v0.1.0

The addition of the option to toggle bandlimited interpolation, rather than rely on cubic interpolation, to extract field values away from the vertices/spatial grid.

What's Changed

• Introduction of band-limited interpolation schemes by @willGraham01, @samcunliffe, @prmunro.
• Refactoring of the testing framework by @dstansby, @willGraham01.

Users can how specify on the command-line whether tdms should use the cubic interpolation methods (which were the only interpolation methods available in v0.0.1), or the band-limited interpolation methods (detailed below).

Band-limited interpolation has replaced cubic interpolation as the default interpolation method. The -c or --cubic-interpolation flags can be passed to the tdms executable to override this behaviour and force the use of cubic interpolation again.

Band-limited interpolation (BLi)

When a Yee cell has a spatial dimensions of the order of $\lambda/6$, where $\lambda$ is the shortest wavelength of interest, interpolation schemes such as cubic interpolation that only use 4 sample points are suitable. It is often the case however that larger Yee cell dimensions are being used, in which case a BLi scheme (which uses 8 sample points) is required. Much like cubic interpolation, the formula for the interpolated field value when using BLi is a weighted sum of the sample points, although derived from ideas in Fourier analysis as opposed to fitting a polynomial through the provided sample points.

Since BLi requires 8 samples to produce an interpolated value, there are some situations where the executable will be forced to fall back on cubic interpolation if running particularly "thin" simulations (with $<8$ Yee cells in a particular dimension). Interpolation is done on a directional basis, so tdms will always attempt to use BLi if there are $>8$ Yee cells available in a particular dimension even if there are less than this in other dimensions.

Full changelog: https://github.com/UCL/TDMS/commits/v0.1.0