← Back to GSA docs

teeline · algorithms/gsa

Gravitational Search Algorithm

Each candidate TSP tour is an agent with a mass proportional to its fitness — better tours weigh more. Every step, the active agent is pulled toward the k-best (heaviest) neighbours via swap-move velocity. The gravitational constant G decays over epochs, shifting the swarm from broad exploration toward fine-grained exploitation.

01234567891011
Agents
01868
#0 · dist=1868 · mass=0.146
12004
#1 · dist=2004 · mass=0.114
22051
#2 · dist=2051 · mass=0.104
31724
#3 · dist=1724 · mass=0.179
41639
#4 · dist=1639 · mass=0.199
51813
#5 · dist=1813 · mass=0.159
62069
#6 · dist=2069 · mass=0.099
72500
#7 · dist=2500 · mass=0.000
● gbest tour● active agent● swarm● city kbest source
Press Step or Run to begin
G (gravitational constant) = 20.00
High — broad exploration
epoch
0
step
0/8
gbest dist
1638.8
avg dist
1959
Balanced swarm size
cities: 12agents: 8G₀: 20α: 1kbest: ⌈8/2⌉=4