UCVM Prototype

Introduction

We are targeting the development of a new UCVM (tentatively slated for release in September 2016). This version will represent an overhaul of the existing UCVM framework and include a number of major technological changes:

• Python 3-based
  • Compatible with Anaconda
• Embedded Python APIs for interacting with the platform
• Easier installation (python setup.py)
• Must work on the latest supercomputers

The goal would be to minimize the amount of C code that we need to keep in the platform. We will have to evaluate the speed of the Python version and ensure that it can generate the e-trees efficiently.

Supported Models

The models that UCVM will support moving forward include:

• CVM-S4
• CVM-H 15.1.0
• CCA
• CVM-S4.26
• CVM-S4.26.M01
• USGS Bay Area
• Lin-Thurber
• Broadband 1D model format (including Whittier Narrows)
• SCEC 1D background model (modified Hadley-Kanamori)

Development Timeline

Proposed initial sprint:

July 5 - July 8, 2016

Merge the new plotting utilities and the very basic prototype of UCVM that I had developed earlier. Develop the following capabilities:

• Develop a Python CCA model code that works with iteration 6 of the model
• A Python-based ucvm_query-like utility that can read this model as well as CVM-S4 (the legacy Fortran code)
• Make the setup.py script auto download the new CCA and legacy CVM-S4 code and build the Fortran code like before
• Make the plotting scripts use this new cvm_query-like utility and Python code to extract the vertical profiles from:
  • The SRL paper
  • The CCA profiles for the CyberShake model
• Develop tests for all of this to ensure it is working correctly using the standard Python unit test framework