gully

OK, I fleshed out some of the details here. We definitely _don't_ want to use the non-normalized (_i.e._ raw) model grids in the spectral emulation process, for the reasons I...

Ok, we figured this out after some deliberation. Note the portion of the code in `grid_tools.py` ```python #If we want to normalize the spectra, we must do it now since...

Short update on the flux standardization. I'm tracking scalar values in front of the emulated spectra so that we can maintain absolute fluxes. In `PCAGrid.create` in `emulator.py`: ```python # Normalize...

Demo of `flux_scalars` over a grid for a range in Teff (shown), and small range in logg and [Fe/H] for a user-input wavelength segment and resolution: (Don't read too much...

Here is a commit that implements the flux scalars approach in `emulator.py` on my fork: 7b2afc7 This code **breaks backwards compatibility**, since it inserts a new positional argument that is...

In short, I think HDF5 is fine, and if we wanted to switch I would be inclined to go with some form of ASCII, either csv or JSON. The new...

I am seeing that this PR breaks 1 test: `py.test -vsk test_filenames_are_set_properly` Currently investigating...

I fixed the failed test which in hindsight was an edge case of a None header. Also added a warning logger. I'm a bit muddled on this PR. On one...

Here is an example of [wobble working on Google colab](https://colab.research.google.com/drive/1ujSc9JTfQwml0XWg0LnJk8ZuX-5OEiuq). [Colab supports several external data sources](https://colab.research.google.com/notebooks/io.ipynb), including Google Drive, which I've used here. Unfortunately that means that the hardcoded data...

The absolute flux units are lost in the emulation process for the relatively unavoidable reason that the spectral grid has to be [prewhitened](https://en.wiktionary.org/wiki/prewhiten). The procedure is described to some extent...