Test Cases
A two-dimensional Branin function is tested for the one-shot method, and a three-dimensional Hartmann function is tested for the sequential sampling method. In all cases, the Halton sequence is applied to ensure consistent starting points for all methods.
One-shot method - Branin function
The minimization of the two-dimensional Branin function is formulated as:
with constraints \(x_{d1} \in [0, 8]\) and \(x_{d2} \in [0, 8]\). The goal is to find a global minimum of 0.39 at \((\pi, 2.275)\).
Sequential sampling method - Hartmann function
The minimization of the three-dimensional Hartmann function is expressed as:
where \(\alpha = [1.0, 1.2, 3.0, 3.2]^T\), and matrices \(\mathbf{A}\) and \(\mathbf{P}\) are defined as:
The function is constrained by \(\mathbf{x}_{d_L} = (0,0,0)\) and \(\mathbf{x}_{d_U} = (1,1,1)\), with one global minimum of -3.86 at \((0.11, 0.56, 0.85)\) and four local minima.