def parmfit(datfname, idxfname, param):
fnames, enes, weights = tools.readdat(datfname)
idxs, folds = readidx(idxfname)
params = getparams(idxs, param)
A = []
B = []
for fname, E_qm, weight in zip(fnames, enes, weights):
mol = read.readxyz(file(fname))
tors = analyze.gettorsbytype(mol, idxs)
newmol, E_mm = tinker.minimize_file(fname, param)
E_tor = getetor(newmol, tors, params)
E_fit = E_qm - E_mm + E_tor
weight = math.sqrt(weight)
B.append(E_fit * weight)
A.append([x*weight for x in getA(newmol, idxs, folds, tors)])
A = numpy.array(A)
B = numpy.array(B)
result = numpy.linalg.lstsq(A, B)
newparam = list(result[0][:-1])
errors = numpy.dot(A, result[0]) - B
errors = [error/math.sqrt(weight) for error, weight in zip(errors, weights)]
error = extra_tools.weightedsd(errors, weights)
printprm(newparam, idxs, folds)
print error
def parmeval(datfname, param):
fnames, enes, weights = tortools.readdat(datfname)
writer = csv.writer(sys.stdout)
writer.writerow(['Filename', 'weight', 'E_qm', 'E_mm', 'E_diff', 'RMSD'])
E_diffs = []
rmsds = []
for fname, E_qm, weight in zip(fnames, enes, weights):
mol = read.readxyz(file(fname))
newmol, E_mm = tinker.minimize_file(fname, param)
E_diff = E_mm - E_qm
E_diffs.append(E_diff)
rmsd_ = rmsd.rmsd(mol, newmol)
rmsds.append(rmsd_)
row = [fname, weight, E_qm, E_mm, E_diff, rmsd_]
writer.writerow(row)
writer.writerow([])
writer.writerow(['Weighted Averrage of E_diff',
tools.weightedmean(E_diffs, weights)])
writer.writerow(['Weighted SD of E_diff',
tools.weightedsd(E_diffs, weights)])
writer.writerow(['Weighted Average of RMSD',
tools.weightedmean(rmsds, weights)])