Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

CCA Stability with latest PyQt5 [5.15.0] #3

Open
ispingos opened this issue Sep 18, 2020 · 1 comment
Open

CCA Stability with latest PyQt5 [5.15.0] #3

ispingos opened this issue Sep 18, 2020 · 1 comment
Labels
bug Something isn't working critical issue of critical priority

Comments

@ispingos
Copy link
Owner

One of the latest updates in PyQt5 (currently using 5.15.0) has caused stability issues to the CCA processing (fully automated catalog processing).

The issue causes unexpected crashes while running CCA (both the GUI and the background executed code) without any raised exception or error message.

This is currently worked on and a fix should be posted soon(ish).

@ispingos ispingos added bug Something isn't working critical issue of critical priority labels Sep 18, 2020
ispingos added a commit that referenced this issue Apr 3, 2021
*IMPORTANT* This release is at an early development stage. New functionality is NOT detailed in the documentation. If you intend to use it, please refer to the following comments as well as the included README.nd file.

NEW FEATURE: MULTIPROCESSING
- Added a rough multiprocessing scheme of Cluster Analysis (CA) for Pytheas.
- This includes fixes concerning stability issues (see #3)
- A new thread will spawn additional processes for processing the given event catalogue in multiple cores, in parallel.
- Added a 'number of cores' selector in the CCA preferences window.
- CA for individual event-station pairs through the GUI is DISABLED until further notice. This is a feature we intent to reactivate in the future.
- Parallel processing has been tested in various large datasets (over 10,000 event-station pairs).

NEW FEATURE: CLI - EXPERIMENTAL
- Better be avoided until further notice.
- Moved several functions from `pytheas.py` to `tools.py`
- Revised and updated several functions related to CA/CCA
- Removed `QThread` functionality from `clustering.py`

NEW FEATURE: ALTERNATIVE PICKS
- Added the feature of using different picking methods, if the observed ones are missing from the catalogue, for CCA.

NEW FEATURE: SPECTROGRAM
- There is now an option to view the spectrogram of the current stream

ACQUISITION SCRIPTS
- Renamed multiprocessing module(s) for clarity.
- Added multiple providers in `fdsn_get_waveform_data.py`, to enable downloading datta from multiple nodes in one session.
- Cleaned logging in `fdsn_get_waveform_data.py`

EXCEPTIONS
- added an `exceptions.py` module for custom exception classes

SNR-FILTER FIXES
- Moved the SNR and automatic filtering processes to the `tools` module
- Peformed various cleanup actions for the SNR/auto-filter processes
- The new SNR should now properly be displayed when applying/removing filters

GENERAL
- Updated the model with better calculated densities
- Spectra of the horizontal channels will now represent the current state of the stream, instead of the initial. This means that if you now filter a stream, you should see the corresponding frequencies in the spectra.
- Added some system information logging
- Added a new function `tools.grouper` to permit more fine grouping of iterables.
- Added a new function 'tools.grid_dimensions` to permit the estimation of the 'best' grid dimensions for figures/layout
- Did a pass on the prerequisites installer scripts.
- Added a workaround in cases where detrending fails. This should help in cases where event/station pairs would be omitted because `detrend` raised an exception.
- The number of C95 contours should now be properly determined!
- Minor fixes/adjustments in CCA.
- Fixed various issues with switching between events and/or stations.
- Even if the angle of incidence changes after opening a new station (i.e. the angle is recalculated), the order of the stations should no longer change.
- The Calinski-Harabasz plot should now REALLY not show up in the CA figures (for real this time!).
- Swapped the Fast-Slow channels with the Radial-Transverse ones, in the EV/ME/RC plots. This should provide a clearer picture of the quality of the correction (since the transverse component is now shown).
- In the EV/ME/RC plot, the Radial-Transverse channels are either `RT` or `QT`. This is defined by the user in the `Splitting` toolbar.
- Cleaned up the logging a bit (removed all those bloating `findfont` debug messages from `matplotlib`).
@ispingos
Copy link
Owner Author

ispingos commented Apr 3, 2021

After an overhaul of the processing scheme, this issue now seems to be fixed in 138f50b. However, it has not made the master releases yet (will do after validation, in the near future).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working critical issue of critical priority
Projects
None yet
Development

No branches or pull requests

1 participant