Work Log

2 hrs.

1/24/11

4hrs.

• Since the results were very poor, I changed some parameters and other aspects of the experiment and I am rerunning it. Now the strengths of the vortices are known. I am looking at $\Delta t = 1.0, 1.5, 2.0, 3.0$ and $\sigma = 0.02, 0.03, 0.05$. I am using the same starting location for the tracer.
• I also changed the time increment used for integration. Now my integration used a $dt$ of 1/200 instead of 1/100.

1/20/11

2 hrs.

• Wrote a script to perform an experiment to look at how the particle filter performs with different noise parameters and time step size. I used time steps of $\Delta t = 1.0, 1.5, 2.0$ and $\sigma = 0.02, 0.03, 0.05$. The tracer started at the location $0.3 - 0.6 i$. The vortices both have a strength of $2\pi$. Results are poor; they are much worse than those from Chris's paper using the same parameters. In my experiment the strengths of the vortices were unknown, but in Chris's they were known. Next, I will run the same experiment with the strengths known.

1/19/11

5 hrs, 30 mins.

• Cleaned up the code a lot. Previously, an entirely different cloud of particles (with different realizations of the random noise process used during integration) was used for the Lagrangian frame, instead of just appropriately rotating the cloud to get the Lagrangian frame. The particle filter now uses the same cloud for plotting the movement of the cloud in both frames, but simply rotates the axes through time according to the constant angular velocity of the rotation of the vortices.
• Particle filter now computes and records a time series of the variances of the weighted distributions of particles (for each state variable).
• TODO: Add papers to the wiki page!

1/18/11

• Tested code in Octave. It works!
• TODO: Create script to save parameters as a text file or something.

1/13/11

2:00 - 7:30

• Created my Wiki page.
• Organized research files.
• Created generic particle filter function to accept parameters as arguments. Only allows two vortices of unknown strength.
• Set up experiment script to run multiple simulations and save the resulting distribution of error time series. Each experiment now uses the same seed for random number generation.

Meeting with Dr. Spiller

• Set up 2:30 meetings on Tuesdays and Fridays.
• Need to use the same parameters as Chris's paper and same criteria for 'filter divergence' and see if I get the same results.
• Systematically look at the change in the error distributions as the initial distribution of particles changes.
• Test code in 'Octiv'.
• Email Feng Xishou for 'Pere' cluster permission.