Supplement 1. Matlab code containing the MCMC algorithms used in the example.

File List cluster_MCMC.txt -- Matlab code for MCMC clustering algorithm Description cluster_MCMC.txt performs MCMC clustering and source assignment for multivariate data sets following the procedures described in the main text and appendix. To use in Matlab, change the file extension from .txt to .m and place in the Matlab working directory. cluster_MCMC is then executed from the Matlab command line; required input arguments are the data matrix ("X"), the proposed number of sources ("K"), the number of MCMC iterations to perform ("M"), the number of iterations to discard as burn-in ("burn"), and the number of iterations for the relabelling algorithm ("Unswitch"). The function returns a structure variable with fields containing source assignments and a variety of summary statistics, including DIC. Users should try a range of values of K, making multiple runs fo a given K to ensure convergence. DIC values for different values of K should then be compared to identify the correct number of sources. Additional instructions are embedded in the file.

文件列表：cluster_MCMC.txt —— 用于马尔可夫链蒙特卡洛（Markov Chain Monte Carlo, MCMC）聚类算法的Matlab代码

本文件cluster_MCMC.txt可按照正文及附录所述流程，对多变量数据集执行MCMC聚类与源分配任务。若需在Matlab环境中使用，请将该文件的扩展名从.txt修改为.m，并放置于Matlab工作目录内。随后可在Matlab命令行执行cluster_MCMC；其所需输入参数包括：数据矩阵("X")、拟设的源数量("K")、需执行的MCMC迭代总次数("M")、作为烧失期需舍弃的迭代次数("burn")，以及重标记算法的迭代次数("Unswitch")。该函数将返回一个结构体变量，其字段涵盖源分配结果与各类汇总统计量，其中包括偏差信息准则（DIC）。用户应当尝试多组不同的K值，并对同一K值开展多次运行以确保算法收敛。后续可对比不同K值对应的DIC值，以确定准确的源数量。文件内部嵌入了额外的操作说明。