Move-file format

Movefile example

The move-file is used in GDS to define which graph-moves can be used to generate new reaction-path end-points during each GDS graph-change step.

A typical move-file (usually called moves.in, although can be called anything) looks like this:

move
natom 2
-
0 1
1 0
-
0 0
0 0
-
labels * *
prob 0.4

move
natom 2
-
0 0
0 0
-
0 1
1 0
-
labels * *
prob 0.4

move
natom 3
-
0 1 0
1 0 1
0 1 0
-
0 0 1
0 0 1
1 1 0
-
labels * * *
prob 0.1

move
natom 3
-
0 1 0
1 0 1
0 1 0
-
0 0 1
0 0 0
1 0 0
-
labels * * *
prob 0.1

Each move is defined in a block, which looks like this.

move
natom 2
-
0 1
1 0
-
0 0
0 0
-
labels * *
prob 0.4

In the example above, the proposed graphmove involves two atoms; these are randomly selected in the code. The first 2x2 matrix is the required bonding pattern of the two atoms; in this case, the moves requires that the two atoms are bonded (i.e. there is a 1, indicating bonding, on the off-diagonal element). The second 2x2 matrix indicates the target final graph; in thise case, there is a 0 on the off-diagonal, indicating that there is no bond between the two atoms in the final structure. Overall, therefore, this is a simple bond-breaking reaction.

The labels line indicates the required atom labels for this move; in this case, we have indicated " * ", which means that any atom can participate. An alternative would be

labels C O

which would imply that only C and O can be involved for this graph move.

The prob value gives the relative probability of this move taking place during a given GDS run; not that the probabilities of all moves are all normalised in the code (i.e. they are re-scaled so that they sum to 1), so they do not have to add to 1 in the movefile.

A second example, this time involving 3 atoms, is as follows:

move
natom 3
-
0 1 0
1 0 1
0 1 0
-
0 0 1
0 0 0
1 0 0
-
labels * * *
prob 0.1

The above reaction is A-B-C --> A-C + B.