State Estimation with Stable Errors
and Whisker Particles

(Revised 3/10/21.)

Paper: "State Estimation with Stable Errors and Whisker Particles"

(Revised 8/18/20) to eliminate the "failures" in the Mixture Kalman Filter, and 3/10/21 for presentation.)

Matlab Programs

Stable Whisker Particle Filter Routine: StableParticleFilter.m
Supporting subroutines
- DrawCEsample.m
- HhatCE.m
- Nolan STABLE package
- Matlab 2016+ Statistics Toolbox
Illustrative driver program producing Figure 7, with plots = 1 and N = 100. For bubble graphs, set plots = 2 and N = 10 or 20 in program, then set xlim([51 58]) etc. in command window.
Simulation program for Whisker Filter, set to alpha = 1.7 etc, N = 100, 25% adaptive.
Stock program
- Stock data
- T-bill data (for XS returns)
Inflation program
- Inflation data
Bitcoin program
- Bitcoin data

Note: In the following programs, the Mixture Kalman Filter is referred to as the "Kernel Filter."

Mixture Kalman Filter Routine: StableKernelFilter.m (Revised 8/18/20 to eliminate "failure" cases.)
- Additional subroutines:
Mixed Kalman Filter demo program
Mixed Kalman Filter simulation program

Page written Aug. 8, 2020
revised March 10, 2021

Up to JHM homepage.