Author: bugman
Date: Wed Mar 6 11:11:05 2013
New Revision: 18657
URL: http://svn.gna.org/viewcvs/relax?rev=18657&view=rev
Log:
Removed the prune parameter from the backend of the
monte_carlo.error_analysis user function.
This was a dangerous parameter used to mimic the 'Trim' parameter from the
Modelfree4 program. The
result is bad statistics. The probable reason for the 'Trim' parameter was
the failure of
model-free models in the simulations, but this issue was solved using model
elimination (see
http://www.nmr-relax.com/refs.html#dAuvergneGooley06).
Modified:
trunk/generic_fns/monte_carlo.py
Modified: trunk/generic_fns/monte_carlo.py
URL:
http://svn.gna.org/viewcvs/relax/trunk/generic_fns/monte_carlo.py?rev=18657&r1=18656&r2=18657&view=diff
==============================================================================
--- trunk/generic_fns/monte_carlo.py (original)
+++ trunk/generic_fns/monte_carlo.py Wed Mar 6 11:11:05 2013
@@ -1,6 +1,6 @@
###############################################################################
#
#
-# Copyright (C) 2004-2005, 2007-2009 Edward d'Auvergne
#
+# Copyright (C) 2004-2013 Edward d'Auvergne
#
#
#
# This file is part of the program relax (http://www.nmr-relax.com).
#
#
#
@@ -111,7 +111,7 @@
pack_sim_data(data_index, random)
-def error_analysis(prune=0.0):
+def error_analysis():
"""Function for calculating errors from the Monte Carlo simulations.
The standard deviation formula used to calculate the errors is the
square root of the
@@ -126,12 +126,6 @@
- n is the total number of simulations.
- Xi is the parameter value for simulation i.
- Xav is the mean parameter value for all simulations.
-
-
- @keyword prune: The amount to prune the upper and lower tails the
distribution. If set to
- 0.0, no simulations will be pruned. If set to 1.0,
all simulations will be
- pruned. This argument should be set 0.0 to avoid
meaningless statistics.
- @type prune: float
"""
# Test if the current data pipe exists.
@@ -143,8 +137,6 @@
# Model loop, return simulation chi2 array, return selected simulation
array, return simulation parameter array, and set error functions.
model_loop = get_specific_fn('model_loop', cdp.pipe_type)
- if prune > 0.0:
- return_sim_chi2 = get_specific_fn('return_sim_chi2', cdp.pipe_type)
return_selected_sim = get_specific_fn('return_selected_sim',
cdp.pipe_type)
return_sim_param = get_specific_fn('return_sim_param', cdp.pipe_type)
set_error = get_specific_fn('set_error', cdp.pipe_type)
@@ -153,31 +145,6 @@
for model_info in model_loop():
# Get the selected simulation array.
select_sim = return_selected_sim(model_info)
-
- # Initialise an array of indices to prune (an empty array means no
pruning).
- indices_to_skip = []
-
- # Pruning.
- if prune > 0.0:
- # Get the array of simulation chi-squared values.
- chi2_array = return_sim_chi2(model_info)
-
- # The total number of simulations.
- n = len(chi2_array)
-
- # Create a sorted array of chi-squared values.
- chi2_sorted = sorted(deepcopy(chi2_array))
-
- # Number of indices to remove from one side of the chi2
distribution.
- num = int(float(n) * 0.5 * prune)
-
- # Remove the lower tail.
- for i in range(num):
- indices_to_skip.append(chi2_array.index(chi2_sorted[i]))
-
- # Remove the upper tail.
- for i in range(n-num, n):
- indices_to_skip.append(chi2_array.index(chi2_sorted[i]))
# Loop over the parameters.
index = 0
@@ -198,10 +165,6 @@
if not select_sim[i]:
continue
- # Prune.
- if i in indices_to_skip:
- continue
-
# Increment n.
n = n + 1
@@ -212,10 +175,6 @@
if not select_sim[i]:
continue
- # Prune.
- if i in indices_to_skip:
- continue
-
# Sum.
Xsum = Xsum + param_array[i]
@@ -230,10 +189,6 @@
for i in range(len(param_array)):
# Skip deselected simulations.
if not select_sim[i]:
- continue
-
- # Prune.
- if i in indices_to_skip:
continue
# Sum.
r18657 - /trunk/generic_fns/monte_carlo.py