Docker>> Ensemble-Kalman-Filter>> 返回
项目作者： Kogl1n

项目描述 ：
EnKF in Matlab for a system of two PDEs modeling burglar behavior (model of Short et al.)
高级语言： Matlab
项目地址: git://github.com/Kogl1n/Ensemble-Kalman-Filter.git
创建时间： 2018-03-06T11:56:23Z
项目社区：https://github.com/Kogl1n/Ensemble-Kalman-Filter
开源协议：
下载


Ensemble-Kalman-Filter

EnKF in Matlab for a system of two PDEs modeling burglar behavior (model of Short et al.)

Note:

I have experienced several bugs in Matlab’s solvepde() function on Matlab R2016a.

  • For instance, the stated restart from the last time point does not work for systems of PDEs.
    My workaround consists in defining an anonymous function that takes the value of the last (point of time of the) solution. This function is taken as the new initial value in the solution process.
  • Saving the workspace after a successful simulation, clearing all variables and loading the workspace will result in an error (“initial conditions either do not exist or are empty”) trying to delete the initial condition model.deleteInitialCondition and setting a new one using SetInitialConditions via anonymous function on the loaded model. I guess some internal variable is missing. Workaround: Run the MAIN function for a couple of seconds and abort. This will apparently load the missing data.
  • When doing simulations, clearing all variables results in a small chance some internal variable still persists.

Since I don’t own the Matlab license for the university version, my bugreport won’t be even looked at.

Files

  1. Short.m: Solves Short's et al. PDE system with the following coefficients and initial value:
  2. acoeffunction.m: Coefficient a function for our PDE system
  3. ccoeffunction.m: Coefficient c function for our PDE system
  4. fcoeffunction.m: Coefficient f function for our PDE system
  5. u0fun.m: Initial value u0 function for our PDE system
  7. EnKF.m: Ensemble Kalman-Filter performing the aggregation between last aggregated solution simulated 
  8.         to the next time point with the observational data for that point.
  9. EnKF_ens_create.m: creates the ensemble
  10. EnKF_main.m: MAIN
  12. save_fig.m: saves all open plot windows as portable network graphics
  13. subplot4.m: Arranges the solutions as four subplots: artificial solution, 
  14.             noisy observation, solution prior to EnKF and aggregated solution (EnKF).
  15. short_bsp.m: Example script for the unfiltered solution of Short's model.