Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

I am looking for a reference for a result of the following form:

I have a sequence of discrete probability distributions, $p_N$, where the $N$th distribution has associated state space {$k/N, 1 \leq k \leq N$}. The value $Np_N(k)$ (probability of the state $k/N$, divided by the "size" of the state) is defined by a difference equation. I want to conclude that $N p_N(Nx) \to \pi(x)$ for some density $\pi$ on $[0,1]$, described by the "limiting" ODE. (Strictly speaking, I just need to show that the set of "discrete densities" has compact closure, i.e. that all subsequences have a convergent sub-subsequence; showing that the limit of a convergent subsequence is described by the ODE is easy.)

Equivalently, I have a sequence of solutions $\tilde p_N$ (where $\tilde p_N$ corresponds to $Np_N$ in the previous paragraph) to (a sequence of) difference equations. I interpret the solutions as the values of the function $\tilde p_N$ at points $k/N, 1 \leq k \leq N$. The difference equations (and the initial conditions) "converge" to an ODE (with initial conditions), which has a unique solution. I would like to claim that the functions $\tilde p_N$ converge to this solution of the ODE. (Again, all I really need is that the functions $\tilde p_N$ converge somewhere, ideally somewhere absolutely continuous.)

Just to clarify: I'm convinced the result is true, and has been proven a dozen times elsewhere; I'm hoping someone will tell me where.

share|improve this question

1 Answer 1

up vote 1 down vote accepted

A friend pointed out that (in the opposite direction, i.e. going from an ODE to a discrete approximation) this is essentially Euler's method, and convergence theorems for it can be found in any reasonable numerical analysis textbook (e.g., Chapter 7 of Bradie, A Friendly Introduction to Numerical Analysis).

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.