an.tsouchlos/ba-thesis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

Reimplemented gradient computation with more numpy builtin functions

Browse Source
This commit is contained in:
Andreas Tsouchlos 2022-11-07 19:37:46 +01:00
parent b7c4b4e359
commit 97acaaafd3
1 changed files with 8 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
sw/decoders/proximal.py
View File

@ -7,8 +7,6 @@ class ProximalDecoder:
""" """
# TODO: Is 'R' actually called 'decoding matrix'? # TODO: Is 'R' actually called 'decoding matrix'?
# TODO: How large should eta be?
# TODO: How large should step_size be?
def __init__(self, H: np.array, R: np.array, K: int = 100, step_size: float = 0.1, def __init__(self, H: np.array, R: np.array, K: int = 100, step_size: float = 0.1,
gamma: float = 0.05, eta: float = 1.5): gamma: float = 0.05, eta: float = 1.5):
"""Construct a new ProximalDecoder Object. """Construct a new ProximalDecoder Object.
@ -48,24 +46,20 @@ class ProximalDecoder:
# TODO: Is this correct? # TODO: Is this correct?
def _grad_h(self, x: np.array) -> np.array: def _grad_h(self, x: np.array) -> np.array:
"""Gradient of the code-constraint polynomial. See 2.3, p. 2.""" """Gradient of the code-constraint polynomial. See 2.3, p. 2."""
# Calculate first term # Pre-computations
result = 4 * (x**2 - 1) * x k, _ = self._H.shape
# Calculate second term A_prod_matrix = np.tile(x, (k, 1))
A_prods = np.prod(A_prod_matrix, axis=1, where=self._H > 0)
for k, x_k in enumerate(x): # Calculate gradient
sum_result = 0
for i in self._B[k]: sums = np.dot(A_prods**2 - A_prods, self._H)
prod = np.prod(x[self._A[i]])
sum_result += prod**2 - prod
term_2 = 2 / x_k * sum_result result = 4 * (x**2 - 1) * x + (2 / x) * sums
result[k] += term_2 return result
return np.array(result)
# TODO: Is the 'projection onto [-eta, eta]' actually just clipping? # TODO: Is the 'projection onto [-eta, eta]' actually just clipping?
def _projection(self, x): def _projection(self, x):