Parallel C++ 11 implementation of a nonparametric simulated maximum likelihood estimation. The implementation is done according to the study: Kristensen, Dennis, and Yongseok Shin. "Estimation of dynamic models with nonparametric simulated maximum likelihood." Journal of Econometrics 167.1 (2012): 76-94.
The library provides functionality for estimating Vasicek process with an analytical method and the NPMSLE on simulated or empirical data. It also provides functionality for estimating a 3-D process, with Vasicek process as a base incorporated to the Heston stochastic volatility model, also with simulated and empirical counterparts. The final option provided is a replication of the original jump-diffusion process presented in the paper, section 4.1.2.
The parameters of the simulations and optimizations can be set in the Globals header. Also, options for logging and loading can be established in that header. Currently, data need to have form of 3 comma-separated columns with titles (the last column used for the Vasicek process).
OpenMP version 4.7
NLopt Library (https://github.com/stevengj/nlopt) version 2.4.2
Makefile is provided for default building, with a necessity of changing the include (with nlopt headers and headers of this project) and library (with nlopt libraries) directories.
Run main.exe -TEST.
cmake version 2.8.12.2
gcc version 7.2.1
This project is licensed under the MIT License - see the LICENSE.md file for details.