Seismic-2d

This is a simplification of a model originally used for global-scale seismology. The full model can be read here:

NET-VISA: Network Processing Vertically Integrated Seismic Analysis. Nimar S. Arora, Stuart Russell, Erik Sudderth. Bulletin of the Seismological Society of America (BSSA) April 2013, vol. 103 no. 2A pp709-729.

Please cite the above paper in scientific research using this model or the included data.

Files

• description.tex -- a description of the model, also see the downloadable file description.pdf.

• generate.py -- generates the physics of a 2-D world and some episodes

• util.py -- some geophysical utility functions

• solve.py -- a sample solver that learns the physics and solves the episodes

• pysolve.py -- a python-based solver

• csolve.c -- a C-based solver, identical to pysolve.py but much faster

• evaluate.py -- evaluates a solution versus a reference

• mwmatching.py -- utility script for max-weight max cardinality matching

• short_data/ or large_data/ -- data directories created by uncompressing the downloadable file data.tar.gz. These directories contain the following files:

  • physics.data -- physics for the training and test episodes
  • training.data -- 100 (or 10K) episodes for training
  • test.data -- another 100 (or 10K) episodes for testing
  • test.blind -- the test data with the event-to-detection mapping omitted
  • test.solution -- a sample solution on the test data

Overview

The simplified model is completely described in description.tex and this should be translated in the Probabilistic Programming Language of your choosing. Your model can be tested on the included data files in the Downloads section of this repository. The unlabeled test data in test.blind and the labeled training data in training.data comprises the observations to the model. The query of interest is the seismic bulletin for each of the observed episodes in the test data.

Once all the bulletins have been produced, the script evaluate.py can be used to produce the reports on the accuracy versus the reference script test.data. One can also compare the results versus the baseline in test.solution. For example to check the accuracy of the sample solution:

$> python evaluate.py short_data/test.data short_data/test.solution 

188 matchable events, 243 guess events, and 140 matched

Precision 57.6 % , Recall 74.5 % , F1 65.0

Time Errors mean 8.4 std 8.0

Dist Errors mean 1.6 std 1.3

Mag Errors mean 0.3 std 0.2

The files generate.py and solve.py have only been provided for convenience. These files shouldn't normally be used. However, if you want to check the performance of your model on more than just the provided data you may generate more as needed. The sample solver is based loosely on the published greedy algorithm, and may be used as a very simple baseline.

The sample solver first tries to run the C-based solver, and if that is not found it attempts the python version. Both versions of the solver are identical except that the C one is a lot faster. Note that the output of the solvers is already included with the provided data, so it is only necessary to run the solver for additional data that you might generate.

Note that the sample solver cheats a little bit by directly looking at the underlying physics rather than learning it from the training data. Your solver should not be reading any of the following files -- physics.data, test.data, or test.solution.

Compilation Instructions For C-Based Solver

After building the C-based solver please run the script solve.py as usual, and it will automatically use the newly built extension csolve.

In order to build the C-based solver the following command must be executed:

python setup.py build_ext --inplace

This will generate a file csolve.so (or csolve.pyd on Windows). Compiling Python C extensions on Windows usually involves additional steps. We strongly recommend using a 32-bit Python even on 64-bit Windows, as the steps are a lot easier.

• Compiling on 32-bit Python using MinGW:

  • Install MinGW

  • Compile using the --compiler flag:

    python setup.py build_ext --inplace --compiler=mingw32

• Compiling on 64-bit Python:

  • Install Visual Studio 10.0 Express

  • Install Microsoft Windows SDK for Windows 7 and .NET Framework 4

  • Create the following file:

    C:\Program Files (x86)\Microsoft Visual Studio 10.0\vc\bin\amd64\vcvars64.bat with the single-line content:

    CALL "C:\Program Files\Microsoft SDKs\Windows\v7.1\Bin\SetEnv.cmd" /x64

  • Compile as usual

    python setup.py build_ext --inplace

Authors

• Nimar S. Arora, Bayesian Logic Inc., nimar.arora@gmail.com
• Stuart Russell, Deptt. of Computer Science, Berkeley.