Contrary to the interface of the other acquisitions, the evaluate and evaluate_with_gradients methods of MultipointExpectedImprovement acquisition return the negated acquisition value and gradient.
Example in evaluate:
|
return -self._get_acquisition(mean.flatten(), variance, y_minimum)[0] |
and in evaluate_with_gradients:
Making an issue to summarize this aspect of the discussion in: #303 (comment)
Contrary to the interface of the other acquisitions, the
evaluateandevaluate_with_gradientsmethods ofMultipointExpectedImprovementacquisition return the negated acquisition value and gradient.Example in
evaluate:emukit/emukit/bayesian_optimization/acquisitions/expected_improvement.py
Line 185 in 9f6763f
and in
evaluate_with_gradients:emukit/emukit/bayesian_optimization/acquisitions/expected_improvement.py
Line 201 in 9f6763f
Making an issue to summarize this aspect of the discussion in: #303 (comment)