Golden Blood Optimization (GBO)

The algorithm is inspired by the idea of rare golden donors whose unique genes are selectively transfused into the population

Intuition / metaphor

• Population = donors. Each donor is a candidate solution vector.
• A few individuals become Golden Donors (elite, rare) thanks to exceptional fitness.
• Transfusion operator: parts of golden donors are “transfused” into other donors (crossover/exploitation).
• Plasma diffusion: small random perturbations spread through the population (exploration).
• Antigen suppression: weak/duplicate donors are suppressed/replaced to keep diversity.
• Balances exploitation (use golden donors) and exploration (diffusion + random donors).

2) Main operators (how it works)

1. Initialization: random donor population.
2. Evaluation: fitness for all donors.
3. Select Golden Donors: top g_ratio fraction (1–3 individuals typically).
4. Transfusion: for each donor, replace some genes with genes sampled from a randomly chosen golden donor (probability depends on donor’s relative weakness).
5. Plasma diffusion: add Gaussian or Lévy perturbation to some donors (controls exploration).
6. Antigen suppression (replacement): replace a fraction of worst donors with new random donors or mutated copies of golden donors.
7. Elitism: preserve the best golden donor(s) across iterations.
8. Stop after MaxIt.

3) Pseudocode (short)

initialize population P (nPop x dim)

evaluate fitness

for t = 1:MaxIt

select golden donors G (top g_count)

for each donor i in P

choose a golden donor g_rand

perform transfusion: replace k genes (prob transf_prob) with g_rand genes

perform plasma diffusion: donor = donor + sigma(t)*random_perturb

enforce bounds

evaluate fitness

apply antigen suppression: replace worst r_frac donors

with new samples

update global best

end

return best solution, convergence curve