Non-cat changes to master from cat branch, 1 #1538
Commits on Jan 31, 2019
-
Add a (failing) unit test for subgraph isomorphism checks of disjoint…
… graphs This is what's currently causing us a problem in RMG-Cat for van-der-Waals bonds, but this is a minimal test case to reproduce the problem in the vf2 algorithm. Unsure at this stage if it's reasonable to fix the vf2 algorithm or if we'll have to work around it at a higher level.
-
Fix vf2 subgraph isomorphism for disconnected graphs.
If you have two graphs [1-2-3 4] and [1-2-3 4] then you can show they're isomorphic OK, but if you search for all the subgraph isomorphisms, it crashed. Now if there are no terminal atoms to choose from in, it chooses pairs only from non-matched vertices, instead of all vertices. I also added some commentary from the original paper.
-
-
-
Improved error logging in Kinetics database when loading reaction fam…
…ilies. Replace print() statements with logging.error() when loading bad reaction families.
-
-
-
Move where a comment line is, I think to clarify.
Unless I misunderstood, I think this is clearer. (And if I did misunderstand, then it needs to be clearer!)
-
-
Create a getConversionFactorFromSItoCM method on Units.
The idea is that if the object has units of 'm^3/mol/s' then but you want to get it in chemkin format you need the value in 'cm^3/mol/s'. This gives the factor required. Add a warning in docstring about potential bug.
-
Chemkin writing uses Units.getConversionFactorFromSItoCM()
This will allow us to do surface reactions with correct units. (Which will be implemented in a later commit) For gas-phase reactions, I put in a bunch of assert statements to check that the new implementation has the same effect as the old. If this is speed-critical code, they could be deprecated/removed.
-
-
Fix rate constant unit conversion to cm/mol/s
Replaced the method getConversionFactorFromSItoCM which used to leave things like Molecules in their original form, with a new method getConversionFactorFromSItoCmMolS which converts everything into S (apart from length becomes cm). This means that all chemkin files are now written in CM/MOL/S combinations even if the original rate was in m3/molecules/minute. A few reaction libraries and training rules are stored in molecules instead of moles, but the chemkin file must be consistent.
-
Fix molecule drawing rotation bug. (master)
When rotating by matrix algebra, we need to keep self.coondinates and coordinates pointing to the same numpy array, otherwise other things that reference or update one will not see or change the other. This commit just does the non-cat instance, so can be applied to master branch.
-
-
Add cdef type declarations for base solver class
numSurfaceSpecies and numSurfaceReactions
-
These don't seem to be catalyst-specific. (Though will be helpful for ones that are)
-
Force simpleReactor initialization to use keyword arguments.
Reduces the chances of passing the wrong thing by using positional arguments.
-
Test the A-factor units on reaction library kinetics.
So far this does: - Arrhenius - Third Body - Falloff (Lindemann, Troe) - PDepArrhenius (PLOG) - MultiArrhenius - MultiPDepArrhenius in reaction libraries. This could also be done for other instances of kinetics (training depositories, etc.)
-
Add units check for kinetics depositories to database test
Make checkLibraryRateUnits more generic Add test definition for kinetics depositories
