you know e.g. boost mutex 
more than me -- that won't be difficult -- maybe you will !)


Jérôme

------
Chargé de Recherche / Research Associate
Irstea, RECOVER
3275 route Cezanne – CS 40061 13182 Aix-en-Provence Cedex 5 FRANCE
+33 (0)4 42 66 99 21
www.irstea.fr/duriez

On 07/10/2019 16:46, William Chèvremont wrote:
> Dear Yade dev,
>
> I'm facing an issue implementing a new interaction law into Yade.
>
> To be quick, I would like to provide normal force norm from python and 
> integrate this value into the lubrication law.
>
> The current state of my work is on branch 'potential' on gitlab. 
> What's new in this branch:
>
> - pkg/dem/LubricationWithPotential.{cpp, hpp} : Lubrication law with 
> arbitraty contact law + potential solver, and some potential coded in 
> c++. This part work perfectly.
>
> - py/wrapper/yadeWrapper.cpp: I've written a wrapping class and 
> boost.python exposing expressions. I want to inherit this class in 
> python in order to provide the value to the solver above.
>
> It works perfectly if I call O.step();. Problems are coming as soon as 
> I launch O.run(), I got a segfault signal when the c++ code call the 
> python overrided function.
>
> What's wrong???
>
> Summary of the code:
>
> :::python:::
>
> classSimplePotential(GenericPotential):
>     def contactForce(self, u):
> #        print("pyContactForce!");
>         return 0;
>     def potentialForce(self, u):
> #        print("pyPotentialForce!");
>         return 1;
>     def hasContact(sefl, u):
>         return False;
>
> :::C++:::
>
>
> class pyGenericPotential : public GenericPotential, public 
> py::wrapper<GenericPotential> {
> public:
>
>     Real potential(Real const& u, LubricationPhys const&) const {
>         TRACE;
>         return contactForce(u) + potentialForce(u);
>     }
>
>     void applyPotential(Real const& u, LubricationPhys& phys, Vector3r 
> const&n) {
>         TRACE;
>         phys.normalContactForce = contactForce(u)*n;
>         phys.normalPotentialForce = potentialForce(u)*n;
>         phys.contact = hasContact(u);
>     }
>
>     virtual Real contactForce(Real const& u) const {
>         TRACE;
>         return get_override("contactForce")(u);
>     }
>
>     virtual Real potentialForce(Real const& u) const {
>         TRACE;
>         return get_override("potentialForce")(u);
>     }
>
>     virtual bool hasContact(Real const& u) const {
>         TRACE;
>         return get_override("hasContact")(u);
>     }
> };
>
>
>     py::class_<pyGenericPotential, boost::noncopyable>("GenericPotential")
> .def("contactForce",py::pure_virtual(&pyGenericPotential::contactForce),"This 
> function should return contact force norm.")
> .def("potentialForce",py::pure_virtual(&pyGenericPotential::potentialForce),"This 
> function should return potential force norm.")
> .def("hasContact",py::pure_virtual(&pyGenericPotential::hasContact),"This 
> function should return true if there are contact.");
>
> --
> <http://www.univ-grenoble-alpes.fr>
> 	
>
> William Chèvremont
>
> Doctorant, PhD student
> 04 56 52 01 86
>
> Laboratoire de Rhéologie et Procédés
> Bureau Bureau B-365
> 363 Rue de la chimie - bâtiment B
> Domaine Universitaire - BP 53 - 38041 Grenoble cedex 9
> www.univ-grenoble-alpes.fr <http://www.univ-grenoble-alpes.fr/>
> http://www.laboratoire-rheologie-et-procedes.fr/
>
>
> _______________________________________________
> Mailing list: https://launchpad.net/~yade-dev
> Post to     : yade-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~yade-dev
> More help   : https://help.launchpad.net/ListHelp