def run_fit_procedure(args):
# run_fit_proecdure runs the actual fitting procedure, currently TSVD method.
# load files and set necessary directories
cwd = os.getcwd()
new = model.langevin_model(args.name) # load the model
cwd = os.getcwd()
new = model.langevin_model(args.name)
iterdir = "%s/iteration_%d" % (cwd, new.iteration)
newtondir = "%s/newton" % iterdir
os.chdir(newtondir)
lambda_index = get_lambda_index(args.cutoff) # get the lambda index for the cutoff value
param_changes = np.loadtxt("xp_%d.dat" % lambda_index) # load the correct file
# now re-scale the solution so that it is not too large.
if np.max(param_changes) > args.scale:
param_changes *= args.scale / np.max(param_changes)
# apply the changes to the parameters
count = 0
for i in new.fit_indices:
new.params[i] += param_changes[count]
count += 1
# save the changes
np.savetxt("params", new.params)
def run_fit_procedure(args):
#run_fit_proecdure runs the actual fitting procedure, currently TSVD method.
#load files and set necessary directories
cwd = os.getcwd()
new = model.langevin_model(args.name) #load the model
cwd = os.getcwd()
new = model.langevin_model(args.name)
iterdir = "%s/iteration_%d" % (cwd, new.iteration)
newtondir = "%s/newton" % iterdir
os.chdir(newtondir)
lambda_index = get_lambda_index(
args.cutoff) #get the lambda index for the cutoff value
param_changes = np.loadtxt("xp_%d.dat" %
lambda_index) #load the correct file
#now re-scale the solution so that it is not too large.
if np.max(np.abs(param_changes)) > args.scale:
param_changes *= (args.scale / np.max(np.abs(param_changes)))
#apply the changes to the parameters
count = 0
for i in new.fit_indices:
new.params[i] += param_changes[count]
count += 1
new.params[new.params < args.pmin] = args.pmin
#save the changes
np.savetxt("params", new.params)
def run_jac_procedure(args):
# run_jac runs a jacobian calculating procedure. Currently only implemented for a FRET fitting
# Set file locations and load necessary files
target_feature = np.loadtxt(args.target)
cwd = os.getcwd()
new = model.langevin_model(args.name) # load the model
iterdir = "%s/iteration_%d" % (cwd, new.iteration)
newtondir = "%s/newton" % iterdir
try:
os.mkdir(newtondir)
except:
pass
os.chdir(iterdir)
x = np.loadtxt("position.dat")
range_hist = (-1.0 * args.histrange, 1.0 * args.histrange)
sim_feature, bincenters, slices, spacing = compute.plot_x_histogram(
x, "iteration_%d" % new.iteration, nbins=args.nbins, histrange=range_hist
) # plot histogram and return histogram values
Jacobian = compute.compute_jacobian(
new, x, slices, sim_feature, args.nbins, spacing
) # compute the jacobian the FRET way
# save the files
os.chdir(newtondir)
np.savetxt("sim_feature.dat", sim_feature)
np.savetxt("target_feature.dat", target_feature)
np.savetxt("Jacobian.dat", Jacobian)
compute.fit_jacobian()
def run_jac_procedure(args):
#run_jac runs a jacobian calculating procedure. Currently only implemented for a FRET fitting
#Set file locations and load necessary files
target_feature = np.loadtxt(args.target)
cwd = os.getcwd()
new = model.langevin_model(args.name) #load the model
iterdir = "%s/iteration_%d" % (cwd, new.iteration)
newtondir = "%s/newton" % iterdir
try:
os.mkdir(newtondir)
except:
pass
os.chdir(iterdir)
x = np.loadtxt("position.dat")
range_hist = (-1.0 * args.histrange, 1.0 * args.histrange)
sim_feature, bincenters, slices, spacing = compute.plot_x_histogram(
x,
"iteration_%d" % new.iteration,
nbins=args.nbins,
histrange=range_hist) #plot histogram and return histogram values
Jacobian = compute.compute_jacobian(
new, x, slices, sim_feature, args.nbins,
spacing) #compute the jacobian the FRET way
#save the files
os.chdir(newtondir)
np.savetxt("sim_feature.dat", sim_feature)
np.savetxt("target_feature.dat", target_feature)
np.savetxt("Jacobian.dat", Jacobian)
compute.fit_jacobian()
def run_next_step(args):
#run_next_step takes the previously found params in iteration_i/newton/params and sets up the ini file for iteration i+1
#set necessary directories
cwd = os.getcwd()
new = model.langevin_model(args.name) #load the model
cwd = os.getcwd()
new = model.langevin_model(args.name)
iterdir = "%s/iteration_%d" % (cwd, new.iteration)
newtondir = "%s/newton" % iterdir
new.iteration += 1
new.save_params_file = "%s/params" % newtondir
#save ini file with new location of the new model params
if args.start:
new.save_ini_file()
def run_next_step(args):
# run_next_step takes the previously found params in iteration_i/newton/params and sets up the ini file for iteration i+1
# set necessary directories
cwd = os.getcwd()
new = model.langevin_model(args.name) # load the model
cwd = os.getcwd()
new = model.langevin_model(args.name)
iterdir = "%s/iteration_%d" % (cwd, new.iteration)
newtondir = "%s/newton" % iterdir
new.iteration += 1
new.save_params_file = "%s/params" % newtondir
# save ini file with new location of the new model params
if args.start:
new.save_ini_file()
def run_sim(args):
# run_sim runs a stanard simulation based on the args given.
cwd = os.getcwd()
new = model.langevin_model(args.name) # load the model
x, v = compute.run_langevin(new, nsteps=args.steps) # run the actual computation
# save the results below and make the appropriate directories
iterdir = "iteration_%d" % new.iteration
try:
os.mkdir(iterdir)
except:
pass
np.savetxt("%s/position.dat" % iterdir, x)
np.savetxt("%s/velocity.dat" % iterdir, v)
os.chdir(iterdir)
compute.plot_x_inverted(new, x)
new.save_parameter_files()
def run_sim(args):
#run_sim runs a stanard simulation based on the args given.
cwd = os.getcwd()
new = model.langevin_model(args.name) #load the model
x, v = compute.run_langevin(new,
nsteps=args.steps) #run the actual computation
#save the results below and make the appropriate directories
iterdir = "iteration_%d" % new.iteration
try:
os.mkdir(iterdir)
except:
pass
np.savetxt("%s/position.dat" % iterdir, x)
np.savetxt("%s/velocity.dat" % iterdir, v)
os.chdir(iterdir)
compute.plot_x_inverted(new, x)
new.save_parameter_files()
def run_energy_plotting(name, savedir, iters, plot_range, plot_step, axis, title):
#plot the potnetial energy of several different iterations
cwd = os.getcwd()
new = model.langevin_model(name)
plot_values = np.arange(plot_range[0], plot_range[1], plot_step)
netdir = "%s/net_potential" % savedir
paramdir = "%s/individual_potentials" % savedir
if not os.path.isdir(netdir):
os.mkdir(netdir)
if not os.path.isdir(paramdir):
os.mkdir(paramdir)
individual_scaled_values = []
individual_scaled_potentials = []
label = []
for j in range(len(new.params)):
individual_scaled_values.append(plot_values)
individual_scaled_potentials.append(new.potential_functions[j](plot_values))
label.append("p-%d"%j)
os.chdir(savedir)
plot_simple(individual_scaled_values, individual_scaled_potentials, label, "all-potentials", "position", "potential energy", axis=axis, marker_type="", use_legend=False)
for i in iters:
os.chdir(cwd)
individual_scaled_values = []
individual_scaled_potentials = []
label = []
new.params = np.loadtxt("iteration_%d/params"%i)
potential_values = new.net_potential(plot_values)
os.chdir(netdir)
plot_one_simple(plot_values, potential_values, "net-potential_iteration-%d"%i, "position", "potential energy", axis=axis)
os.chdir(paramdir)
for j in range(len(new.params)):
individual_scaled_values.append(plot_values)
individual_scaled_potentials.append(new.params[j]*new.potential_functions[j](plot_values))
label.append("p-%d"%j)
plot_simple(individual_scaled_values, individual_scaled_potentials, label, "potentials_iteration-%d"%i, "position", "potential energy", axis=axis, marker_type="", use_legend=False)
os.chdir(cwd)
def run_jac_procedure(args):
#run_jac runs a jacobian calculating procedure.
#Set file locations and load necessary files
target_feature = np.loadtxt(args.target)
cwd = os.getcwd()
new = model.langevin_model(args.name) #load the model
iterdir = "%s/iteration_%d" % (cwd, new.iteration)
newtondir = "%s/newton" % iterdir
try:
os.mkdir(newtondir)
except:
pass
os.chdir(iterdir)
x = np.loadtxt("position.dat")
range_hist = (-1.0 * args.histrange, 1.0 * args.histrange)
sim_feature, bincenters, slices, spacing = compute.plot_x_histogram(
x,
"iteration_%d" % new.iteration,
nbins=args.nbins,
histrange=range_hist) #plot histogram and return histogram values
if args.data_type == "FRET":
Jacobian = compute.compute_jacobian(
new, x, slices, sim_feature, args.nbins,
spacing) #compute the jacobian the FRET way
else:
sim_feature, bincenters, slices, ran_size, spacing = compute.plot_tmatrix(
x,
"iteration_%d" % new.iteration,
nbins=args.nbins,
histrange=range_hist,
framestep=args.fstep)
slices -= (ran_size[0] + 1)
nbins_use = np.shape(sim_feature)[0] #number of bins in trimmed matrix
sim_feature = np.ndarray.flatten(sim_feature)
#trim and flatten target_feature
target_feature = target_feature[ran_size[0]:ran_size[1],
ran_size[0]:ran_size[1]]
if not np.shape(target_feature)[0] == nbins_use:
raise ValueError(
"target_feature and sim_feature have different dimensions!")
target_feature = np.ndarray.flatten(target_feature)
if args.data_type == "tmatrix":
Jacobian = compute.compute_tmatrix_jacobian(
new,
x,
slices,
sim_feature,
nbins_use,
spacing,
framestep=args.fstep) #compute Jacobian the tmatrix way
if args.data_type == "new_tmatrix":
Jacobian = compute.compute_tmatrix_jac_new(
new,
x,
slices,
sim_feature,
nbins_use,
spacing,
framestep=args.fstep) #testing new tmatrix procedure
#save the files
os.chdir(newtondir)
np.savetxt("sim_feature.dat", sim_feature)
np.savetxt("target_feature.dat", target_feature)
np.savetxt("Jacobian.dat", Jacobian)
if args.data_type == "new_tmatrix":
compute.fit_temp()
else:
compute.fit_jacobian()