Homework 3 was due this week, we were studying the spread and biases of various location and dispersion estimators on Gaussian data and data with outliers. We also started talking about jacknife and bootstrap methods.
Still working on debugging the VL+CT integrator. Here’s what I accomplished this week
- I found a couple of style errors, now whenever I have a pointer or the half time or standard grid it’s always named
workingGridinstead of a mix of
- I fixed the bug that was occuring at the right boundary. It turns out that since the magnetic fields are stored on cell faces you need 1 extra array element to store the face on the rightmost (or leftmost) side so that all the cells have magnetic fields on both faces.
- The PLM function was treating all the magnetic fields the same and reconstructing all of them when it should have been getting the \( x \)-direction field directly from the grid.
- Checked the entire integrator except the CT field calculations for bugs.
- Checked the 3D virtual grid I use for calculating the CT fields
The current results of the Brio & Wu shocktube are below,