Algorithm 1: MTLLSO

Input: the number of particles of two-task NP, random mating probability rmp, number of levels L, level size LS, control parameter ϕ, the maximum number of fitness evaluation FES.   /* Initialization */ fes = 0 /* This variable counts the used fitness evaluations. */ Randomly initialize P1 and P2 for two tasks respectively and evaluate the fitness values for P1 and P2; fes + = 2*NP; While fes < FES do P1 and P2 are divided into L group; For POP = {P1, P2} do /* This variable the population evolving in the current iteration. */ Generate offspring particles OP through Algorithm 2; Evaluate the generated offspring particles OP; fes + = (NP − LS); POP = POP ∪ OP; Sort P1 and P2 in ascending order of fitness; Select the fittest NP particles from P1 and P2 to form the next P1 and P2; End while

Output: The best solutions of P1 and P2