Is it not possible to have periodic boundary conditions and DG elements?
I am trying to solve a bunch of advection-diffusion-reaction equations using 'DG' elements in one and two dimensions, and in a periodic domain (to begin with).
The following code
from dolfin import *
class PBC1D(SubDomain):
def __init__(self, xmin=0, xmax=1):
SubDomain.__init__(self)
assert(xmax>xmin)
self.xmin=xmin
self.xmax=xmax
def inside(self, x, on_boundary):
return bool(near(x[0], self.xmin) and on_boundary)
def map(self, x, y):
y[0] = x[0] - self.xmax + self.xmin
mesh = UnitIntervalMesh(64)
pbc = PBC1D()
X = FunctionSpace(mesh, "DG", 1, constrained_domain=pbc)
generates this error
[turing:16366] *** Process received signal ***
[turing:16366] Signal: Segmentation fault (11)
[turing:16366] Signal code: Address not mapped (1)
[turing:16366] Failing at address: 0x7f2800000020
[turing:16366] [ 0] /lib/x86_64-linux-gnu/libpthread.so.0(+0xfbd0) [0x7f2880ac4bd0]
[turing:16366] [ 1] /usr/lib/libdolfin.so.1.2(_ZN6dolfin13DofMapBuilder32build_constrained_vertex_indicesERKNS_4MeshERKSt3mapIjSt4pairIjjESt4lessIjESaIS5_IKjS6_EEERSt6vectorImSaImEE+0x110) [0x7f287a2ea960]
[turing:16366] [ 2] /usr/lib/libdolfin.so.1.2(_ZN6dolfin13DofMapBuilder9build_ufcERNS_6DofMapERSt3mapIiiSt4lessIiESaISt4pairIKiiEEERKNS_4MeshEN5boost10shared_ptrIKS3_IjS3_IjS6_IjjES4_IjESaIS6_IKjSH_EEESI_SaIS6_ISJ_SM_EEEEENSG_IKNS_11RestrictionEEE+0x18f) [0x7f287a2edd3f]
[turing:16366] [ 3] /usr/lib/libdolfin.so.1.2(_ZN6dolfin13DofMapBuilder5buildERNS_6DofMapERKNS_4MeshEN5boost10shared_ptrIKSt3mapIjS8_IjSt4pairIjjESt4lessIjESaIS9_IKjSA_EEESC_SaIS9_ISD_SG_EEEEENS7_IKNS_11RestrictionEEE+0xfb) [0x7f287a2f00bb]
[turing:16366] [ 4] /usr/lib/libdolfin.so.1.2(_ZN6dolfin6DofMapC1EN5boost10shared_ptrIKN3ufc6dofmapEEERKNS_4MeshENS2_IKNS_9SubDomainEEE+0x678) [0x7f287a2cfb78]
[turing:16366] [ 5] /usr/lib/python2.7/dist-packages/dolfin/cpp/_fem.so(+0x7ce78) [0x7f285d254e78]
[turing:16366] [ 6] python(PyEval_EvalFrameEx+0x216a) [0x47d80a]
[turing:16366] [ 7] python(PyEval_EvalCodeEx+0x19a) [0x4e09aa]
[turing:16366] [ 8] python() [0x4e140a]
[turing:16366] [ 9] python() [0x4cabfe]
[turing:16366] [10] python() [0x49ebf4]
[turing:16366] [11] python() [0x4b1a1e]
[turing:16366] [12] python(PyEval_EvalFrameEx+0xafd) [0x47c19d]
[turing:16366] [13] python(PyEval_EvalCodeEx+0x19a) [0x4e09aa]
[turing:16366] [14] python() [0x4e140a]
[turing:16366] [15] python() [0x4cabfe]
[turing:16366] [16] python(PyEval_EvalFrameEx+0xafd) [0x47c19d]
[turing:16366] [17] python(PyEval_EvalCodeEx+0x19a) [0x4e09aa]
[turing:16366] [18] python() [0x4e153a]
[turing:16366] [19] python() [0x4cabfe]
[turing:16366] [20] python() [0x49ebf4]
[turing:16366] [21] python() [0x4b1a1e]
[turing:16366] [22] python(PyEval_EvalFrameEx+0xafd) [0x47c19d]
[turing:16366] [23] python(PyEval_EvalCodeEx+0x19a) [0x4e09aa]
[turing:16366] [24] python(PyEval_EvalCode+0x32) [0x540412]
[turing:16366] [25] python() [0x54088b]
[turing:16366] [26] python(PyRun_FileExFlags+0x92) [0x4658f6]
[turing:16366] [27] python(PyRun_SimpleFileExFlags+0x2d8) [0x465e26]
[turing:16366] [28] python(Py_Main+0xb4e) [0x466b9b]
[turing:16366] [29] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5) [0x7f287fdebea5]
[turing:16366] *** End of error message ***
Segmentation fault
Change the element to 'CG' and everything goes through smoothly. However neither DG0 nor DG1 seem to work.
I have no idea what is happening. Perhaps I am missing something which I do not understand.
Is there any demo code for solving a transient advection-diffusion equation using DG elements? In periodic domains?
