clberube / bisip Goto Github PK

Feature request:
Add a function to invResults.py that extracts the traces of each parameter and saves to a single csv file.

Code draft:

import numpy as np
def csv_traces(sol):
    # Saves the traces contained in mcmcinv object sol to a single csv file.
    trace_matrix = np.empty((len(sol.params), iterations-burnin))
    for i in range(len(sol.params)):
        trace_matrix[i] = sol.MDL.trace(sol.params[i])[:]
    headers = list(sol.params).join(',')
    np.savetxt('traceofparameters.csv', trace_matrix, delimiter=',', header=headers, comments="")

Dear Charles,
great work that can be of value for many SIP people! Thanks for sharing it.
I noticed that it only works for Python 2 (I guess only 2.7 as the pyd is linked against it). However, after long years of coexistence of Python 2 and Python 3, the days of Python 2 are over and all distributions are moving to Python 3 only as planned many years ago. So did we, not distributing any 2.7 stuff anymore.

Making the code compatible with both Python 2 and 3 should be easy using the future module (http://python-future.org) or 2to3 tools. I would appreciate it and I am sure it would increase users a lot.
Thomas

I have a .dat file that I want to inverse with BISIP.

When I try to do the inversion with a Pelton Cole-Cole model ( 2 modes) on the BISIP executable, it works fine.
But when I want to do it with the BISIP python shell, it doesn't fit anymore for the exact same model.

Any Idea on how to solve this problem?

Is it possible to have a function that calculates the misfit of the data once the model has be drawn?