I've made a fresh start on the noise budget code, using much of the work and many of the functions from aligoNB, but in a code that should be overall simpler for us to work with. This exists in a repo here: simplenb.
Noises estimated using the excess power method can be read in either using gps times or from a dtt template. .xml files are tracked using git lfs and read in by dttxml. If a list of gps times is provided then the code will download the data using gwpy and save asds as a .pkl file, so that the data doesn't need to be downloaded each time the budget is run. Deleteing the pkl file from the local copy will cause it to redownload the data if the user wants to change times or resolution. The excess power projection is done using the frequency resolution of the dtt file, or right now it is hard coded to an fft length of 3 seconds for those made using a list of gps times. The noises are rebinned to a frequency vector specified for the budget after the projections are made.
Thermal noises are imported from gwinc. Quantum noises are also calculated using gwinc, which is why this code needs to be run in an environment that uses Kevin's superQK branch of gwinc. I've added a function save_quantum_params to my quantum noise modeling repo that takes a template of a yaml file including all the parameters you think are important for modeling quantum noise, and a ifo struct, which saves a yaml file with the current parameters. This function is also available in the quantum_utils.py function in this repo, so that people can use it with any method they like of generating a gwinc model of quantum noise to incorporate into this noise budget. I think it is best to leave the quantum noise budgeting separate from this repo for now, because the code for that is still evolving. For now I've used the quantum parameters where all the missing squeezing is explained by frequency dependent loss from 85942, the next step is to add mode mismatches to that model.
Running the budget.py script will generate noise budget plots, you can choose how many layers deep to go in plotting sub-budgets by setting that parameter in the make_all_plots function call in the final lines. The plots attached were made using the injections that Camilla made this morning (86550), and a reference time of 20:00 UTC.
Still to do:
- Add missing noises:
- residual gas
- DAC noise
- PD dark noise
- Newtonian Noise
- remainder of ASC noises
- Correct some noise terms:
- intensity noise has double counting around 1kHz (also an issue in old noise budget)
- frequency noise loop correction (it is measured with only one sensor in loop, in full lock we use 2 sensors)
- Would be nice:
- specifiy the ftt length for excess power noises using gps times for each noise
- get Craig's interactive svg working, which isn't working in either model yet
- add some useful information to plot, like gps time of reference, maybe some quantum model parameters
