Bond orders and optimizer issues

[benedictsaunders]

I'm building COFs which have double bonds between the building blocks, but due to the nature of the repeat units, I have had to define by own functional groups. However, using GenericReactionFactory to define the bond orders, it defaults to 1 such that I have trivalent carbons.

Additionally, both MCHammer and Collapser result in slight distortions to the hexagonal geometry; the bonds remain as long as ever!

Any help or suggestions would be greatly appreciated!

    newNHFG = stk.SmartsFunctionalGroupFactory(
        smarts="[C,c]=[N]([H])",
        bonders=(0,),
        deleters=(1, 2),
    )

    bbAldehyde = stk.BuildingBlock(
        smiles=bb1,
        functional_groups=[
            stk.AldehydeFactory(),
        ],
    )

    bbNH = stk.BuildingBlock(
        smiles=bb2,
        functional_groups=[
            newNHFG,
        ],
    )

    cof_stk = stk.ConstructedMolecule(
        topology_graph=stk.cof.Honeycomb(
            building_blocks=(bbNH, bbAldehyde),
            lattice_size=(3, 3, 1),
            reaction_factory=stk.GenericReactionFactory(
                bond_orders={
                    frozenset({newNHFG, stk.Aldehyde}): 2,
                },
            ),
            optimizer=stk.Collapser(
                distance_threshold=1.35,
            ),
        )
    )
    cof_stk.write(f"{str(ID)}_stk.mol")

[lukasturcani]

Hi @benedictsaunders, welcome to stk!

bonder orders

I think the issue here is that in your mapping

    reaction_factory=stk.GenericReactionFactory(
        bond_orders={
            frozenset({newNHFG, stk.Aldehyde}): 2,
        },
    ),

Your set consists of newNHFG and stk.Aldehyde. The issue is that newNHFG is not a FunctionalGroup class, it is a FunctionalGroupFactory instance, but bond_oders expects the set to hold two FunctionalGroup classes. The SmartsFunctionalGroupFactory creates instances of GenericFunctionalGroup so if you do

    reaction_factory=stk.GenericReactionFactory(
        bond_orders={
            frozenset({stk.GenericFunctionalGroup, stk.Aldehyde}): 2,
        },
    ),

it should create double bonds bonds between GenericFunctionalGroup instances (ie those created by the SmartsFunctionalGroupFactory) and the Aldehyde instances.

optimizers

For your second question, I think thats becuse I haven't merged this PR by @andrewtarzia (#282) so I will merge it today and you should be able to use features from there from tomorrow.

[lukasturcani]

Hey, sorry for the delay. Yeah I've managed to get some time to work on the PR yesterday, and I will also have some time today. So hopefully today, but to give myself a more relaxed window let's say next Monday for sure. I'll keep you updated.

[lukasturcani]

Hey hey, I think I'm almost done, so expect a new version of stk tomorrow.

[lukasturcani]

New release is out, if you're using pip you can run pip install stk --upgrade -> the version should be 2021.04.26.0.

You can make an optimized COF with something like

bb1 = stk.BuildingBlock('BrCCBr', [stk.BromoFactory()])
bb2 = stk.BuildingBlock('BrCC(CBr)CBr', [stk.BromoFactory()])
cof = stk.ConstructedMolecule(
    topology_graph=stk.cof.PeriodicHoneycomb(
        building_blocks=(bb1, bb2),
        lattice_size=(3, 3, 1),
        optimizer=stk.PeriodicCollapser(),
    ),
)
stk.MolWriter().write(
    molecule=cof,
    path='example.mol',
)

[lukasturcani]

I just noticed that your example does not use a periodic cof. In this case, I believe the only thing you need to do is
set the scale_steps parameter to False.

I.e. taking your example

    cof_stk = stk.ConstructedMolecule(
        topology_graph=stk.cof.Honeycomb(
            building_blocks=(bbNH, bbAldehyde),
            lattice_size=(3, 3, 1),
            reaction_factory=stk.GenericReactionFactory(
                bond_orders={
                   frozenset({stk.GenericFunctionalGroup, stk.Aldehyde}): 2,
                },
            ),
            optimizer=stk.Collapser(
                scale_steps=False,
            ),
        )
    )

See the Example section of https://stk.readthedocs.io/en/latest/stk.molecular.topology_graphs.cof.cof.html#stk.molecular.topology_graphs.cof.cof.Cof

[benedictsaunders]

Thanks for the update!