Installation¶
Install DeepC¶
You may install and update the DeepC package using gsm - gempa software management tool [12] or by unpacking a package file which you need to download first. Our preferred installation path works with gsm.
Installation using gsm:
First time only (works also for SeisComP and all other gempa modules):
Get and install gsm [12] as outlined in the gsm documentation .
Set up gsm as outlined in the gsm documentation .
Regular DeepC installation and update:
cd $HOME/install/gsm ./gsm update ./gsm install deepc
This installs four packages:
deepc - this application
dlbase - dependency package for deep learning applications
dlmodels-class - DL models, installed according to the gsm configuration of data_dir in
gsm.conf, typically /home/data/dlmodelsdlmodels-pick - more DL models, installed according to the gsm configuration of data_dir in
gsm.conf, typically /home/data/dlmodels
The DL models are described in section Available trained models.
The following packages are optional but recommended:
dlbase-extend - CUDA dependencies, can be omitted if already present on your system
dltools - only necessary if you plan to use your own pretrained model(s)
Unpacking the DeepC package file (not recommended):
Download the DeepC package TAR ball from our package server. The server address and login credentials will be communicated with you. The file name will take a structure like
deepc-<version>-seiscomp-<SC version>-<OS>-<OS version>-<architecture>.tar.gz. The downloaded file must match your current version of SeisComP and your operating system.Unpack the file and extract the package it into your SeisComP installation, e.g.:
cd $SEISCOMP_ROOT/.. tar zxvf deepc-<version>-seiscomp-<SC version>-<OS>-<OS version>-<architecture>.tar.gz
The DL models and the dependency packages must also be installed accordingly.
Install dependencies¶
On Debian-based systems:
On RedHat systems: