Table 1 Numerical complexities of the algorithms compared at each iteration (for a given t)
From: A comparison of mixed-variables Bayesian optimization approaches
Mixed space search | Vanilla LV-EGO | ALV-EGO-g | ALV-EGO-l | |
|---|---|---|---|---|
(Alg. 1) | (Alg. 2) | |||
GP learning | \((n_{c}+\sum _{i=1}^{n_{d}} m_{i})\times t^3\) | \((n_{c}+q\times \sum _{i=1}^{n_{d}} m_{i})\times t^3\) | \((n_{c}+q\times \sum _{i=1}^{n_{d}} m_{i})\times t^3\) | \((n_{c}+q\times \sum _{i=1}^{n_{d}} m_{i})\times t^3\) |
Max acquisition | \( (\prod _{i=1}^{n_{d}} m_{i}) \times n_{c}\times t^2\) | \((n_{c}+ q\times \sum _{i=1}^{n_{d}} m_{i}) \times t^2\) | \((N_{\text {DoE}}' + n_{c}+ q\times \sum _{i=1}^{n_{d}} m_{i}) \times t^2\) | \((n_{c}+ q\times \sum _{i=1}^{n_{d}} m_{i}) \times t^2\) |
Pre-image | 0 | \((\prod _{i=1}^{n_{d}} m_{i}) \times t^2\) | \((\prod _{i=1}^{n_{d}} m_{i}) \times t^2\) | \( (\prod _{i=1}^{n_{d}} m_{i}) \times t^2\) |