Metaheuristics Explorer

Problem & Controls

Bit Length

Search space: x ∈ [0, 2^bits − 1]

Objective: f(x) = a x³ + b x² + c x + d

a (x³)

b (x²)

c (x)

d (const)

• Apply：更新係數並重畫 landscape
• Analyze：掃描整個空間找 global optimum 與 local maxima（bit-flip 鄰域）

Landscape Analysis
(按下 Analyze Landscape 後顯示)

Algorithm Selection

Algorithm Initial x

SA Parameters

T0 alpha (cooling)

TS Parameters

Tabu size

建議教學：先用 II 看卡住，再切 SA/TS 看 escape。

Batch Test

Max iterations (per run) Number of runs Random seed

手動

未勾選「手動」：每次 Batch 自動產生隨機 seed。

• 每次 run 會從隨機初始解開始（可重現：由 seed 決定）
• SA/TS 會回報 best-so-far（不一定是最後一代）

Log

Batch Results

Symbol Explanation
x: current solution (decoded integer)
bits: binary encoding of x (length = bitLength)
f: objective value f(x)
best: best-so-far objective value
Δ: f(neighbor) − f(current) (SA step shows it)
T: temperature (SA only; cooled by T ← α·T)
p: acceptance probability exp(Δ/T) (Δ<0)
r: random draw in [0,1] (accept if r < p)
tabu: forbidden recent states (TS memory)