Skip to content

P
pytensor
提交 da0fd398 authored 作者: David Warde-Farley's avatar David Warde-Farley
浏览文件
操作

Assert CholeskyGrad has same layout as Cholesky.

Since one of the inputs here is the Cholesky factorization itself, it's important they both be computing the lower triangle or the upper triangle.
上级 05007f20
隐藏空白字符变更
内嵌 并排
正在显示 包含 3 行增加 和 0 行删除
theano/sandbox/linalg/ops.py
浏览文件 @ da0fd398
...@@ -387,6 +387,9 @@ class CholeskyGrad(Op): ...@@ -387,6 +387,9 @@ class CholeskyGrad(Op):
x = as_tensor_variable(x) x = as_tensor_variable(x)
l = as_tensor_variable(l) l = as_tensor_variable(l)
dz = as_tensor_variable(dz) dz = as_tensor_variable(dz)
assert l.owner.op.lower == self.lower, (
"lower/upper mismatch between Cholesky op and CholeskyGrad op"
)
return Apply(self, [x, l, dz], [x.type()]) return Apply(self, [x, l, dz], [x.type()])
def perform(self, node, inputs, outputs): def perform(self, node, inputs, outputs):
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论