utils.py

'''
utility file provide helper function
'''
import numpy as np
import tensorflow as tf


def discounted_rewards(r, gamma):
    """ take 1D float array of rewards and compute discounted reward """
    discounted_r = np.zeros_like(r)
    running_sum = 0
    for i in reversed(range(0,len(r))):
        discounted_r[i] = running_sum * gamma + r[i]
        running_sum = discounted_r[i]
    return list(discounted_r)


def add_FC_layer(inputs,in_size,out_size, activation_function = None):
    Weights = tf.Variable(tf.random.normal([in_size,out_size]),tf.float32)
    Bias = tf.Variable(tf.zeros([1,out_size]),tf.float32)
    Wx_plus_b = tf.matmul(inputs,Weights) + Bias
    if activation_function == None:
        output = Wx_plus_b
    else:
        output = activation_function(Wx_plus_b)

    return output

def get_normal(mean,variance):
    tfd = tfp.distributions
    dist = tfd.Normal(loc=mean, scale=variance)
    
    return dist


def add_arguments(parser):
    parser.add_argument("--num_asset", type=int, default=4, help="asset number.")
    parser.add_argument("--horizon", type=int, default=50, help="time step.")
    parser.add_argument("--num_signal_feature",type=int, default = 4, help = "number of signal axiuliary feature")
    parser.add_argument("--learning_rate_policy_net", type=float, default=1e-3, help="Learning rate.")
    parser.add_argument("--learning_rate_value_net", type=float, default=1e-3, help="Learning rate.")
    parser.add_argument("--sigma",type = float, default = 0.05,help = "sigma")
    parser.add_argument("--variance_decay",type = float,default = 5e-2,help = "decay rate")
    parser.add_argument("--transcation_cost", type = float, default =25e-3)
    parser.add_argument("--discount_rate", type = float, default = 1, help = 'dicounted rate for reward')
    parser.add_argument("--num_traj",type = int,default = 30, help = "number of trajectory")
    parser.add_argument("--iteration",type = int,default = 150, help = "number of iteration")
    parser.add_argument("--max_train",type = int, default = 100,help = "maximum training per trajectory")
    parser.add_argument("--with_model", action="store_true", help="Continue from previously saved model")

    parser.add_argument("--cs",type = float, default = 0, help="cost of sell")
    parser.add_argument("--cp",type = float, default = 0, help="cost of purchase")
    parser.add_argument("--step_size",type = int, default = 25, help="env rollling step size")