Alan, Deirdre, Jan, Soumen, Sayantani, Ish, Parthapratim, Poulomi, and Arun
## Agenda
Demonstration of the [code](https://git.ligo.org/sayantani.datta/pca_tgr/-/blob/main/pca/code_demo/transform.ipynb) for calculating PCA parameters for individual events.
## Minutes
- Minute taker: Parthapratim
- Minutes from last call: [link](https://git.ligo.org/sayantani.datta/pca_tgr/-/wikis/June%202,%202023)
- Alan- LALINFERENCE was used in 03 and results are stored in .hdf5 file. BILBY will be used in O4 and results will be stored in .json file. Write a code that reads both .hdf5 file and .json file.
- Sayantani- We will write it.
- Jan- PESummary is the solution.
- Soumen- You can use PESummary. It can read any file .hdf5/.json.
- Alan- PE group is going to change a lot of things in 04. Did TGR group talk with PE group ? Check that !!
- Sayantani- We will look into that.
- Alan- Check the covariance matrix. Check if all the eigenvalues are positive. Check!! Eye balls won't catch it ! Write a code to check it.
- Sayantani- This is just a consistency check. This code is not a part of the main script.
- Alan- Important check! Need to be integrated with the main script. The eigenvalues are telling you, if you need to take the first 2 PCA, first 3 PCA, or first 4 PCA parameters. The eigenvalues will tell you how many PCA parameters should be considered for different events.
- Arun- Super important! Add this code with main script.
- Arun- Demo script shows the result from real event run or, injection ?
- Sayantani- Injection with GW151226 like systems.
- Soumen- While writing [\theta_{PCA}] = [U]^T * [\theta] in the code, you can use single np. instead of multiple np.
- Sayantani- Ok.
- Sayantani- Should I add this demo notebook in the wiki page ?
- Alan- Yes, add that.
- Alan- It's nice and simple. Few generic comments. Look at your imports! None of them are specific to LIGO except BILBY. But you are not using BILBY anywhere in the main script. Just remove it from import! In general for LIGO specific packages write separately in the imports and mention that. Write the name of the contact person in the beginning of the notebook. Write 'what does the code do' briefly at the beginning. Write about each cells using Markdown in Jupyter. These are important as the codes are to be made public.
- Arun- In these codes there is nothing specific to TIGER/FTI ?
- Sayantani- No. General to all; TIGER/FTI.
- Arun- Reviewers, please make sure that while reviewing the code. Phase deformation parameters are differently named in TIGER and FTI.
- Sayantani- I will check that and modify the code accordingly.
- Arun- Take few examples of FTI and run by yourself. Get it done ASAP.
## Action items
- Write a code for reader that can read both .hdf5 and .json file. PESummary may be the solution.
- Add the code that checks the covariance matrix (if all eigenvalues are positive) with the main script.
- Decorate the notebook as par Alan's suggestion.
- Phase deformation parameters are differently named in TIGER and FTI. Need to modify the code accordingly.
- For Sayantani: Take few examples (multiparameter runs) of FTI and run by yourself. Get it done ASAP.
