Ruzsa’s genus-one problem

Found by a dedicated fixed-cardinality search at modulus 503. Simulated annealing first produced a 15-element near-solution whose only conflict edge was {126, 249, 495} (two ordered nontrivial solutions). Starting there, a tabu search that evaluated every one-residue swap crossed the local barrier and found this set after 11,956 iterations. Direct enumeration gives exactly 15 solutions to a+3b=2c+2d, all diagonal.