Graph Data Conversion

To run the test case, simply execute:

cd ${OPENDIS_DIR}
cd examples/02_frank_read_src
python3 -i test_data_convert.py

Expected Behavior

This test case first run the same simulation as in test_frank_read_src_exadis.py, and then converts the dislocation network (graph) to various formats and check for self-consistency. If all tests are successful, you should see

1. test G1 sanity check PASSED
2. test G1_eq_G2 PASSED
3. test G1_eq_G4 PASSED

Conversion between ExaDiS and PyDiS using DisNetManager

At the end of the simulation, the final dislocation network is managed by the DisNetManager called net. The following line in the python code

G0 = net.get_disnet(ExaDisNet)

extracts the dislocation graph G0 in the ExaDisNet format (which is the native format for the ExaDiS library).

The following line in the python code

G1 = net.get_disnet(DisNet)

converts the dislocation graph to G1 in the DisNet format, which may be more convenient to interact with for some purposes. For example, the DisNet has a function for sanity checks (e.g. the Burgers vectors for all dislocation lines going out of any node sum up to zero), which can be called as follows.

G1_sanity_check = G1.is_sane()

If G1_sanity_check is True, then the first test passes.

Conversion using export_data() and import_data() Functions

Both DisNet and ExaDisNet implement the export_data() and import_data() functions that can be used to convert data to different formats. As a self-consistency check, we first create an empty DisNet object G2, and then import the data exported by G1 as follows.

G2 = DisNet()
G2.import_data(G1.export_data())

The result G2 should be a copy that is identical to the original graph G1. We can verify this by calling the is_equivalent() function implemented in DisNet.

G1_eq_G2 = G1.is_equivalent(G2) and G2.is_equivalent(G1)

If G1_eq_G2 is True, then the second test passes.

Conversion to and from Networkx

Networkx is a widely used network library. DisNet has utility functions to export its graph to networkx format (DiGraph) and to import from networkx. As a self-consistency check, we first export the dislocation graph to networkx object G3, and then import data from G3 into a new DisNet object G4 as follows.

G3 = G1.to_networkx()
G4 = DisNet()
G4.from_networkx(G3)

The result G4 should again be a copy that is identical to the original graph G1. We can verify this by calling the is_equivalent() function implemented in DisNet.

G1_eq_G4 = G1.is_equivalent(G4) and G4.is_equivalent(G1)

If G1_eq_G4 is True, then the third test passes.