The Parameter tree and OPTIM keywords

This page describes how to translate an OPTIM odata file into the pele Parameter tree. If a keyword is not mentioned, then it is either not relevant, or I just haven’t gotten to it yet.

The documentation for OPTIM keywords can be found here, http://www-wales.ch.cam.ac.uk/OPTIM.doc/node4.html

Generally shortened names are used for the Parameter tree, e.g. lowestEigenvectorQuenchParams instead of params.double_ended_connect.local_connect_params.tsSearchParams.lowestEigenvectorQuenchParams. See here for a full description of the Parameter tree.

NEWCONNECT NConMax NTriesMax ImageDensity IterDensity ImageMax ImageIncr RMStol:

double_ended_connect.niter = NConnMax
(local_connect.NEBattempts = NTriesMax not passable)
NEBparams.image_density =  ImageDensity
NEBparams.iter_density =  IterDensity
NEBparams.max_images =  ImageMax
(ImageIncr not implemented)
NEBparams.NEBquenchParams.tol = RMStol

NEBK nebk:

NEBparams.k = nebk

MAXERISE x1 x2:

(x1 is for normal quenches)
pushoff_params.quenchParams.maxErise = x1 (maybe other places also?)
lowestEigenvectorQuenchParams.maxErise = x2

BFGSTS nevs ncgmax1 ncgmax2 ceig nevl:

lowestEigenvectorQuenchParams.nsteps = nevs
tsSearchParams.nsteps_tangent1 = ncgmax1
tsSearchParams.nsteps_tangent2 = ncgmax2
lowestEigenvectorQuenchParams.tol = ceig
(nevl not used)

BFGSMIN gmax:

(tolerance for standard quenches)
pushoff_params.quenchParams.tol = gmax (maybe other places also?)

MAXSTEP x1 x2:

(x1 is for eigenvector-following, but is it used for other things?)
(x2 ? not documented)

MAXMAX x:

tsSearchParams.max_uphill_step = x

BFGSCONV gmax:

tsSearchParams.tangentSpaceQuenchParams.tol = gmax (I think)

PUSHOFF x:

local_connect_params.pushoff_params.stepmin = x

STEPS n:

?

BFGSSTEPS n:

(number of steps for standard quenches)
local_connect_params.pushoff_params.nsteps = n

MAXBFGS x1 x2 x3 x4:

(max step length)
(x1 for normal quenches)
local_connect_params.pushoff_params.maxstep = x1 (maybe other places also?)
lowestEigenvectorQuenchParams.maxstep = x2
(x3 for putting structures in closest coincidence with ministd)
NEBparams.NEBquenchParams.maxstep = x4

UPDATES mupdate1 mupdate2 mupdate3 mupdate4:

(M for lbfgs)
(mupdate1 is for standard quenches)
pushoff_params.quenchParams.M = mupdate1 (maybe other places also?)
(??? mupdate2 is currently ignored by mind)
lowestEigenvectorQuenchParams.M = mupdate3
NEBparams.NEBquenchParams.M = mupdate4

CHECKCHIRALITY:

(not implemented, but should be)

RANROT n:

(niter in the stochastic minpermdist routine.  not in parameter tree yet)