From bcc82f335d006e1f85d8294dae377d7fab7902bd Mon Sep 17 00:00:00 2001 From: "John Chodera (MSKCC)" Date: Sun, 4 Oct 2015 23:09:08 -0400 Subject: [PATCH 1/6] Draft implementation of FIRE minimizer --- openmmtools/integrators.py | 151 +++++++++++++++++++++++++- openmmtools/tests/test_integrators.py | 22 +++- 2 files changed, 170 insertions(+), 3 deletions(-) diff --git a/openmmtools/integrators.py b/openmmtools/integrators.py index f19283f6d..5b376f974 100644 --- a/openmmtools/integrators.py +++ b/openmmtools/integrators.py @@ -41,7 +41,7 @@ import simtk.unit as units import simtk.openmm as mm -from .constants import kB +from openmmtools.constants import kB #============================================================================================= # INTEGRATORS @@ -184,6 +184,154 @@ def __init__(self, initial_step_size=0.01 * units.angstroms): # Update step size. self.addComputeGlobal("step_size", "step_size * (2.0*accept + 0.5*(1-accept))") +class FIREMinimizationIntegrator(mm.CustomIntegrator): + """Fast Internal Relaxation Engine (FIRE) minimization. + + Notes + ----- + This integrator is taken verbatim from Peter Eastman's example appearing in the CustomIntegrator header file documentation. + + References + ---------- + + Erik Bitzek, Pekka Koskinen, Franz Gaehler, Michael Moseler, and Peter Gumbsch. + Structural Relaxation Made Simple. PRL 97:170201, 2006. + http://dx.doi.org/10.1103/PhysRevLett.97.170201 + + Examples + -------- + + Create a FIRE integrator with default parameters. + + >>> integrator = FIREMinimizationIntegrator() + + """ + + def __init__(self, timestep=1.0 * simtk.unit.femtoseconds, tolerance=None, alpha=0.1, dt_max=10.0 * simtk.unit.femtoseconds, f_inc=1.1, f_dec=0.5, f_alpha=0.99, N_min=5): + """Construct a Fast Internal Relaxation Engine (FIRE) minimization integrator. + + Parameters + ---------- + timestep : simtk.unit.Quantity compatible with femtoseconds, optional, default = 1*femtoseconds + The integration timestep. + tolerance : simtk.unit.Quantity compatible with kilojoules_per_mole/nanometer, optional, default = None + Minimization will be terminated when RMS force reaches this tolerance. + alpha : float, optional default = 0.1 + Velocity relaxation parameter, alpha \in (0,1). + dt_max : simtk.unit.Quantity compatible with femtoseconds, optional, default = 10*femtoseconds + Maximum allowed timestep. + f_inc : float, optional, default = 1.1 + Timestep increment multiplicative factor. + f_dec : float, optional, default = 0.5 + Timestep decrement multiplicative factor. + f_alpha : float, optional, default = 0.99 + alpha multiplicative relaxation parameter + N_min : int, optional, default = 5 + Limit on number of timesteps P is negative before decrementing timestep. + + Notes + ----- + Velocities should be set to zero before using this integrator. + + """ + + # Check input ranges. + if not ((alpha > 0.0) and (alpha < 1.0)): + raise Exception("alpha must be in the interval (0,1); specified alpha = %f" % alpha) + + if tolerance is None: + tolerance = 0 * units.kilojoules_per_mole / units.nanometers + + super(FIREMinimizationIntegrator, self).__init__(timestep) + + self.addGlobalVariable("alpha", alpha) # alpha + self.addGlobalVariable("P", 0) # P + self.addGlobalVariable("N_neg", 0.0) + self.addGlobalVariable("fmag", 0) # |f| + self.addGlobalVariable("fmax", 0) # max|f_i| + self.addGlobalVariable("ndof", 0) # number of degrees of freedom + self.addGlobalVariable("ftol", tolerance.value_in_unit_system(units.md_unit_system)) # convergence tolerance + self.addGlobalVariable("freg", 0.001) # regularization to add to |f| + self.addGlobalVariable("vmag", 0) # |v| + self.addGlobalVariable("converged", 0) # 1 if convergence threshold reached, 0 otherwise + self.addPerDofVariable("x0", 0) + self.addPerDofVariable("x1", 0) + + # Enclose everything in a block that checks if we have already converged. + self.beginIfBlock('converged < 1') + + # Update Context. + # TODO: Remove this? + self.addUpdateContextState() + + # Store old positions + self.addComputePerDof('x0', 'x') + + # MD: Take a velocity Verlet step. + self.addComputePerDof("v", "v+0.5*dt*f/m") + self.addComputePerDof("x", "x+dt*v") + self.addComputePerDof("x1", "x") + self.addConstrainPositions() + self.addComputePerDof("v", "v+0.5*dt*f/m+(x-x1)/dt") + self.addConstrainVelocities() + + # Compute fmag = |f| + self.addComputeGlobal('fmag', '0.0') + self.addComputeSum('fmag', 'f*f') + self.addComputeGlobal('fmag', 'sqrt(fmag)') + # Compute vmag = |v| + self.addComputeGlobal('vmag', '0.0') + self.addComputeSum('vmag', 'v*v') + self.addComputeGlobal('vmag', 'sqrt(vmag)') + + # Check for convergence of force magnitude. + # TODO: Check componentwise or RMS? + self.addComputeSum('ndof', '1') + #self.addComputeSum('converged', 'step(ftol - abs(f)) / ndof') + self.beginIfBlock('fmag/sqrt(ndof) < ftol') + self.addComputeGlobal('converged', '1') # signal convergence + self.endBlock() + + ## Check for NaN and try to recover. + #self.beginIfBlock('abs(step(energy)-step(1-energy)) = 0') + ## Reset positions and set velocities to zero. + #self.addComputePerDof('x', 'x0') + #self.addComputePerDof('v', '0.0') + ## Reset P counter. + #self.addComputeGlobal('N_neg', '0.0') + ## Scale down timestep. + #self.addComputeGlobal('dt', 'dt*%f' % f_dec) + #self.endBlock() + + # F1: Compute P = F.v + self.addComputeSum('P', 'f*v') + + # F2: set v = (1-alpha) v + alpha \hat{F}.|v| + # Update velocities. + self.addComputePerDof('v', '(1-alpha)*v + alpha*(f/(fmag+freg))*vmag') + + # F3: If P > 0 and the number of steps since P was negative > N_min, + # Increase timestep dt = min(dt*f_inc, dt_max) and decrease alpha = alpha*f_alpha + self.beginIfBlock('P > 0') + # Update count of number of steps since P was negative. + self.addComputeGlobal('N_neg', 'step(P) * (N_neg + 1)') + # If we have enough steps since P was negative, scale up timestep. + self.beginIfBlock('N_neg > %d' % N_min) + self.addComputeGlobal('dt', 'min(dt*%f, %f)' % (f_inc, dt_max.value_in_unit_system(units.md_unit_system))) # TODO: Automatically convert dt_max to md units + self.addComputeGlobal('alpha', 'alpha * %f' % f_alpha) + self.endBlock() + self.endBlock() + + # F4: If P < 0, decrease the timestep dt = dt*f_dec, freeze the system v=0, + # and set alpha = alpha_start + self.beginIfBlock('P < 0') + self.addComputeGlobal('N_neg', '0.0') + self.addComputeGlobal('dt', 'dt*%f' % f_dec) + self.addComputePerDof('v', '0.0') + self.endBlock() + + # Close block that checks for convergence. + self.endBlock() class VelocityVerletIntegrator(mm.CustomIntegrator): @@ -229,7 +377,6 @@ def __init__(self, timestep=1.0 * simtk.unit.femtoseconds): self.addComputePerDof("v", "v+0.5*dt*f/m+(x-x1)/dt") self.addConstrainVelocities() - class AndersenVelocityVerletIntegrator(mm.CustomIntegrator): """Velocity Verlet integrator with Andersen thermostat using per-particle collisions (rather than massive collisions). diff --git a/openmmtools/tests/test_integrators.py b/openmmtools/tests/test_integrators.py index d7bb125e6..dd50da87b 100644 --- a/openmmtools/tests/test_integrators.py +++ b/openmmtools/tests/test_integrators.py @@ -101,7 +101,7 @@ def test_integrator_decorators(): integrator = integrators.HMCIntegrator(timestep=0.05 * unit.femtoseconds) testsystem = testsystems.IdealGas() nsteps = 25 - + context = openmm.Context(testsystem.system, integrator) context.setPositions(testsystem.positions) context.setVelocitiesToTemperature(300 * unit.kelvin) @@ -111,3 +111,23 @@ def test_integrator_decorators(): assert integrator.n_accept == nsteps assert integrator.n_trials == nsteps assert integrator.acceptance_rate == 1.0 + +def test_fire_minimization(): + integrator = integrators.FIREMinimizationIntegrator() + testsystem = testsystems.LennardJonesFluid() + nsteps = 100 + + context = openmm.Context(testsystem.system, integrator) + context.setPositions(testsystem.positions) + context.setVelocitiesToTemperature(300 * unit.kelvin) + + initial_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole + + integrator.step(nsteps) + + final_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole + + print "FIRE minimizer" + print "Initial energy: %12.3f kcal/mol" % initial_energy + print "Final energy: %12.3f kcal/mol" % final_energy + From 5bda1749813533768e8fa37794041007b1114890 Mon Sep 17 00:00:00 2001 From: "John Chodera (MSKCC)" Date: Sun, 4 Oct 2015 23:13:41 -0400 Subject: [PATCH 2/6] Added test script for FIRE minimizer --- openmmtools/scripts/test_fire_minimizer.py | 445 +++++++++++++++++++++ 1 file changed, 445 insertions(+) create mode 100644 openmmtools/scripts/test_fire_minimizer.py diff --git a/openmmtools/scripts/test_fire_minimizer.py b/openmmtools/scripts/test_fire_minimizer.py new file mode 100644 index 000000000..e64bcb74f --- /dev/null +++ b/openmmtools/scripts/test_fire_minimizer.py @@ -0,0 +1,445 @@ +#!/usr/local/bin/env python + +#============================================================================================= +# MODULE DOCSTRING +#============================================================================================= + +""" +Test the FIRE minimizer against the OpenMM minmizer. + +DESCRIPTION + +COPYRIGHT + +@author John D. Chodera + +All code in this repository is released under the GNU General Public License. + +This program is free software: you can redistribute it and/or modify it under +the terms of the GNU General Public License as published by the Free Software +Foundation, either version 3 of the License, or (at your option) any later +version. + +This program is distributed in the hope that it will be useful, but WITHOUT ANY +WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A +PARTICULAR PURPOSE. See the GNU General Public License for more details. + +You should have received a copy of the GNU General Public License along with +this program. If not, see . + +TODO + +""" + +#============================================================================================= +# PYTHON 3 COMPATIBILITY CRAP +#============================================================================================= + +from __future__ import print_function + +#============================================================================================= +# ENABLE LOGGING +#============================================================================================= + +import logging +logger = logging.getLogger(__name__) + +def config_root_logger(verbose, log_file_path=None, mpicomm=None): + """Setup the the root logger's configuration. + The log messages are printed in the terminal and saved in the file specified + by log_file_path (if not None) and printed. Note that logging use sys.stdout + to print logging.INFO messages, and stderr for the others. The root logger's + configuration is inherited by the loggers created by logging.getLogger(name). + Different formats are used to display messages on the terminal and on the log + file. For example, in the log file every entry has a timestamp which does not + appear in the terminal. Moreover, the log file always shows the module that + generate the message, while in the terminal this happens only for messages + of level WARNING and higher. + Parameters + ---------- + verbose : bool + Control the verbosity of the messages printed in the terminal. The logger + displays messages of level logging.INFO and higher when verbose=False. + Otherwise those of level logging.DEBUG and higher are printed. + log_file_path : str, optional, default = None + If not None, this is the path where all the logger's messages of level + logging.DEBUG or higher are saved. + mpicomm : mpi4py.MPI.COMM communicator, optional, default=None + If specified, this communicator will be used to determine node rank. + + """ + + class TerminalFormatter(logging.Formatter): + """ + Simplified format for INFO and DEBUG level log messages. + This allows to keep the logging.info() and debug() format separated from + the other levels where more information may be needed. For example, for + warning and error messages it is convenient to know also the module that + generates them. + """ + + # This is the cleanest way I found to make the code compatible with both + # Python 2 and Python 3 + simple_fmt = logging.Formatter('%(message)s') + default_fmt = logging.Formatter('%(levelname)s - %(name)s - %(message)s') + + def format(self, record): + if record.levelno <= logging.INFO: + return self.simple_fmt.format(record) + else: + return self.default_fmt.format(record) + + # Check if root logger is already configured + n_handlers = len(logging.root.handlers) + if n_handlers > 0: + root_logger = logging.root + for i in xrange(n_handlers): + root_logger.removeHandler(root_logger.handlers[0]) + + # If this is a worker node, don't save any log file + if mpicomm: + rank = mpicomm.rank + else: + rank = 0 + + if rank != 0: + log_file_path = None + + # Add handler for stdout and stderr messages + terminal_handler = logging.StreamHandler() + terminal_handler.setFormatter(TerminalFormatter()) + if rank != 0: + terminal_handler.setLevel(logging.WARNING) + elif verbose: + terminal_handler.setLevel(logging.DEBUG) + else: + terminal_handler.setLevel(logging.INFO) + logging.root.addHandler(terminal_handler) + + # Add file handler to root logger + if log_file_path is not None: + #file_format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s' + file_format = '%(asctime)s: %(message)s' + file_handler = logging.FileHandler(log_file_path) + file_handler.setLevel(logging.DEBUG) + file_handler.setFormatter(logging.Formatter(file_format)) + logging.root.addHandler(file_handler) + + # Do not handle logging.DEBUG at all if unnecessary + if log_file_path is not None: + logging.root.setLevel(logging.DEBUG) + else: + logging.root.setLevel(terminal_handler.level) + +#============================================================================================= +# GLOBAL IMPORTS +#============================================================================================= + +import os +import os.path +import sys +import math + +from simtk import openmm, unit + +from openmmtools import testsystems + +#============================================================================================= +# SUBROUTINES +#============================================================================================= + +# These settings control what tolerance is allowed between platforms and the Reference platform. +ENERGY_TOLERANCE = 0.06*unit.kilocalories_per_mole # energy difference tolerance +FORCE_RMSE_TOLERANCE = 0.06*unit.kilocalories_per_mole/unit.angstrom # per-particle force root-mean-square error tolerance + +def assert_approximately_equal(computed_potential, expected_potential, tolerance=ENERGY_TOLERANCE): + """ + Check whether computed potential is acceptably close to expected value, using an error tolerance. + + ARGUMENTS + + computed_potential (simtk.unit.Quantity in units of energy) - computed potential energy + expected_potential (simtk.unit.Quantity in units of energy) - expected + + OPTIONAL ARGUMENTS + + tolerance (simtk.unit.Quantity in units of energy) - acceptable tolerance + + EXAMPLES + + >>> assert_approximately_equal(0.0000 * unit.kilocalories_per_mole, 0.0001 * unit.kilocalories_per_mole, tolerance=0.06*unit.kilocalories_per_mole) + + """ + + # Compute error. + error = (computed_potential - expected_potential) + + # Raise an exception if the error is larger than the tolerance. + if abs(error) > tolerance: + raise Exception("Computed potential %s, expected %s. Error %s is larger than acceptable tolerance of %s." % (computed_potential, expected_potential, error, tolerance)) + + return + +def compute_potential_and_force(system, positions, platform): + """ + Compute the energy and force for the given system and positions in the designated platform. + + ARGUMENTS + + system (simtk.openmm.System) - the system for which the energy is to be computed + positions (simtk.unit.Quantity of Nx3 numpy.array in units of distance) - positions for which energy and force are to be computed + platform (simtk.openmm.Platform) - platform object to be used to compute the energy and force + + RETURNS + + potential (simtk.unit.Quantity in energy/mole) - the potential + force (simtk.unit.Quantity of Nx3 numpy.array in units of energy/mole/distance) - the force + + """ + + # Create a Context. + kB = unit.BOLTZMANN_CONSTANT_kB + temperature = 298.0 * unit.kelvin + kT = kB * temperature + beta = 1.0 / kT + collision_rate = 90.0 / unit.picosecond + timestep = 1.0 * unit.femtosecond + integrator = openmm.LangevinIntegrator(temperature, collision_rate, timestep) + context = openmm.Context(system, integrator, platform) + # Set positions + context.setPositions(positions) + # Evaluate the potential energy. + state = context.getState(getEnergy=True, getForces=True) + potential = state.getPotentialEnergy() + force = state.getForces(asNumpy=True) + + return [potential, force] + +def compute_potential_and_force_by_force_index(system, positions, platform, force_index): + """ + Compute the energy and force for the given system and positions in the designated platform for the given force index. + + ARGUMENTS + + system (simtk.openmm.System) - the system for which the energy is to be computed + positions (simtk.unit.Quantity of Nx3 numpy.array in units of distance) - positions for which energy and force are to be computed + platform (simtk.openmm.Platform) - platform object to be used to compute the energy and force + force_index (int) - index of force to be computed (all others ignored) + + RETURNS + + potential (simtk.unit.Quantity in energy/mole) - the potential + force (simtk.unit.Quantity of Nx3 numpy.array in units of energy/mole/distance) - the force + + """ + + forces = [ system.getForce(index) for index in range(system.getNumForces()) ] + + # Get original force groups. + groups = [ force.getForceGroup() for force in forces ] + + # Set force groups so only specified force_index contributes. + for force in forces: + force.setForceGroup(1) + forces[force_index].setForceGroup(0) # bitmask of 1 should select only desired force + + # Create a Context. + kB = unit.BOLTZMANN_CONSTANT_kB + temperature = 298.0 * unit.kelvin + kT = kB * temperature + beta = 1.0 / kT + collision_rate = 90.0 / unit.picosecond + timestep = 1.0 * unit.femtosecond + integrator = openmm.LangevinIntegrator(temperature, collision_rate, timestep) + context = openmm.Context(system, integrator, platform) + # Set positions + context.setPositions(positions) + # Evaluate the potential energy. + state = context.getState(getEnergy=True, getForces=True, groups=1) + potential = state.getPotentialEnergy() + force = state.getForces(asNumpy=True) + + # Restore original force groups. + for index in range(system.getNumForces()): + forces[index].setForceGroup(groups[index]) + + return [potential, force] + +def compute_potential_and_force_by_force_group(system, positions, platform, force_group): + """ + Compute the energy and force for the given system and positions in the designated platform for the given force group. + + ARGUMENTS + + system (simtk.openmm.System) - the system for which the energy is to be computed + positions (simtk.unit.Quantity of Nx3 numpy.array in units of distance) - positions for which energy and force are to be computed + platform (simtk.openmm.Platform) - platform object to be used to compute the energy and force + force_group (int) - index of force group to be computed (all others ignored) + + RETURNS + + potential (simtk.unit.Quantity in energy/mole) - the potential + force (simtk.unit.Quantity of Nx3 numpy.array in units of energy/mole/distance) - the force + + """ + + forces = [ system.getForce(index) for index in range(system.getNumForces()) ] + + # Create a Context. + kB = unit.BOLTZMANN_CONSTANT_kB + temperature = 298.0 * unit.kelvin + kT = kB * temperature + beta = 1.0 / kT + collision_rate = 90.0 / unit.picosecond + timestep = 1.0 * unit.femtosecond + integrator = openmm.LangevinIntegrator(temperature, collision_rate, timestep) + context = openmm.Context(system, integrator, platform) + # Set positions + context.setPositions(positions) + # Evaluate the potential energy. + groupmask = 1 << (force_group + 1) + state = context.getState(getEnergy=True, getForces=True, groups=groupmask) + potential = state.getPotentialEnergy() + force = state.getForces(asNumpy=True) + + return [potential, force] + +def get_all_subclasses(cls): + """ + Return all subclasses of a specified class. + + Parameters + ---------- + cls : class + The class for which all subclasses are to be returned. + + Returns + ------- + all_subclasses : list of class + List of all subclasses of `cls`. + + """ + + all_subclasses = [] + + for subclass in cls.__subclasses__(): + all_subclasses.append(subclass) + all_subclasses.extend(get_all_subclasses(subclass)) + + return all_subclasses + +#============================================================================================= +# MAIN AND TESTS +#============================================================================================= + +def main(): + import doctest + import argparse + + parser = argparse.ArgumentParser(description="Benchmark the FIRE minimizer against the OpenMM minimizer for a suite of test systems.") + parser.add_argument('-o', '--outfile', dest='logfile', action='store', type=str, default=None) + parser.add_argument('-v', dest='verbose', action='store_true') + parser.add_argument('-p', '--platform', dest='platformname', action='store', type=str, default=None) + args = parser.parse_args() + + verbose = args.verbose # Don't display extra debug information. + config_root_logger(verbose, log_file_path=args.logfile) + + # Print version. + logger.info("OpenMM version: %s" % openmm.version.version) + logger.info("") + + # List all available platforms + logger.info("Available platforms:") + for platform_index in range(openmm.Platform.getNumPlatforms()): + platform = openmm.Platform.getPlatform(platform_index) + logger.info("%5d %s" % (platform_index, platform.getName())) + logger.info("") + + # Select specified platform. + platform = None + if args.platformname: + logger.info("Using platform '%s'..." % args.platformname) + platform = openmm.Platform.getPlatformByName(args.platformname) + + # Test performance on a variety of testsystems. + tests_failed = 0 # number of times tolerance is exceeded + tests_passed = 0 # number of times tolerance is not exceeded + #logger.info("%16s%16s %16s %16s %16s %16s" % ("platform", "precision", "potential", "error", "force mag", "rms error")) + testsystem_classes = get_all_subclasses(testsystems.TestSystem) + for testsystem_class in testsystem_classes: + class_name = testsystem_class.__name__ + logger.info(class_name) + + try: + testsystem = testsystem_class() + except ImportError as e: + logger.info(e) + logger.info("Skipping %s due to missing dependency" % class_name) + continue + + # Create test system instance. + [system, initial_positions] = [testsystem.system, testsystem.positions] + + logger.info("%s (%d atoms)" % (class_name, testsystem.system.getNumParticles())) + + maxits = 500 + forcetol = 1.0 * unit.kilojoules_per_mole / unit.angstrom + + # Create integrator. + from openmmtools import integrators + integrator = integrators.FIREMinimizationIntegrator(tolerance=forcetol) + + # Build list of parameters. + global_variables = { integrator.getGlobalVariableName(index) : index for index in range(integrator.getNumGlobalVariables()) } + + # Create context. + if platform: + context = openmm.Context(system, integrator, platform) + else: + context = openmm.Context(system, integrator) + + # Minimize with FIRE minimizer + import time + context.setPositions(initial_positions) + fire_initial_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole + initial_time = time.time() + integrator.step(maxits) + final_time = time.time() + elapsed_time = final_time - initial_time + fire_final_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole + logger.info("FIRE %8.3f s" % elapsed_time) + logger.info("initial energy = %12.3f kcal/mol" % fire_initial_energy) + logger.info("final energy = %12.3f kcal/mol" % fire_final_energy) + logger.info("converged = %f" % integrator.getGlobalVariable(global_variables['converged'])) + + # Time LocalEnergyMinimizer + from simtk.openmm import LocalEnergyMinimizer + context.setPositions(initial_positions) + fire_initial_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole + initial_time = time.time() + LocalEnergyMinimizer.minimize(context, forcetol, maxits) + final_time = time.time() + elapsed_time = final_time - initial_time + fire_final_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole + logger.info("LocalEnergyMinimizer %8.3f s" % elapsed_time) + logger.info("initial energy = %12.3f kcal/mol" % fire_initial_energy) + logger.info("final energy = %12.3f kcal/mol" % fire_final_energy) + + + # Clean up + logger.info("Cleaning up.") + del context + logger.info("") + + logger.info("%d tests failed" % tests_failed) + logger.info("%d tests passed" % tests_passed) + + if (tests_failed > 0): + # Signal failure of test. + sys.exit(1) + else: + sys.exit(0) + +if __name__ == "__main__": + main() From 92e5e1b39ae31564f2b5ed1dea531b50c940d4bb Mon Sep 17 00:00:00 2001 From: John Chodera Date: Wed, 14 Oct 2015 17:28:02 -0400 Subject: [PATCH 3/6] Updates. --- openmmtools/scripts/test_fire_minimizer.py | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/openmmtools/scripts/test_fire_minimizer.py b/openmmtools/scripts/test_fire_minimizer.py index e64bcb74f..a535f2e6e 100644 --- a/openmmtools/scripts/test_fire_minimizer.py +++ b/openmmtools/scripts/test_fire_minimizer.py @@ -367,6 +367,7 @@ def main(): tests_passed = 0 # number of times tolerance is not exceeded #logger.info("%16s%16s %16s %16s %16s %16s" % ("platform", "precision", "potential", "error", "force mag", "rms error")) testsystem_classes = get_all_subclasses(testsystems.TestSystem) + testsystem_classes = [getattr(testsystems, name) for name in ('LysozymeImplicit', 'AMOEBAProteinBox', 'SrcImplicit', 'SrcExplicit')] for testsystem_class in testsystem_classes: class_name = testsystem_class.__name__ logger.info(class_name) @@ -383,15 +384,17 @@ def main(): logger.info("%s (%d atoms)" % (class_name, testsystem.system.getNumParticles())) - maxits = 500 + maxits = 100 forcetol = 1.0 * unit.kilojoules_per_mole / unit.angstrom # Create integrator. from openmmtools import integrators + logger.info('Creating FIRE integrator...') integrator = integrators.FIREMinimizationIntegrator(tolerance=forcetol) + logger.info('Done.') # Build list of parameters. - global_variables = { integrator.getGlobalVariableName(index) : index for index in range(integrator.getNumGlobalVariables()) } + #global_variables = { integrator.getGlobalVariableName(index) : index for index in range(integrator.getNumGlobalVariables()) } # Create context. if platform: @@ -411,7 +414,7 @@ def main(): logger.info("FIRE %8.3f s" % elapsed_time) logger.info("initial energy = %12.3f kcal/mol" % fire_initial_energy) logger.info("final energy = %12.3f kcal/mol" % fire_final_energy) - logger.info("converged = %f" % integrator.getGlobalVariable(global_variables['converged'])) + #logger.info("converged = %f" % integrator.getGlobalVariable(global_variables['converged'])) # Time LocalEnergyMinimizer from simtk.openmm import LocalEnergyMinimizer From 0fabb74ab9a79f0a56780ede02e0f94e1f80760d Mon Sep 17 00:00:00 2001 From: "John Chodera (MSKCC)" Date: Sat, 17 Oct 2015 18:52:28 +0200 Subject: [PATCH 4/6] Edits. --- openmmtools/scripts/test_fire_minimizer.py | 3 ++ openmmtools/sobol.py | 33 ++++++++++++---------- openmmtools/testsystems.py | 8 +++++- 3 files changed, 28 insertions(+), 16 deletions(-) diff --git a/openmmtools/scripts/test_fire_minimizer.py b/openmmtools/scripts/test_fire_minimizer.py index e64bcb74f..94c6cb027 100644 --- a/openmmtools/scripts/test_fire_minimizer.py +++ b/openmmtools/scripts/test_fire_minimizer.py @@ -362,11 +362,14 @@ def main(): logger.info("Using platform '%s'..." % args.platformname) platform = openmm.Platform.getPlatformByName(args.platformname) + # TODO: Try minimizing pdbfixer-retrieved systems + # Test performance on a variety of testsystems. tests_failed = 0 # number of times tolerance is exceeded tests_passed = 0 # number of times tolerance is not exceeded #logger.info("%16s%16s %16s %16s %16s %16s" % ("platform", "precision", "potential", "error", "force mag", "rms error")) testsystem_classes = get_all_subclasses(testsystems.TestSystem) + testsystem_classes = [ getattr(testsystems, name) for name in ('SrcImplicit', 'SrcExplicit') ] for testsystem_class in testsystem_classes: class_name = testsystem_class.__name__ logger.info(class_name) diff --git a/openmmtools/sobol.py b/openmmtools/sobol.py index 823583a97..fb33dbac8 100644 --- a/openmmtools/sobol.py +++ b/openmmtools/sobol.py @@ -236,21 +236,24 @@ def i4_sobol ( dim_num, seed ): # # Output, real QUASI(DIM_NUM), the next quasirandom vector. # - global atmost - global dim_max - global dim_num_save - global initialized - global lastq - global log_max - global maxcol - global poly - global recipd - global seed_save - global v - - if ( not 'initialized' in globals().keys() ): - initialized = 0 - dim_num_save = -1 +# global atmost +# global dim_max +# global dim_num_save +# global initialized +# global lastq +# global log_max +# global maxcol +# global poly +# global recipd +# global seed_save +# global v + +# if ( not 'initialized' in globals().keys() ): +# initialized = 0 +# dim_num_save = -1 + + initialized = 0 + dim_num_save = -1 if ( not initialized or dim_num != dim_num_save ): initialized = 1 diff --git a/openmmtools/testsystems.py b/openmmtools/testsystems.py index b2e7d13c3..79e6841f6 100644 --- a/openmmtools/testsystems.py +++ b/openmmtools/testsystems.py @@ -155,7 +155,7 @@ def halton_sequence(p, n): return u -def subrandom_particle_positions(nparticles, box_vectors, method='sobol'): +def subrandom_particle_positions(nparticles, box_vectors, method='random'): """Generate a deterministic list of subrandom particle positions. Parameters @@ -205,6 +205,12 @@ def subrandom_particle_positions(nparticles, box_vectors, method='sobol'): l = box_vectors[dim][dim] positions[:, dim] = unit.Quantity(x[dim, :] * l / l.unit, l.unit) + elif method == 'random': + # Generate random sequence. + x = np.random.uniform(size=[nparticles]) + for dim in range(3): + l = box_vectors[dim][dim] + positions[:, dim] = unit.Quantity(x * l / l.unit, l.unit) else: raise Exception("method '%s' must be 'halton' or 'sobol'" % method) From 3a9cb3b2f312cc17f292558af605230bbb348cae Mon Sep 17 00:00:00 2001 From: "John Chodera (MSKCC)" Date: Mon, 2 Nov 2015 17:52:42 -0800 Subject: [PATCH 5/6] Restore sobol as default method for placing particles. --- openmmtools/testsystems.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/openmmtools/testsystems.py b/openmmtools/testsystems.py index 79e6841f6..b1e46fb88 100644 --- a/openmmtools/testsystems.py +++ b/openmmtools/testsystems.py @@ -155,7 +155,7 @@ def halton_sequence(p, n): return u -def subrandom_particle_positions(nparticles, box_vectors, method='random'): +def subrandom_particle_positions(nparticles, box_vectors, method='sobol'): """Generate a deterministic list of subrandom particle positions. Parameters From 92b0b235f862c0728b0c703e607be3663e19df23 Mon Sep 17 00:00:00 2001 From: John Chodera Date: Sun, 15 Nov 2015 21:25:56 -0500 Subject: [PATCH 6/6] Update test with timing test (not a real test yet). --- openmmtools/tests/test_integrators.py | 62 ++++++++++++++++++++++----- 1 file changed, 52 insertions(+), 10 deletions(-) diff --git a/openmmtools/tests/test_integrators.py b/openmmtools/tests/test_integrators.py index dd50da87b..5c47cd4bc 100644 --- a/openmmtools/tests/test_integrators.py +++ b/openmmtools/tests/test_integrators.py @@ -14,7 +14,7 @@ #============================================================================================= import re -import numpy +import numpy as np from simtk import unit from simtk import openmm @@ -58,7 +58,7 @@ def check_stability(integrator, test, platform=None, nsteps=100, temperature=300 # Check that simulation has not exploded. state = context.getState(getEnergy=True) potential = state.getPotentialEnergy() / kT - if numpy.isnan(potential): + if np.isnan(potential): raise Exception("Potential energy for integrator %s became NaN." % integrator.__doc__) del context @@ -113,9 +113,25 @@ def test_integrator_decorators(): assert integrator.acceptance_rate == 1.0 def test_fire_minimization(): - integrator = integrators.FIREMinimizationIntegrator() - testsystem = testsystems.LennardJonesFluid() - nsteps = 100 + tolerance = 10.0 * unit.kilojoules_per_mole / unit.nanometer + + # Disorder the system + testsystem = testsystems.DHFRExplicit() + nparticles = testsystem.system.getNumParticles() + temperature = 300 * unit.kelvin + collision_rate = 90.0 / unit.picosecond + timestep = 2.0 * unit.femtosecond + nsteps_equil = 2500 + integrator = openmm.LangevinIntegrator(temperature, collision_rate, timestep) + context = openmm.Context(testsystem.system, integrator) + context.setPositions(testsystem.positions) + integrator.step(nsteps_equil) + testsystem.positions = context.getState(getPositions=True).getPositions(asNumpy=True) + del context, integrator + + # Minimize. + integrator = integrators.FIREMinimizationIntegrator(tolerance=tolerance) + maxits = 100 context = openmm.Context(testsystem.system, integrator) context.setPositions(testsystem.positions) @@ -123,11 +139,37 @@ def test_fire_minimization(): initial_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole - integrator.step(nsteps) + import time + initial_time = time.time() + integrator.step(maxits) + final_time = time.time() + elapsed_time = final_time - initial_time + + final_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilojoules_per_mole + final_gnorm = np.sqrt( ((context.getState(getForces=True).getForces(asNumpy=True) / (unit.kilojoules_per_mole/unit.nanometer))**2).sum() / nparticles) + + print "FIRE minimizer (%.3f s)" % elapsed_time + print "Initial energy: %12.3f kJ/mol" % initial_energy + print "Final energy: %12.3f kJ/mol" % final_energy + print "Final gnorm: %12.3f kJ/mol/nm" % final_gnorm + + # Reset + context.setPositions(testsystem.positions) + initial_time = time.time() + openmm.LocalEnergyMinimizer.minimize(context, tolerance, maxits) + final_time = time.time() + elapsed_time = final_time - initial_time + + final_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilojoules_per_mole + final_gnorm = np.sqrt( ((context.getState(getForces=True).getForces(asNumpy=True) / (unit.kilojoules_per_mole/unit.nanometer))**2).sum() / nparticles) - final_energy = context.getState(getEnergy=True).getPotentialEnergy() / unit.kilocalories_per_mole + print "LocalEnergyMinimizer (%.3f s)" % elapsed_time + print "Initial energy: %12.3f kJ/mol" % initial_energy + print "Final energy: %12.3f kJ/mol" % final_energy + print "Final gnorm: %12.3f kJ/mol/nm" % final_gnorm - print "FIRE minimizer" - print "Initial energy: %12.3f kcal/mol" % initial_energy - print "Final energy: %12.3f kcal/mol" % final_energy + del context, integrator + +if __name__ == '__main__': + test_fire_minimization()