Abstract:We investigate the concatenation of Markov processes. Our primary concern is to utilize processes constructed in this manner for Monte Carlo integration. To enable this using conventional methods, it is essential to demonstrate the Markov property and invariance with respect to a given target distribution. We provide mild sufficient conditions for this. Our main result is the identification of the generator of the concatenation of Markov processes. This result provides the theoretical foundation for Monte Carlo methods based on this construction.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is about the splicing of Markov processes and its application in Monte Carlo integration. Specifically, the author focuses on how to construct new Markov processes by splicing multiple Markov processes with finite - time execution, and ensure that these new processes satisfy the Markov property and the invariance with respect to a given target distribution. ### Main problem description 1. **Splicing of Markov processes**: - The author explores how to splice multiple Markov processes together in sequence to form a new Markov process. Each process is replaced by the next process after it terminates at its specified time. - The splicing process needs to ensure that each individual process has a finite lifespan during its execution, so as to ensure that the subsequent processes can be executed smoothly. 2. **Markov property and invariance**: - In order for the spliced process to be used in Monte Carlo integration, it must be proved that the spliced process still maintains the Markov property and is invariant with respect to a given target distribution. - The author provides mild sufficient conditions to ensure the establishment of these properties. 3. **Identification of the generator**: - The main result of the author is to identify the generator of the spliced Markov process. This result provides a theoretical basis for Monte Carlo methods based on such constructions. ### Formula summary - **Definition of Markov process**: \[ (X_t)_{t \geq 0} \text{ is a } (E, \mathcal{E})\text{-value process} \Leftrightarrow \begin{cases} (X_t)_{t \geq 0} \text{ is a process adapted to } (\mathcal{F}_t)_{t \geq 0} \\ X_{s + t}=X_t\circ\theta_s \quad P\text{-almost everywhere} \end{cases} \] - **Generator of the spliced process**: \[ \frac{Q_t f(x)-f(x)}{t}\xrightarrow{t \to 0^+}(Lf)(x)-c(x)f(x) \] for all \( (x, f)\in E_M\times D(L) \), where \( L \) is the point - wise generator of the original Markov semigroup, and \( c(x) \) is the rate function determined by the splicing rules. ### Conclusion Through the above research, the author provides a strict theoretical framework for the splicing of Markov processes and shows its potential application in Monte Carlo integration. This not only extends the relevant work in the existing literature, but also provides a solid foundation for future research.

Concatenation of Markov processes

Generator Matching: Generative modeling with arbitrary Markov processes

Parallelization of Markov Chain Generation and its Application to the Multicanonical Method

The Concatenator: A Bayesian Approach To Real Time Concatenative Musaicing

Markov Processes Follow from the Principle of Maximum Caliber

Markovian Maximal Coupling of Markov Processes

On a Markov construction of couplings

One-sided Markov additive processes with lattice and non-lattice increments

Some characterizations for Markov processes as mixed renewal processes

Continuous flows driving Markov processes and multiplicative $L^p$-semigroups

Supplement to "Markov Chain Monte Carlo Based on Deterministic Transformations"

Long Time Behavior of General Markov Additive Processes

Contraction of Markovian Operators in Orlicz Spaces and Error Bounds for Markov Chain Monte Carlo

Markov Chain Approximations to Singular Stable-like Processes

Improving the convergence of Markov chains via permutations and projections

Explicit computations for some Markov modulated counting processes

Ensemble preconditioning for Markov chain Monte Carlo simulation

Parallel and interacting Markov chains Monte Carlo method

Convergence of sequential Markov Chain Monte Carlo methods: I. Nonlinear flow of probability measures

Asymptotic properties for linear processes of functionals of reversible Markov Chains

Gelation in Vector Multiplicative Coalescence and Extinction in Multi-Type Poisson Branching Processes