lbl-camera / gpcam Goto Github PK

In gpcam/surrogate_model.py

lines#150-152 is a call to evaluate_acquisition_function, and lines#170-171 is the definition of evaluate_acquisition_function.

As-is, cost_function in line# 152 is mapped to number_of_maxima_sought in line#170.

To fix this, I suggest specifying all keyword arguments as keyword arguments in the call (lines#150-152), i.e. origin=origin,cost_function=cost_function,cost_function_parameters=cost_function_parameters, instead of relying on their positions in the list of arguments.

Thanks.
Russell Fung (University of Wisconsin Milwaukee)

In gpcam/surrogate_model.py, lines# 303-305, in the devo(partial(…)) call, if vectorized is True & updating is ‘immediate’, scipy resets updating to ‘deferred’ automatically and displays a warning.

Setting updating ourselves will remove this nuisance warning.

I suggest adding

updating = 'deferred' if vectorized else 'immediate'

before the res = devo(partial(…)) call,and adding updating=updating in the list of input arguments.

Thanks.
Russell Fung (University of Wisconsin Milwaukee)

fvgp is currently pinned (scipy == 1.12.0) higher than gpcam (scipy == 1.11.2). Cursory testing showed no issues from upgrading the pin on gpCAM.

The sparse_mode kwarg is no longer present in fvgp/gp.py:GP.__init__. If this is no longer an available option, remove it from gpcam/gp_optimizer.py:GPOptimizer.__init__.

Currently, the following error occurs when instancing GPOptimizer:

...
  File "C:\Users\rp\PycharmProjects\gpcam\gpcam\gp_optimizer.py", line 343, in __init__
    super().__init__(
TypeError: GP.__init__() got an unexpected keyword argument 'sparse_mode'

Edit: The same seems to apply to the normalize_y kwarg.