Machine learning force field generator implemented according to Botu et. al.'s pioneering works (Ref. 1, 2).
Keyword |
Type |
Meaning |
Neta |
int |
Dimension of eta grid |
eta |
float;float |
Uniformly generating a logarithmic-law grid. |
lat_len |
float;float;float |
lattice const: a, b, c |
lat_ang |
float;float;float |
lattice const: alpha, beta, gamma |
Rc |
float |
cut off radius |
shuffle |
bool |
shuffle the fingerprint or not |
inp_path |
string;string;string |
path;start_num;end_num |
write |
bool |
Writing the fingerprint files? |
out_path |
string |
path to write fingerprint files |
Keyword |
Type |
Meaning |
Ntrain |
int |
training samples |
k |
int |
k-fold cross validation |
Nlbd |
int |
number of lambda's |
lbd |
float;float |
lambda grid is generated in the same way as eta grid |
Keyword |
Type |
Meaning |
tau0 |
float |
In formulat (tau0 + t)^(-kappa) |
kappa |
float |
In formulat (tau0 + t)^(-kappa) |
MAXITER |
int |
maximum number of iterations |
Nbatch |
int |
size of batch for batch SGD |
Format: Vx(eta_1);Vx(eta_2);...;V(eta_Neta);Fx