# _*_ coding: utf-8 _*_

# GPU command:

# THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python script.py

from __future__ import absolute_import

from __future__ import print_function

import numpy as np

import time

import sys

import theano

import cPickle

import os

from keras.models import Sequential

from keras.layers.core import Dense, Dropout

from keras.layers.recurrent import DropoutLSTM

from keras.callbacks import ModelCheckpoint, ModelTest

from keras.regularizers import l2

import matplotlib.pyplot as plt

from reader import read_data_sets, construct_time_steps

from missing_value_processer import missing_check

from feature_processor import data_coordinate_angle

root_path = '/home/clliao/workspace/python/socket/rnn_regression/'

sys.path.insert(0, "/usr/local/cuda-7.5/bin")

sys.path.insert(0, root_path + "keras-BayesianRNN") # point this to your local fork of https://github.com/yaringal/keras

sys.path.insert(0, "../Theano")

# Create ram disk: mount -t tmpfs -o size=512m tmpfs /mnt/ramdisk

# Use flag THEANO_FLAGS='base_compiledir=/mnt/ramdisk' python script.py

print('Theano version: ' + theano.__version__ + ', base compile dir: '

+ theano.config.base_compiledir)

theano.config.mode = 'FAST_RUN'

theano.config.optimizer = 'fast_run'

theano.config.reoptimize_unpickled_function = False

def time_spent_printer(start_time, final_time):

print((int(spent_time) % 3600) % 60, 'seconds')

def target_level(target, kind='PM2.5'):

input('error value: %d' % target)

pollution_site_map = {

}

def rnn(pollution_kind, local, city, target_site, training_year, testing_year, training_duration, testing_duration, interval_hours, data, is_training):

return mean_y_train + std_y_train * prob_mean