Cvmbycols

From SCECpedia
Revision as of 20:31, 8 March 2011 by Patrices (talk | contribs)
Jump to navigationJump to search

Overview

Cvmbycols is a set of codes for producing compacted Etrees for geographic regions with material properties drawn from a community velocity model. Serial and MPI versions of the program are provided.

Algorithm Description

Optimized Etree Layering to Support Freq, PPWL, and Min Vs

  • Divide region in set of x,y columns, where the length of each column equals its width
  • Starting at depth = 0m, extract a 2D grid for the column at the resolution needed to support the max frequency, pounts/wavelength, and minimum Vs.
    • If the minimum Vs is in this 2D grid allows for a lower resolution, and the current z value is divisible by the lower resolution, requery the CVM at the lower resolution. Otherwise, if the min Vs in this grid needs a higher resolution, increase the resolution to what is required (up to the maximum resolution dictated by the configured supported minimum Vs) and requery the CVM.
    • Write this layer to an Etree (either a final etree or an etree local to this core)
    • Step the depth down by the resolution of this 2D grid and return loop.


Programs

Cvmbycols Serial

Divides etree region into columns, then extracts each column from the underlying CVM and inserts the octants into the final Etree.


Cvmbycols MPI

There are two main steps.

Extraction: Division of the etree region into X columns for extraction. This is an embarrassingly parallel operation. A dispatcher (rank 0) farms out each column to a worker in a pool of N cores for extraction. Each worker writes its own local etree. After program execution, there are N small etrees. The extractor must be run on 2^N + 1 cores.

Merging: Merging of the N small etrees into the final Etree. This is essentially a merge sort on the keys from the Etree addresses extracted from the local etrees. The cores at the lowest level of the merge tree each read in octants from two etrees with a pre-order traversal, merge sort the addresses, then pass the locally sorted list of addresses to a parent node for additional merging. This proceeds until the points rise to rank 1 which has a completely sorted list of etree addresses. Rank 0 takes this sorted list and performs a transactional append on the final etree.

Merging is a very fast opertion. Speeds in excess of 3M points/sec have been achieved on NICS Kraken.


Defining properties of an Etree

The following properties are defined:

  • Bounding box of geographic region in lon,lat coordinates
  • Extents of Etree volume in 3D, in km
  • Number of columns in x,y axis to partition 3D space. The resulting columns must be square along the x-y axis.
  • Frequency limit to support in Hz
  • Number of points per wavelength to support
  • Minimum Vs to support
  • Maximum octant size to allow
  • Name of the output etree file
  • CVM to use for material properties (from UCVM)


User Guide

Source Code

SVN: https://source.usc.edu/svn/cvmbycols


Supported CVMs

The cvmbycols utility is built upon UCVM. As a result, it can extract etrees from the following CMV sources:

  • CVM-H
  • CVM-S
  • USGS Cencal (which itself is an etree)
  • 1D