I guess theory is considered as toy models that illustrate features of phenomena. The simulation activity would then be trying to a priori predict the behaviour of an actual system, given that these systems are more intricate,complex the hope of describing its accurate behaviour by on paper methods seems hard but you could try to ‘brute force’ it with computers.

I guess very theoretically oriented people could retorque that the relevant properties you want to describe might be obfuscated by blind intensive calculation when the relevant information you want might be obtained by a reasoning pattern and even proven.