Fixes bugs in optimizer docstrings #1053
Conversation
|
Hello. You may have forgotten to update the changelog!
|
| tuple[list [array], float]: the new variable values :math:`x^{(t+1)}` and the objective | ||
| function output prior to the step. | ||
| If single arg is provided, list [array] is replaced by array. |
There was a problem hiding this comment.
Continuation lines in Returns: should never be indented; if they are, sphinx will treat them as a glossary entry and bold the first line.
| @@ -106,8 +106,8 @@ def step(self, objective_fn, *args, grad_fn=None, **kwargs): | |||
|
|
|||
| @staticmethod | |||
| def compute_grad(objective_fn, args, kwargs, grad_fn=None): | |||
| r"""Compute gradient of the objective_fn at the point x and return it along with the | |||
There was a problem hiding this comment.
The docstring conventions discourage the use of variable names in the first line/summary line of the docstring. Hence replacing objective_fn with 'objective function'.
Codecov Report
@@ Coverage Diff @@
## master #1053 +/- ##
=======================================
Coverage 97.75% 97.75%
=======================================
Files 153 153
Lines 11514 11514
=======================================
Hits 11255 11255
Misses 259 259
Continue to review full report at Codecov.
|
Context:
Description of the Change:
Removes rendering bugs in the
Returns:statement by removing indentation in continuation lines. While theArgs:section requires continuation lines be indented since there might be multiple arguments, there is only ever a single return, and so the return is never indented.In Allowing more flexible cost functions for optimizers #959, most of the optimizers were updated to remove restrictions on cost function signatures. As a result, the
xargument was changed to*argsto be more general. Some strayxmentions in the docstrings were removed.Benefits: Docstrings render correctly.
Possible Drawbacks:
I noticed while working on this that the
QNGOptimizeris the one optimizer that continues to only accept cost functions with a single argument. This was intentional in #959, as at the time themetric_tensormethod did not allow multiple QNode arguments.However, the new
qml.metric_tensorfunction that was added in #1014 removes this restriction, so we are now unblocked from also updating theQNGOptimizer.Related GitHub Issues: n/a