coupled_chains: Coupled MCMC chains
In pierrejacob/unbiasedpathsampling: Unbiased Path Sampling with couplings

Sample two MCMC chains, each following single_kernel marginally, and coupled_kernel jointly, until min(max(tau, m), max_iterations), where tau is the first time at which the two chains meet (i.e. take the same value exactly). Or more precisely, they meet with a delay of one, i.e. X_t = Y_t-1. The chains are initialized from the distribution provided in rinit.

See get_hmc_kernel for an example of function returning the appropriate kernels.

1 2	coupled_chains(single_kernel, coupled_kernel, rinit, m = 1, max_iterations = Inf, preallocate = 10)

`single_kernel`	function taking a state (in a vector), its log density and an iteration, and returning a list with a key named `chain_state` containing the next state and its log density `current_pdf`.
`coupled_kernel`	function taking two states (in two vectors), their log densities and an iteration, and returning a list with keys `chain_state1`, `chain_state2`, `current_pdf1` and `current_pdf2`.
`rinit`	function taking no arguments are returning an initial state for a Markov chain.
`m`	number of iterations desired (will be proportional to the computing cost if meeting occurs before `m`, default to 1).
`max_iterations`	number of iterations at which the function stops if it is still running (default to Inf).
`preallocate`	expected number of iterations, used to pre-allocate memory (default to 10).
`logtarget`	function evaluating the log target density