Multivariate Optimization of ILC Parameters

We present results of multiobjective optimization of the International Linear Collider (ILC) which seeks to maximize luminosity at each given total cost of the linac (capital and operating costs of cryomodules, refrigeration and RF). Evolutionary algorithms allow quick exploration of optimal sets of parameters in a complicated system such as ILC in the presence of realistic constraints as well as investigation of various what-if scenarios in potential performance. Among the parameters we varied there were accelerating gradient and Q of the cavities (in a coupled manner following a realistic Q vs. E curve), the number of particles per bunch, the bunch length, number of bunches in the train, etc. We find an optimum which decreases (relative to TESLA TDR baseline) the total linac cost by 22%, capital cost by 25% at the same luminosity of 3 × 10³⁸m^–2s^–1. For this optimum the gradient is 35 MV/m, the final spot size is 3.6 nm, and the beam power is 15.9MW. Changing the luminosity by 10³⁸m^–2s^–1results in 10% change in the total linac cost and 4% in the capital cost. We have also explored the optimal fronts of luminosity vs. cost for several other scenarios using the same approach.