Simon Often maths gives an unassailably correct answer. I agree, but this isn’t my world.
Then one gets into nonlinear equations, such as for gravitation, and a numerical scheme is in order. But these have error estimates, so what is produced can be accurate but not exact. The topic of error estimates is huge and people even devote their whole careers to it.
Then we have things like weather-forecasting. We are now entering the realm of 3D in space, unsteady, three velocity components, temperature, humidity, air density changes, possible relatively local effects due to clouds and/or cloud cover and possibly precipitation. And is the precipitation water, hail or snow? And the whole system is chaotic, meaning that two very slightly different initial conditions (as interpolated from weather stations, more errors) can yield very different results two days from now. It is a complete nightmare here in the UK. It certainly requires extremely powerful computers. So sometimes the weather is far less chaotic (hence forecasters pronounce confidently what will happen a week from now) and other times much more chaotic (hence forecasters hedge their bets even for tomorrow). Technically, it is down to size of the Lyapunov exponent, and this can be estimated by running the model loads of times with slightly perturbed initial conditions and see how divergent they are. One can tell by the confidence of the forecaster that this has been done. Yes, maths, but chaos makes things nonunique, partly because of the system itself being unstable and partly because we don’t know everything. Maths is only exact in school!