files / problems / c7-shannon / compute / shift_search.txt
best 367 shift (40, 123, 40, 123, 40) note stopped after recovering the published translation; other sampled translations at den=109 scored 357-360 (M around 325-329, residual I 29-35). structured (a,b,a,b,a) grid and 80 random translations were started (343 planned) but exact MIS on published-scale leftovers was too slow to finish. No 368-set appeared in the completed prefix.