core/incremental_mc.py

This example shows how to do incremental scoring with Monte Carlo. Incremental scoring can be significantly faster than non-incremental scoring when using moves that only move a few particles at a time.

## \example core/incremental_mc.py
# This example shows how to do incremental scoring with Monte Carlo.
# Incremental scoring can be significantly faster than non-incremental
# scoring when using moves that only move a few particles at a time.

import IMP.core
import IMP.container
import IMP.algebra
import IMP.display
import sys

IMP.base.setup_from_argv(sys.argv, "Optimize balls example")

if IMP.base.get_is_quick_test():
    num_balls = 2
    num_mc_steps = 10
else:
    num_balls = 20
    num_mc_steps = 1000

m = IMP.kernel.Model()
bb = IMP.algebra.BoundingBox3D(IMP.algebra.Vector3D(0, 0, 0),
                               IMP.algebra.Vector3D(30, 30, 30))
ps = []
for i in range(0, num_balls):
    ps.append(IMP.kernel.Particle(m))
    d = IMP.core.XYZR.setup_particle(ps[-1])
    d.set_radius(10)
    d.set_coordinates(IMP.algebra.get_random_vector_in(bb))
    d.set_coordinates_are_optimized(True)

psl = IMP.container.ListSingletonContainer(ps)
r = IMP.core.ExcludedVolumeRestraint(psl, 1)
rb = IMP.container.SingletonsRestraint(
    IMP.core.BoundingBox3DSingletonScore(
        IMP.core.HarmonicUpperBound(0, 1), bb),
    psl)
mc = IMP.core.MonteCarlo(m)
isf = IMP.core.IncrementalScoringFunction(ps, [r, rb])
mc.set_incremental_scoring_function(isf)

mvs = [IMP.core.BallMover([p], 5) for p in ps]
sm = IMP.core.SerialMover(mvs)
mc.add_mover(sm)
IMP.base.set_log_level(IMP.base.SILENT)
print "initial", isf.evaluate(False)
after = mc.optimize(num_mc_steps)
print "final", after
name = IMP.base.create_temporary_file_name("incremental_mc", ".pym")
w = IMP.display.PymolWriter(name)
for p in ps:
    g = IMP.core.XYZRGeometry(p)
    w.add_geometry(g)
print "pymol", name