log10_op

→TensorImplPtr

int log10_op(const int & a)

source edit

Python wrappers

pylucid.ops.log10function

Base-10 logarithm: $y = \log_{10}(x) = \ln x / \ln 10$ .

Composite over log_op + div_op. Materialises $\ln 10$ as a same-shape constant so the divide does not need scalar broadcasting. Gradient flows through LogBackward and DivBackward.

Math

y = \frac{\ln x}{\ln 10}, \qquad \frac{\partial y}{\partial x} = \frac{1}{x \ln 10}

See Also

log_op, log1p_op.

Parameters

aTensorImplPtr

Input tensor. Domain:

x > 0

Returns

TensorImplPtr

Tensor of the same shape and dtype as a.

Notes

Inherits the precision of the underlying log_op; no specialised base-10 kernel is used.