I think the datasets (.txt) files have disappeared from the supplementary/Data folder

the link https://usgs.github.io/geobipy/ takes a person to a 404 page.

For users running this code on windows, there are multiple calls to np.float128. This datatype is not available for windows machines. This results in continuous exceptions and constant rejection of models.

Suggest changing to "np.longdouble". This will default to float128 on linux and float64 on windows, which is the best you can do.

Changes should be made to "utilities" line 670 and "inference1D" line 382, though there may be other instances.

I want to use the 'Agg' backend to make plots (such that I can create and save figures remotely, without access to an X-server), as I use a remote server to run GeoBiPy.

But, if I do

import matplotlib as mpl
mpl.use('Agg')
import matplotlib.pyplot as plt

from geobipy import LineResults

I get an error impoerting LineResults from geobipy:

(geobipy) tmeha@Arizona:~/$ python pl_results.py
Traceback (most recent call last):
  File "pl_results.py", line 15, in <module>
    from geobipy import LineResults
  File "/mnt/PROGRAMMING/geobipy/geobipy/__init__.py", line 20, in <module>
    from .src.base import customPlots
  File "/mnt/PROGRAMMING/geobipy/geobipy/src/base/customPlots.py", line 2, in <module>
    mpl.use('TkAgg')
  File "/home/tmeha/anaconda3/envs/geobipy/lib/python3.7/site-packages/matplotlib/__init__.py", line 1391, in use
    switch_backend(name)
  File "/home/tmeha/anaconda3/envs/geobipy/lib/python3.7/site-packages/matplotlib/pyplot.py", line 222, in switch_backend
    newbackend, required_framework, current_framework))
ImportError: Cannot load backend 'TkAgg' which requires the 'tk' interactive framework, as 'headless' is currently running

A simple trick to allow a user to select the backend is to comment out line 2 in /geobipy/src/base/customPlots.py:

# mpl.use('TkAgg')

• Thomas

there appears to be a problem with the plotDatapoints.py script when one or more soundings have failed to converge (burn in). In this case, you cannot plot any of the points and the following error shows up:

[pbedrosian@yeti-login20 plotting] python plotDatapoints.py ../Data/ ../results/1650.0.h5 --points 5
Traceback (most recent call last):
File "plotDatapoints.py", line 51, in
fids = LR.iDs[args.points]
IndexError: too many indices for array
[pbedrosian@yeti-login20 plotting] python plotDatapoints.py ../Data/ ../results/1650.0.h5
Traceback (most recent call last):
File "plotDatapoints.py", line 64, in
for fid in fids:
TypeError: iteration over a 0-d array

Hi there! I have installed the serial version of the code. When I followed the instruction to install the parallel version or TDEM modeller, it seems that the tutorial is not for Windows system.

How do you include the DOI as a transparency on top of a line plot? I tried using:

LR.plotDoi(linewidth=0.1, alpha=0.6, percent=40)

but got the following error-

File "plotLine.py", line 63, in
LR.plotDoi(linewidth=0.1, alpha=0.6, percent=40)
File "/cxfs/projects/software/geobipy/geobipy/src/inversion/LineResults.py", line 462, in plotDoi
self.getDOI(percent)
File "/cxfs/projects/software/geobipy/geobipy/src/inversion/LineResults.py", line 209, in getDOI
self.doi[i] = self.depthGrid.cellCentres[zSize - iC]
AttributeError: 'NoneType' object has no attribute 'cellCentres'

Hi there! In the docs, it says:

The code and its processes have been documented in multiple ways. First we have the publication associated with this software release, the citation is below, and presents the application of this package to frequency and time domain electro-magnetic inversion.

But there's no citation... would you all provide a link to the published work? Thanks!

Note: I found this abstract

I followed your instruction in the Documentation, the error messages are:

(base) C:\geobipy-master>setup.py install
Traceback (most recent call last):
File "C:\geobipy-master\setup.py", line 25, in
long_description=readme(),
File "C:\geobipy-master\setup.py", line 18, in readme
return f.read()
UnicodeDecodeError: 'gbk' codec can't decode byte 0x93 in position 870: illegal multibyte sequence

it would be great if gaps in line plots were shown as actual gaps rather than extrapolating the closest models (on either side of the gap) into the gap.

document somewhere allowed column header names. There is a list printed in the .out file when you get one wrong, but can we include them someplace else?

Hi

I run geobipySerial on a remote machine, and I do not use the X-server.
If I use

self.plot = False
self.savePNG = True

then geobipySerial stops with an error when the png file is being created:
"_tkinter.TclError: couldn't connect to display "localhost:12.0"
Is it possible to use another backend with matplotlib in this case (when self.plot=False) that does not need an Xserver?

The code runs fine if I use

self.plot = False
self.savePNG = False

Also, it would be nice if the progress of geobipySerial could be written to the console. Perhaps this is already an option?

Have a great day

• Thomas

add a parameter such as self.thin in the userParameters file that is used to thin the Markov Chain results so that only every nth model is included in output calculations & statistics. For example, with if the chain is run for 400,000 iterations with thin=4, the output statistics would be based on every 4th model (100,000 total). Accept/reject rules apply to all models in the chain as the algorithm runs.

This might help to deal with some 'noisy' outputs where the chain gets stuck occasionally, and will be a step towards future multi-chain runs

docs link receiving 404 error, https://usgs.github.io/geobipy/

Hi

When I import geobipy (fter a fresh install) I get the following:

(geobipy) tmeha@pop-os:~/space/PROGRAMMING/geobipy/documentation_source/source/examples/Models$ python
Python 3.8.1 (default, Jan  8 2020, 22:29:32) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import geobipy
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/mnt/space/PROGRAMMING/geobipy/geobipy/__init__.py", line 21, in <module>
    from .src.base import customFunctions
  File "/mnt/space/PROGRAMMING/geobipy/geobipy/src/base/customFunctions.py", line 5, in <module>
    from ..classes.core import StatArray
  File "/mnt/space/PROGRAMMING/geobipy/geobipy/src/classes/core/StatArray.py", line 12, in <module>
    from ...base.HDF import hdfRead
  File "/mnt/space/PROGRAMMING/geobipy/geobipy/src/base/HDF/hdfRead.py", line 6, in <module>
    from ...classes.data.datapoint.FdemDataPoint import FdemDataPoint
  File "/mnt/space/PROGRAMMING/geobipy/geobipy/src/classes/data/datapoint/FdemDataPoint.py", line 7, in <module>
    from .EmDataPoint import EmDataPoint
  File "/mnt/space/PROGRAMMING/geobipy/geobipy/src/classes/data/datapoint/EmDataPoint.py", line 1, in <module>
    from .DataPoint import DataPoint
  File "/mnt/space/PROGRAMMING/geobipy/geobipy/src/classes/data/datapoint/DataPoint.py", line 1, in <module>
    from cached_property import cached_property
ModuleNotFoundError: No module named 'cached_property'

Am I missing something ?

• Thomas

From CultPenguin.

Setting user defined bounds on the magnitude of the recovered parameter is not possible.
Add an additional uniform prior to the parameters.

I ran into issues when running pip install -e . for installing the serial version. The error seems to be related to mtforward1d_fortranmodule.c, which was not generated during build through f2py. I am not sure how to fix this. Any help will be much appreciated!

I am using python 3.7.1, below is the error message. Python 3.5/3.6 also produced the same error.

$ pip install -e .
Obtaining file:///Users/.../code/geobipy
    Complete output from command python setup.py egg_info:
    running egg_info
    running build_src
    build_src
    building extension "geobipy.src.classes.forwardmodelling.fdemforward1d_fortran" sources
    f2py options: []
      adding 'build/src.macosx-10.7-x86_64-3.7/build/src.macosx-10.7-x86_64-3.7/geobipy/src/classes/forwardmodelling/fortranobject.c' to sources.
      adding 'build/src.macosx-10.7-x86_64-3.7/build/src.macosx-10.7-x86_64-3.7/geobipy/src/classes/forwardmodelling' to include_dirs.
      adding 'build/src.macosx-10.7-x86_64-3.7/geobipy/src/classes/forwardmodelling/fdemforward1d_fortran-f2pywrappers2.f90' to sources.
    building extension "geobipy.src.classes.forwardmodelling.mtforward1d_fortran" sources
    f2py options: []
    f2py:> build/src.macosx-10.7-x86_64-3.7/geobipy/src/classes/forwardmodelling/mtforward1d_fortranmodule.c
    IOError: [Errno 2] No such file or directory: 'geobipy/src/classes/forwardmodelling/mtforward1D_fortran/m_mtforward1D.f90'. Skipping file "geobipy/src/classes/forwardmodelling/mtforward1D_fortran/m_mtforward1D.f90".
    error: f2py target file 'build/src.macosx-10.7-x86_64-3.7/geobipy/src/classes/forwardmodelling/mtforward1d_fortranmodule.c' not generated
    Reading fortran codes...
    Post-processing...
    Post-processing (stage 2)...
    Building modules...
   
    ----------------------------------------
Command "python setup.py egg_info" failed with error code 1 in /Users/.../code/geobipy/

Thanks!

The first time a HDF5 file is used to calculate opacity, store it as a variable such that it can simply be read in if present.

add to comments here what the units are for hard parameter limits (i.e. resistivity, conductivity, log/linear)

when using the plotDataPoints.py script, is there a way to select the sounding to be plotted by either its fiducial value or by the northing or easting value?

With self.reciprocateParameters = True or False, the plot is always conductivity in a geobipySerial run of fdem data

The hyperbolic tangent in Fortran is not defined in the same way on different systems. On my Mac, high positive values produce 1+0j, while on sci linux I got Nan+0j. Same version of gfortran.

I'm trying to plot some single-site results from geobipy. I can create the 'full' site plot using:

R = LR.getResults(LR.iDs[point])
R.plot(forcePlot=True)

Is there an easy way to plot just the component pieces of this figure (say just the histogram plot and the layer-depth historgram?

Is there an option to save all (or every N'th realizations) from the Markov chain?
I do not see these realizations on the hdf5 output file?

• Thomas

posterior plot for data elevation shows just a single value for all locations along a line.

Hi
Is there an example of howto run geobipy from within python, without the need to run it from the command line?

• Thomas

I need to update geobipy since Pandas removed their fast moving average window!

GeoBIPy works with 0.19.2