-
Notifications
You must be signed in to change notification settings - Fork 228
Nodal Distances
Note: This is an explanation for the nodal distances update that has not been pulled yet as of now.
"nodalDistance" is a property of a group object in a kinetics tree that defines the distance between that group and its parent node (see the kinetics estimation page). This affects family averaging and kinetics estimation in general by changing how close different nodes are.
This was developed to solve this type of problem we ran into when simplifying certain trees: (R3_D was eliminated during this process) Ex: Before restructuring: [R3_D;doublebond_intra_pri_2H;radadd_intra_cddouble] finds rate data at [R3_D;doublebond_intra_pri;radadd_intra] After restructuring: [R3;doublebond_intra_pri_2H;radadd_intra_cddouble] finds rate data at [Rn;doublebond_intra_pri_2H;radadd_intra_cddouble]
This results in a rate 8 orders of magnitude higher than the original at 1000 K
The problem is that the movement between Rn and R3 is a much bigger deal than the alternative two movements from doublebond_intra_pri_2H to doublebond_intra_pri and from radadd_intra_cddouble to radadd_intra. So it makes sense to assign a greater nodalDistance value to R3 because it is farther away from Rn.
While we were able to catch this in regression tests, in general, we do not currently have any extensive and robust ways to determine what the nodalDistance attribute should be. So by default nodalDistance is set to one for most nodes. However, tuning nodalDistance attributes