gradient_check.py 文件源码

def eval_numerical_gradient_array(f, x, df, h=1e-5):
  """
  Evaluate a numeric gradient for a function that accepts a numpy
  array and returns a numpy array.
  """
  grad = np.zeros_like(x)
  it = np.nditer(x, flags=['multi_index'], op_flags=['readwrite'])
  while not it.finished:
    ix = it.multi_index

    oldval = x[ix]
    x[ix] = oldval + h
    pos = f(x).copy()
    x[ix] = oldval - h
    neg = f(x).copy()
    x[ix] = oldval

    grad[ix] = np.sum((pos - neg) * df) / (2 * h)
    it.iternext()
  return grad