Original publication: "Automatic Topological Generator for Framework Structures".

This version is under active development. Bug hunting is very much going on, and there are still some old funtionalities that are not yet reimplemented.

Medium priority: - offer an isotropic scaling option - the symmetry axes detection has trouble with planar shapes - more robust mmanalysis module

Low priority:

• unit testing coverage is non-existant
• documentation writing
• better handling of the databases:
  • precomputing sbu-topologies correspondance
  • use better format than pickle

$ pip install --user --upgrade AuToGraFS

For a manual install, first install the dependencies,

1. python >=3.4
2. ase, scipy, numpy<1.15.0

then clone this repository and add it to you pythonpath:

$ cd $HOME
$ git clone https://github.com/DCoupry/autografs.git
$ export PYTHONPATH=$HOME/autografs:$PYTHONPATH`

From any python script or command line:

>>> from autografs import Autografs >>> mofgen = Autografs() >>> mof = mofgen.make(topology_name="pcu", >>> sbu_names=["Zn_mof5_octahedral", "Benzene_linear"]) >>> mof.write()

Custom databases can be accessed by passing the path during instanciation

>>> mofgen = Autografs(topology_path="my_topo_path",sbu_path="my_sbu_path") >>> mof = mofgen.set_topology(topology_name=custom_topology_name,sbu_name=custom_sbu_names)

When looping over both SBU and topologies, it is better to set the topology directly (here, my_topologies and my_sbu_names are appropriate dummy colletions)

>>> for topologi_name in my_topology_names: >>> mofgen.set_topology(topology_name=topology_name) >>> for sbu_names in my_sbu_names: >>> mof = mofgen.make(sbu_names=sbu_names)

It is possible to pass more than one SBU of each shape, optionally with an associated probabilistic weight. This weight defaults to 1.0/(number of similar sbu).

>>> mof = mofgen.make(topology_name="pcu", >>> sbu_names=["Zn_mof5_octahedral", ("Benzene_linear",2.0),("Acetylene_linear",0.5)]) >>> mof.write()

This is particularly helpful in combination with an initial supercell for statistically introducing defects.

>>> mof = mofgen.make(topology_name="pcu", >>> sbu_names=[("Zn_mof5_octahedral",2.0),("defect_octahedral",0.5), "Benzene_linear"], >>> supercell=(3,3,3)) >>> mof.write()

Supercell can also be generated post-alignement, carrying everything done before.

>>> supercell_6x6x6 = mof.get_supercell(m=2) >>> supercell_6x6x6.write()

Defects and modifications can be introduced at any time directly:

>>> # get the site directly >>> sbu = mof[7] >>> # change all hydrogens to Fluorine >>> atoms = sbu.atoms.copy() >>> symbols = atoms.get_chemical_symbols() >>> symbols = [s if s!="H" else "F" for s in symbols] >>> atoms.set_chemical_symbols(symbols) >>> # by setting the atoms back, >>> # mmtypes and bonding are updated. >>> sbu.set_atoms(atoms=atoms,analyze=True) >>> # delete another sbu. H will cap the dangling bits. >>> del mof[8] >>> mof.write()

Methods are also available for the rotation, functionalization and flipping.

>>> # rotate the sbu 7 buy 45 degrees >>> mof.rotate(index=7,angle=45.0) >>> # if a C* axis or reflection plane is detected >>> # in the sbu 8 , flip around it >>> mof.flip(index=8) >>> # replace all functionalizable H sites with NH2 >>> nh2 = mofgen.sbu["NH2_point_group"] >>> sites = mof.list_functionalizable_sites(self,symbol="H") >>> for site in sites: >>> mof.functionalize(where=site,fg=nh2) >>> mof.write()

At any moment, we can monitor the bonding matrix and mmtypes, or get a cleaned version without dummies.

>>> from ase.visualize import view >>> # with the dummies included >>> mmtypes = mof.get_mmtypes() >>> bonds = mof.get_bonds() >>> # without the dummies >>> atoms,bonds,mmtypes = mof.get_atoms(dummies=False) >>> view(atoms)

If you know the shape of each slot and its index within the topology, it is possible to directly pass a dictionary mapping the SBU to a particular slot.

>>> # method to investigate the topology shapes and slots >>> topology = mofgen.get_topology(topology_name="pcu") >>> sbu_dict = {} >>> for slot_index,slot_shape in topology.shapes.items(): >>> # do something to choose an sbu >>> ... >>> sbu_dict[slot_index] = "chosen_sbu_name" >>> # now pass it directly >>> mof = mofgen.make(topology_name="pcu", sbu_dict=sbu_dict) >>> mof.write()

You can access the databases as dictionaries using the following:

>>> sbudict = mofgen.sbu >>> topodict = mofgen.topologies

Or using tools to find compatible objects:

>>> sbu_list = mofgen.list_available_sbu(topology_name="pcu") >>> topology_list = mofgen.list_available_topologies(sbu_names=["Zn_mof5_octahedral", "Benzene_linear"])

AAuToGraFS is also aware of topologically equivalent positions, and can generate multi components frameworks with minimal effort.

>>> sbu_dicts = mofgen.list_available_frameworks() >>> for sbu_dict in sbu_dicts: >>> mof = mofgen.make(sbu_dict=sbu_dict) >>> mof.view()

A useful utility is the Atom typer, which assigns bond orders and UFF atom types to a structure:

>>> from autografs.mmanalysis import analyze_mm >>> bonds, types = analyze_mm(sbu=mofgen.sbu["Zn_mof5_octahedral"])