def __init__(self, system) :

self.system = system

description = 'Fit potential energy surfaces using tensorflow neural networks.'

self.parser = arg.ArgumentParser(description=description)

self.parser.add_argument('epochs', help='The number of training epochs',

type=int, default=-1, nargs='?')

self.parser.add_argument('--plot', help='Plot the network potential and the error',

default=False, action='store_true')

self.parser.add_argument('--ploterror', help='Plot the error',

default=False, action='store_true')

self.parser.add_argument('--plotprogress', help='Plot the cost function against epoch number',

default=False, action='store_true')

self.parser.add_argument('--plotall', help='Plot all the things',

default=False, action='store_true')

self.parser.add_argument('--load', help='Load graph from previous training file',

default=False, nargs='?')

self.parser.add_argument('--save', help='Save the graph and training data to file',

default=False, action='store_true')

self.parser.add_argument('--size', help='[# of hidden layers, # of neurons]',

type=int, nargs=' ')

self.parser.add_argument('--saveeach', help='How many epochs between each graph save',

type=int, default=10)

self.parser.add_argument('--type', help='Which activation functions to use',

choices=['sigmoid','relu','relu-sigmoid'],

default=False, nargs='?')

self.parser.add_argument('--file', help='Data file for use in training',

default=None, nargs='?')

self.args = self.parser.parse_args()

def __call__(self) :

return self.args

def nLayers(self) :

if self().size != None and self().size != False :

return self().size[0]

else :

return 2

def nNodes(self) :

if self().size != None and self().size != False :

return self().size[1]

else :

return 5

def type(self) :

if self().type != None and self().type != False :

return self().type

else :