Hello,
there seems to be a mean obstacle for people who intend to combine discretization schemes based on PolynomialFit and dynamic meshes in general. All simulations where I have tried to use this combination failed with an empty autoPtr as follows:
--> FOAM FATAL ERROR: object is not allocated From function const T& autoPtr<T>::operator()() const in file /opt/foam.debug/foam-extend-3.1/src/foam/lnInclude/autoPtrI.H at line 132. FOAM aborting Aborted (core dumped)
For me it seems, that some stencil? object is spuriously cleared after/with? the first mesh update. Sorry for this superficial analysis, but for me the whole code related to dynamic meshes is still a "closed book".
Even though it is quite "simple" to reproduce the problem with any? *DyMFoam solver, I have attached two selected test cases including output and stack trace:
Please correct me if I'm wrong or if this functionality is desired/not implemented.
Best regards,
Pascal Beckstein