Bug in Psi4 Loading

[vaishakp]

I tried fixing code on Psi4 loading, and I failed. There are some things I don't understand, amongst other things like:

1. Unordered time axis in Psi4 loading code) before interpolation. Throws interpolation error.
2. Every mode now being a pycbc time series (Is this a new sxs thing?). This has some disadvantages.
   a). pycbc time series is sometimes shorter by one element compared to original data, probably associated with rounding off in delta_t. Here in comparison to WaveformModes.n_time and WaveformModes.time
   b). It is unnecessary and expensive to carry a time axis, delta_t and all attributes for every mode when they are identical across them.
3. The number of columns in Psi4 RIT data is 6 not 3 as assumed
4. Interpolation is more accurate in Amp and phase and they are provided in RIT data.
5. When I originally authored the code in https://gitlab.com/vaishakp/waveformtools/-/blob/main/waveformtools/dataIO.py?ref_type=heads#L428
   I found that there are issues sometimes when delta_t is not rounded off, after using scipy.stat.mode to discern it, before constructing a new time axis for uniform sampling

[adivijaykumar]

Every mode now being a pycbc time series

No this is not a sxs thing, rather a nrcatalogtools thing. However that should just be a problem with the get_mode method, and everything that works for WaveformModes in sxs (see this notebook) should ideally work. I think it is worth discussing whether the default return of get_mode should be a pycbc TimeSeries instance.

[adivijaykumar]

Hmm, I think there are other problems in https://github.com/gwnrtools/nr-catalog-tools/blob/master/nrcatalogtools/waveform.py#L83 which are different from the usual sxs behaviour, I'll try to take a look soon.

[vaishakp]

For point 2 please see:

rit_ffi (1).pdf

Psi4 data was loaded here using waveformtools's routine

[vaishakp]

Generating polarizations from Psi4 using nrcatalogtools routine now gives the following error