For a big-theta proof, you don't need the best possible constants. You could make "ridiculous" choices like c1 = 0.001 and c2 = 1000 and n0 = 1000000 and everything would be fine as long as you can finish the proof. (One of my professors liked to do this while lecturing.)
If for whatever reason you want tight constants, then you need to learn how to minimize/maximize a function over an interval. Calculus is useful here. Note that there can be a trade-off between, for example, how tight c1 is and how tight n0 is. In your second example, since 1/n^2 is decreasing, we need c1 <= 10 - 1/n0^2, so the larger n0 is, the closer c1 can be to 10.