Author: bugman
Date: Mon Sep 21 18:44:14 2009
New Revision: 9548
URL: http://svn.gna.org/viewcvs/relax?rev=9548&view=rev
Log:
Complete redesign of the grid_search() method.
The grid ops structure is no longer created, rather the grid is directly
created to be passed to the
grid search algorithm. The __grid_row() method has been added to create the
row of the grid
corresponding to a certain parameter.
Modified:
1.3/specific_fns/frame_order.py
Modified: 1.3/specific_fns/frame_order.py
URL:
http://svn.gna.org/viewcvs/relax/1.3/specific_fns/frame_order.py?rev=9548&r1=9547&r2=9548&view=diff
==============================================================================
--- 1.3/specific_fns/frame_order.py (original)
+++ 1.3/specific_fns/frame_order.py Mon Sep 21 18:44:14 2009
@@ -27,6 +27,7 @@
from copy import deepcopy
from math import pi
from minfx.generic import generic_minimise
+from minfx.grid import grid
from numpy import array, float64, ones, transpose, zeros
from re import search
from warnings import warn
@@ -49,6 +50,35 @@
class Frame_order(Common_functions):
"""Class containing the specific methods of the Frame Order theories."""
+ def __grid_row(self, incs, lower, upper, dist_type=None):
+ """Set up a row of the grid search for a given parameter.
+
+ @param incs: The number of increments.
+ @type incs: int
+ @param lower: The lower bounds.
+ @type lower: float
+ @param upper: The upper bounds.
+ @type upper: float
+ @keyword dist_type: The spacing or distribution type between grid
nodes. If None, then a
+ linear grid row is returned. If 'acos', then an
inverse cos
+ distribution of points is returned (e.g. for
uniform sampling in angular space).
+ @type dist_type: None or str
+ @return: The row of the grid.
+ @rtype: list of float
+ """
+
+ # Init.
+ row = []
+
+ # Loop over the increments.
+ for i in range(incs):
+ # The row.
+ row.append(lower + i * (upper - lower) / (incs - 1.0))
+
+ # Return the row.
+ return row
+
+
def __minimise_setup_tensors(self, sim_index=None):
"""Set up the data structures for optimisation using alignment
tensors as base data sets.
@@ -389,7 +419,7 @@
if not hasattr(cdp, 'pivot'):
raise RelaxError("The pivot point for the cone motion has not
been set.")
- # The cone axis.
+ # The cone axis.
cone_axis = zeros(3, float64)
generate_vector(cone_axis, cdp.theta_axis, cdp.phi_axis)
print(("Cone axis: %s." % cone_axis))
@@ -566,25 +596,27 @@
return names
- def grid_search(self, lower, upper, inc, constraints=False, verbosity=0,
sim_index=None):
+ def grid_search(self, lower=None, upper=None, inc=None,
constraints=False, verbosity=0, sim_index=None):
"""Perform a grid search.
- @param lower: The lower bounds of the grid search which must
be equal to the number of
- parameters in the model.
- @type lower: array of numbers
- @param upper: The upper bounds of the grid search which must
be equal to the number of
- parameters in the model.
- @type upper: array of numbers
- @param inc: The increments for each dimension of the space
for the grid search. The
- number of elements in the array must equal to
the number of parameters
- in the model.
- @type inc: int or array of int
- @param constraints: If True, constraints are applied during the grid
search (eliminating
- parts of the grid). If False, no constraints
are used.
- @type constraints: bool
- @param verbosity: A flag specifying the amount of information to
print. The higher the
- value, the greater the verbosity.
- @type verbosity: int
+ @keyword lower: The lower bounds of the grid search which
must be equal to the
+ number of parameters in the model.
+ @type lower: list of float
+ @keyword upper: The upper bounds of the grid search which
must be equal to the
+ number of parameters in the model.
+ @type upper: list of float
+ @keyword inc: The increments for each dimension of the
space for the grid search.
+ The number of elements in the array must
equal to the number of
+ parameters in the model.
+ @type inc: int or list of int
+ @keyword constraints: If True, constraints are applied during the
grid search (eliminating
+ parts of the grid). If False, no
constraints are used.
+ @type constraints: bool
+ @keyword verbosity: A flag specifying the amount of information
to print. The higher
+ the value, the greater the verbosity.
+ @type verbosity: int
+ @keyword sim_index: The Monte Carlo simulation index.
+ @type sim_index: None or int
"""
# Test if the Frame Order model has been set up.
@@ -596,46 +628,89 @@
# If inc is an int, convert it into an array of that value.
if isinstance(inc, int):
- inc = [inc]*n
-
- # Initialise the grid_ops structure.
- grid_ops = []
+ incs = [inc]*n
+ else:
+ incs = inc
+
+ # Initialise the grid increments structures.
+ default_bounds = False
+ if not lower:
+ lower = []
+ default_bounds = True
+ if not upper:
+ upper = []
+ grid = []
"""This structure is a list of lists. The first dimension
corresponds to the model
- parameter. The second dimension has the elements: 0, the number of
increments in that
- dimension; 1, the lower limit of the grid; 2, the upper limit of the
grid."""
-
- # Set the grid search options.
- for i in xrange(n):
- # Euler angles.
+ parameter. The second dimension are the grid node positions."""
+
+ # Generate the grid.
+ for i in range(n):
+ # Reset the distribution type.
+ dist_type = None
+
+ # Alpha Euler angle.
if cdp.params[i] == 'alpha':
- grid_ops.append([inc[i], 0.0, 2*pi * (1.0 - 1.0/inc[i])])
+ # Set the default bounds.
+ if default_bounds:
+ lower.append(0.0)
+ upper.append(2*pi * (1.0 - 1.0/incs[i]))
+
+ # Beta Euler angle.
if cdp.params[i] == 'beta':
- grid_ops.append([inc[i]/2, 0.0, pi * (1.0 - 1.0/inc[i])])
+ # Change the default increment numbers.
+ if not isinstance(inc, list):
+ incs[i] = incs[i] / 2
+
+ # The distribution type.
+ dist_type = 'acos'
+
+ # Set the default bounds.
+ if default_bounds:
+ lower.append(0.0)
+ upper.append(pi * (1.0 - 1.0/incs[i]))
+
+ # Gamma Euler angle.
if cdp.params[i] == 'gamma':
- grid_ops.append([inc[i], 0.0, 2*pi * (1.0 - 1.0/inc[i])])
+ # Set the default bounds.
+ if default_bounds:
+ lower.append(0.0)
+ upper.append(2*pi * (1.0 - 1.0/incs[i]))
# The isotropic cone model.
if cdp.model == 'iso cone':
- # Cone axis angles and cone angle.
+ # Cone axis polar angle.
if cdp.params[i] == 'theta_axis':
- grid_ops.append([inc[i]/2, 0.0, pi * (1.0 - 1.0/inc[i])])
+ # Change the default increment numbers.
+ if not isinstance(inc, list):
+ incs[i] = incs[i] / 2 + 1
+
+ # The distribution type.
+ dist_type = 'acos'
+
+ # Set the default bounds.
+ if default_bounds:
+ lower.append(0.0)
+ upper.append(pi)
+
+ # Cone axis azimuthal angle.
if cdp.params[i] == 'phi_axis':
- grid_ops.append([inc[i], 0.0, 2*pi * (1.0 - 1.0/inc[i])])
+ # Set the default bounds.
+ if default_bounds:
+ lower.append(0.0)
+ upper.append(2*pi * (1.0 - 1.0/incs[i]))
# The cone angle.
if cdp.params[i] == 'theta_cone':
- grid_ops.append([inc[i], 0.0, pi * (1.0 - 1.0/inc[i])])
-
- # Lower bound (if supplied).
- if lower:
- grid_ops[i][1] = lower[i]
-
- # Upper bound (if supplied).
- if upper:
- grid_ops[i][1] = upper[i]
+ # Set the default bounds.
+ if default_bounds:
+ lower.append(0.0)
+ upper.append(pi * (1.0 - 1.0/incs[i]))
+
+ # Get the grid row.
+ grid.append(self.__grid_row(incs[i], lower[i], upper[i],
dist_type=dist_type))
# Minimisation.
- self.minimise(min_algor='grid', min_options=grid_ops,
constraints=constraints, verbosity=verbosity, sim_index=sim_index)
+ self.minimise(min_algor='grid', min_options=grid,
constraints=constraints, verbosity=verbosity, sim_index=sim_index)
def minimise(self, min_algor=None, min_options=None, func_tol=None,
grad_tol=None, max_iterations=None, constraints=False, scaling=True,
verbosity=0, sim_index=None):
@@ -693,8 +768,14 @@
# Set up the optimisation function.
target = frame_order.Frame_order(model=cdp.model,
full_tensors=full_tensors, red_tensors=red_tensors,
red_errors=red_tensor_err, full_in_ref_frame=full_in_ref_frame)
+ # Grid search.
+ if search('^[Gg]rid', min_algor):
+ results = grid(func=target.func, args=(),
num_incs=min_options[0], lower=min_options[1], upper=min_options[2],
verbosity=verbosity)
+ print results
+
# Minimisation.
- results = generic_minimise(func=target.func, args=(),
x0=param_vector, min_algor=min_algor, min_options=min_options,
func_tol=func_tol, grad_tol=grad_tol, maxiter=max_iterations,
full_output=True, print_flag=verbosity)
+ else:
+ results = generic_minimise(func=target.func, args=(),
x0=param_vector, min_algor=min_algor, min_options=min_options,
func_tol=func_tol, grad_tol=grad_tol, maxiter=max_iterations,
full_output=True, print_flag=verbosity)
# Unpack the results.
self.__unpack_opt_results(results, sim_index)
r9548 - /1.3/specific_fns/frame_order.py