Computing the numerical error as a mesh function

Task: compute the numerical error \( e^n = \uex(t_n) - u^n \)

Exact solution: \( \uex(t)=Ie^{-at} \), implemented as

def exact_solution(t, I, a):
    return I*exp(-a*t)

Compute \( e^n \) by

u, t = solver(I, a, T, dt, theta)  # Numerical solution
u_e = exact_solution(t, I, a)
e = u_e - u