1. Program Discription: The program is for replication of learning the Linear Gaussian Model as discussed in "Learning Made Easy: A Marginalized Resample-Move Approach" by Andras Fulop and Junye Li (2010). The paper can be obtained at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1724203 or at our personal websites. 2. How to use: Run the main program, MainProgLin.m. The particle filter in PF_newobs_lin.m and PF_wholesample_lin.m is a full-adaptive PF as discussed in Pitt and Shepard (1999) and Johannes and Polson (2008). The program is relatively self-contained. When adapting to your own model, you need to design your own PF and correspondingly revise functions PF_newobs_lin.m and PF_wholesample_lin. Of course, the priors also need to be modified in functions PriorSim.m and PriorLogl.m. 3. Further assistance: If you have any problems when using this algorithm, please feel free to contact us at fulop@essec.edu or li@essec.edu
Working Paper (2021)
Fernández-de-Marcos Alberto and García-Portugués Eduardo