在tensorflow程序运行过程中跟踪变量的方法，通过tf.Print实现

# tf.Print(
#     input_,         # input_: A tensor passed through this op.  输入单元
#     data,           # data: A list of tensors to print out when op is evaluated  输出单元
#     message=None,   # message: A string, prefix of the error message. 在输出data之前输出的前缀
#     first_n=None,
#     summarize=None, #Only print this many entries of each tensor. If None, then a maximum of 3 elements are printed per input tensor.如果需要打印矩阵，需要修改这里
#     name=None
#  )
# Prints a list of tensors.
#return A Tensor. Has the same type and contents as input_
#适用于在tensorflow的流数据中追踪变量，但是该函数当前在jupyter不能使用，即无法打印出来
#在pycharm下可用

# tf.while_loop(
#     cond,          # cond is a callable returning a boolean scalar tensor.
#     body,          # body is a callable returning a (possibly nested) tuple, namedtuple or list of 
                     # tensors of the same arity (length and structure) and types as loop_vars.
#     loop_vars,     # loop_vars is a (possibly nested) tuple, namedtuple or list of tensors that is 
                     # passed to both cond and body.
                     # cond and body both take as many arguments as there are loop_vars.
#     shape_invariants=None,
#     parallel_iterations=10,
#     back_prop=True,
#     swap_memory=False,
#     name=None,
#     maximum_iterations=None,
#     return_same_structure=False
# )

# while(condition(tensors)) 
# { 
#    tensors = body(tensors); 
# }  
# 参考 https://stackoverflow.com/questions/37441140/how-to-use-tf-while-loop-in-tensorflow/37444810
# 可以理解为tf.while_loop 将 loop_vars声明为传递变量，在整个循环过程中，loop_vars一直随着数据流图移动，
# 每一步的返回值因此也必须是loop_vars

import tensorflow as tf

i =0
n = 10
with tf.Session() as sess:
    def cond(i, n):
        #print("in cond: ", (i, n))
        i = tf.Print(i, [i], message="in cond") #注意这里需要将返回值赋值给输入单元
        return i < n

    def body(i, n):
        i = i+1
        #print("in body: ", (i, n))
        print(2)
        return (i, n)
    a, k = tf.while_loop(cond, body, [i, n])  #不能直接使用，需要使用session运行
    sess.run(tf.global_variables_initializer())
    print("after while: ", sess.run([a, k]))
    print("after while: ", sess.run([a, k]))
# 因为在tensorflow中采取流数据的方式，因此在运行时，无法通过print直接捕捉到数据，因此需要
# 调用额外的函数实现。

output：
2
after while:  [10, 10]
in cond: [0]
in body: [0]
in cond: [1]
in body: [1]
in cond: [2]
in body: [2]
in cond: [3]
in body: [3]
in cond: [4]
in body: [4]
in cond: [5]
in body: [5]
in cond: [6]
in body: [6]
in cond: [7]
in body: [7]
in cond: [8]
in body: [8]
in cond: [9]
in body: [9]
in cond: [10]
in cond: [0]
in body: [0]
in cond: [1]
in body: [1]
in cond: [2]
in body: [2]
in cond: [3]
in body: [3]
in cond: [4]
in body: [4]
in cond: [5]
in body: [5]
in cond: [6]
in body: [6]
in cond: [7]
in body: [7]
in cond: [8]
in body: [8]
in cond: [9]
in body: [9]
in cond: [10]
after while:  [10, 10]

上述代码主要为了实现跟踪变量i。可以看到如果直接使用print，则只会运行一次。

但是使用tf.Print函数可以实时的进行跟踪i的变化。。这里我主要想实现对程序进行debug，然后采用该种方式实现。