The question I've got from the visitor at ECCE-2011 Exposition: "Simulator with ideal components? Why would anybody need components that do not exist?".
The answer is simple and obvious: "To simulate circuits that do not exist".
With standard SPICE-based simulators, a result of the simulation is considered correct if it matches a known real circuit's behavior. What if real circuit does not exist yet? When you design a new principle, a new topology: how can you make sure the idea is feasible? That's what ideal components are used for.
Making everything as simple as possible, you eliminate problems caused by complex models, methods, and algorithms. Starting with ideal components gives you much more confidence that simulation results you see on the screen is true behavior of your circuit - even if the circuit does not exist.