1: \begin{abstract}

2: We study convergence and convergence rates for resampling schemes. Our first main

3: result is a general consistency theorem based on the notion of negative

4: association, which is applied to establish the almost sure weak convergence of

5: measures output from Kitagawa's \citeyearpar{Kitagawa1996} stratified

6: resampling method. Carpenter et al's \citeyearpar{CarClifFearn} systematic

7: resampling method is similar in structure but can fail to converge depending on

8: the order of the input samples. We introduce a new resampling algorithm based

9: on a stochastic rounding technique of \cite{Srinivasan2001}, which shares some

10: attractive properties of systematic resampling, but which exhibits negative

11: association and therefore converges irrespective of the order of the input

12: samples. We confirm a conjecture made by \cite{Kitagawa1996} that ordering

13: input samples by their states in $\mathbb{R}$ yields a faster

14: rate of convergence; we establish that when particles are ordered using the

15: Hilbert curve in $\mathbb{R}^d$, the variance of the resampling error is

16: $\smallo(N^{-(1+1/d)})$ under mild conditions, where $N$ is the number of

17: particles.

18: We use these results to establish asymptotic properties of particle algorithms

19: based on resampling schemes that differ from multinomial resampling.

20: 

21: \textit{Keywords}: Negative association, resampling, particle filtering

22: 

23: 

24: % **are we allowed citations in the abstract?**

25: % YES! (that's fine for most journals, if not they will tell us anyway)

26: 

27: 

28: 

29: 

30: \end{abstract}

31: