Question

我正在尝试计算期权价格y对输入现货价格的梯度。这是我的代码。现在看来梯度不是没有梯度，但它返回张量误差，我认为我做对了。

def vanilla_price(X):
moneyness_inverse = X.iloc[:, 0]
sig = X.iloc[:, 1]
r = X.iloc[:, 2]
T = X.iloc[:, 3]
d1 = (np.log(moneyness_inverse) + (r + 0.5 * sig ** 2) * T) / (sig* 
np.sqrt(T))
d2 = (np.log(moneyness_inverse) + (r - 0.5 * sig ** 2)* T) / (sig * np.sqrt(T))

    
result = ((moneyness_inverse) * ss.norm.cdf(d1, 0.0, 1.0) -  np.exp(-r * T) * ss.norm.cdf(d2, 0.0, 1.0))

return result

spot = 100.001
spot = 100.001
weight = tf.keras.preprocessing.sequence.pad_sequences(np.array(X), dtype 
='float32')

with tf.GradientTape(persistent=True) as g:
   g.watch(weight)
   y = vanilla_price(weight)
    

jacob_actual = g.gradient(y, weight)[:, 0]
print(jacob_actual)
ValueError                                Traceback (most recent call last)
<ipython-input-231-c20368558cc7> in <module>
  3 
  4 with tf.GradientTape(persistent=True) as g:
  ----> 5     g.watch(weight)
  6     y = vanilla_price(weight)
  7 

  ~\Anaconda3\lib\site-packages\tensorflow_core\python\eager\backprop.py in 
  watch(self, tensor)
  848               pywrap_tensorflow.IsVariable(t)):
  849         raise ValueError("Passed in object of type {}, not 
  tf.Tensor".format(
  --> 850             type(t)))
   851       if not t.dtype.is_floating:
   852         logging.log_first_n(

   ValueError: Passed in object of type <class 'numpy.ndarray'>, not 
   tf.Tensor

谁能给我一个提示或我做错了什么？

Tensorflow梯度不返回

0 个答案: