Usage

Registration

To gain access to VICTOR, you must first register. Currently, we are authenticating users through Github, Gmail, or Outlook, and thus all users are required to possess an account of one of the three. To register, please visit the VICTOR website and click on the Sign Up button on the right of the menu. This will direct you to a form where you can provide the neccesary information. Please note that the authentication process is manual, and thus may take up to 24 hours to complete, possibly longer during weekends or breaks.

Choose an Environment

On initial login to the hub, you will be prompted to select an environment. You can choose a machine with a or without a GPU. The VICTOR team recommends the machines without GPU access unless CUDA or similar high peformance frameworks are used in one’s work.

_images/images.png

There are three images available to choose from. The first image is the default VICTOR environment, and the only one configured to run models. The majority of users should choose this. The Pangeo images prioritize the larger machine learning packages, PyTorch and TensorFlow in particular. These are recommended for advanced users.

_images/type.png

Finally, you must choose a resource allocation for your image. The equivalent to a modern laptop is the 7.4 GB allocation. If running light models or none at all, smaller machines are recommended, and will also save our team money! For heavy workloads, use up to 60.6 GB of RAM. If a need arises to use the the 121.2 GB allocation, please contact victor@ldeo.columbia.edu so we can prepare accordingly.

_images/resource.png

Once you are able to access the hub, you will see a screen very similar to this:

_images/victor.png

The Launcher

The first screen you will be met with is the launcher, providing an variety of environments to utilize. The first row of buttons is comprised of highly interactive Jupyter notebooks, as well as the desktop application. The VICTOR team advises using only the [conda env: notebook] tabs, as this is the environment that includes all the neccesary scientific packages for utilizing and visualizing models.

The second row of buttons opens consoles, a simplified interactive environment where you can enter lines of code and run them, though not able to embed images or renderings such as matplotlib.

The final row of buttons provides quick access to static files in various languages, as well as for a simple text file. Additionally, access to a bash terminal is accessible here, allowing for advanced users to maximize productivity in this environment.

Run Models

To run models, there are a few simple steps. First, navigate to the shared folder. You can either run the the *****_setup.sh files directly, or copy one, either through manual selection (right click + copy) with the mouse, or through the terminal with cp shared/*****_setup.sh . when you are in the home directory.

Note

All files in the shared folder are read/execute only. If you would like to contribute models, data, or ideas for improvement, please contact victor@ldeo.columbia.edu.

Once this runs, you will have all necessary files contained in a new folder in your home directory. Most folders will simply contain the executable and the example notebook. All Juptyter notebook workflows will generate most of the supplemental files necessary for the model to run.

However, DEMs are not automatically included. Users then have 4 primary options: 1. Navigate to the DEMs folder in shared and copy the relevant file to your home directory, if the file needs to be altered. 2. Read the DEM from $HOME/shared/DEMs directly in your workflow, if no changes are required. 3. Import your own DEM from a local machine, dragging and dropping into the file tree. 4. Utilize S3 buckets or an SQL connection (using boto3 or mysql python packages) to load files remotely.

At this point, users can go through the first few cells immediately succeeding the import statement, inputting parameters as needed. Thorough descriptions of each parameter are included. Once finished, the user can simply press the fast forward symbol to run all cells, or shift + enter/return to run each cell individually.

Accessing The Virtual Desktop

Upon clicking the desktop button on the launcher, another tab will open, displaying a screen as seen below. From here, you can access a fully featured QGIS installation, with many other programs scheduled to be added as well. The file system is connected to your personal files, and a web browser is also available for your convenience.

_images/desktop.png

Forum

The VICTOR team has created a forum for the computational volcanology community to ask questions, share ideas, and discuss potential improvements. The forum is catagorized into various subsections of study, from geodesy to seismology to petrology. There are also catagories focused specifically on the VICTOR platform, as well as general questions for the community. Any interested user can sign up with just an email address. Upcoming uses for the Computational Volcanology forum include threads for course assignments and integration with other volcano communities.

To access the forum, please navigate here.

Citations and References

Below are citations and related works used to create this project. To add additional citations or for clarification, contact victor@ldeo.columbia.edu

Conflow

See Conflow/Confort here.

Hazmap

See Hazmap here.

HYSPLIT

See Hysplit here.

IMEX-Lava

See IMEX_LavaFlow & IMEX_SfloW2D_V2 here.

Molasses

See MOLASSES here.

MrLavaLoba

See MrLavaLoba here.

pyFLOWGO

See pyFLOWGO here.

Sulfur_X

See Sulfur_X here.

Scoops3D

See Scoops3D here.

Tephra2

See Tephra2 here.

Titan2D

See Titan2D here.

VENUSS

See VENUSS here.

VolcFlow

See VolcFlow here. Contributing Models and Hub Additions ————————————- If you believe your model would be a good fit for our platform, please email victor@ldeo.columbia.edu with a link to the code on a version-control platform as well as a brief explanation. For additions to the hub itself, please refer to our Github repository. Create an issue for general advice, or create a pull request for specific changes/updates.