def test_item_access(self):
condensed_matrix_1 = CondensedMatrix([1.0, 4.5, 7.2, 8.5, 4.5, 7.8])
condensed_matrix_2 = CondensedMatrix([.0] * 6)
complete_matrix = [[0.0, 1.0, 4.5, 7.2], [1.0, 0.0, 8.5, 4.5],
[4.5, 8.5, 0.0, 7.8], [7.2, 4.5, 7.8, 0.0]]
row_len = condensed_matrix_1.row_length
for i in range(row_len):
for j in range(row_len):
condensed_matrix_2[i, j] = complete_matrix[i][j]
## The access for a complete and a condensed matrix is exactly the same
for i in range(row_len):
for j in range(row_len):
self.assertAlmostEqual(
condensed_matrix_1[i, j], complete_matrix[i][j],
6) #-> there are some errors, as it stores floats instead
# of doubles
## And we can build a condensed matrix as a complete matrix
self.assertItemsEqual(condensed_matrix_1.get_data(),
condensed_matrix_2.get_data())
def test_item_access(self):
condensed_matrix_1 = CondensedMatrix([1.0, 4.5,7.2,
8.5, 4.5,
7.8])
condensed_matrix_2 = CondensedMatrix([.0]*6)
complete_matrix = [[0.0, 1.0, 4.5, 7.2],
[1.0, 0.0, 8.5, 4.5],
[4.5, 8.5, 0.0, 7.8],
[7.2, 4.5, 7.8, 0.0]]
row_len = condensed_matrix_1.row_length
for i in range(row_len):
for j in range(row_len):
condensed_matrix_2[i,j] = complete_matrix[i][j]
## The access for a complete and a condensed matrix is exactly the same
for i in range(row_len):
for j in range(row_len):
self.assertAlmostEqual(condensed_matrix_1[i,j],complete_matrix[i][j],6) #-> there are some errors, as it stores floats instead
# of doubles
## And we can build a condensed matrix as a complete matrix
self.assertItemsEqual(condensed_matrix_1.get_data(), condensed_matrix_2.get_data())
def test_list_creation(self):
MAX_ELEMENTS = 30
DATA_LEN = (MAX_ELEMENTS * (MAX_ELEMENTS-1))/2
numpy_matrix_data = numpy.abs(numpy.random.rand(DATA_LEN))
np_matrix = CondensedMatrix(numpy_matrix_data)
ls_matrix = CondensedMatrix(list(numpy_matrix_data))
self.assertEqual(ls_matrix.row_length, MAX_ELEMENTS)
self.assertEqual(np_matrix.row_length, MAX_ELEMENTS)
numpy.testing.assert_almost_equal(numpy_matrix_data, ls_matrix.get_data(), 7)
numpy.testing.assert_almost_equal(numpy_matrix_data, np_matrix.get_data(), 7)
numpy.testing.assert_almost_equal(np_matrix.get_data(), ls_matrix.get_data(), 7)
def get_rmsds_matrix(conforms, mode, sup, cores):
print "Mode:", mode, "Superposition:", sup, "Number of cores:", cores
if (mode == "cpu_serial" and not sup):
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
"NOSUP_OMP_CALCULATOR", conforms)
elif (mode == "cpu_omp" and not sup):
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
"NOSUP_OMP_CALCULATOR", conforms)
calculator.setNumberOfOpenMPThreads(int(cores))
elif (mode == "cpu_omp" and sup):
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
"QCP_OMP_CALCULATOR", conforms)
calculator.setNumberOfOpenMPThreads(int(cores))
elif (mode == "cuda" and sup):
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
"QCP_CUDA_MEM_CALCULATOR", conforms)
else:
print "Wrong values to pyRMSD ! Please Fix"
exit()
rmsd = calculator.pairwiseRMSDMatrix()
rmsd_matrix = CondensedMatrix(rmsd)
inner_data = rmsd_matrix.get_data()
np.save("Distances_Matrix.data", inner_data)
return inner_data
def get_rmsds_matrix(conforms, mode, sup, cores, symm_groups=None):
print("Mode:", mode, "Superposition:", sup, "Number of cores:", cores)
if (mode == "cpu_serial" and not sup) or (mode == "cpu_omp" and not sup):
calculator_name = "NOSUP_OMP_CALCULATOR"
elif mode == "cpu_omp" and sup:
calculator_name = "QCP_OMP_CALCULATOR"
elif mode == "cuda" and sup:
calculator_name = "QCP_CUDA_MEM_CALCULATOR"
else:
print("Wrong values to pyRMSD ! Please Fix")
exit()
if symm_groups:
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
calculator_name,
fittingCoordsets=conforms,
calculationCoordsets=conforms,
calcSymmetryGroups=symm_groups)
else:
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
calculator_name, conforms)
# additionally set number of cores for parallel calculator
if mode == "cpu_omp":
calculator.setNumberOfOpenMPThreads(int(cores))
rmsd = calculator.pairwiseRMSDMatrix()
rmsd_matrix = CondensedMatrix(rmsd)
inner_data = rmsd_matrix.get_data()
np.save("Distances_Matrix.data", inner_data)
return inner_data
def test_force_sparsity(self):
data = numpy.array([1., 4., 6., 2., 5.,
3., 9., 7., 2.,
4., 1., 1.,
9.,3.,
8.])
W = CondensedMatrix(data)
SpectralTools.force_sparsity(W)
W_expected = [ 0., 0., 6., 0., 5., 0., 9., 7., 0., 0., 0., 0., 9., 0., 8.]
numpy.testing.assert_array_equal(W_expected, W.get_data())
def test_force_sparsity(self):
data = numpy.array([1., 4., 6., 2., 5.,
3., 9., 7., 2.,
4., 1., 1.,
9.,3.,
8.])
W = CondensedMatrix(data)
SpectralTools.force_sparsity(W)
W_expected = [ 0., 0., 6., 0., 5., 0., 9., 7., 0., 0., 0., 0., 9., 0., 8.]
numpy.testing.assert_array_equal(W_expected, W.get_data())
def get_rmsds_matrix(conforms, mode, sup, cores, symm_groups=None):
if (mode == "cpu_serial" and not sup) or (mode == "cpu_omp" and not sup):
calculator_name = "NOSUP_OMP_CALCULATOR"
elif mode == "cpu_omp" and sup:
calculator_name = "QCP_OMP_CALCULATOR"
print("we are using QCP_OMP to compute RMSD")
elif mode == "cuda" and sup:
calculator_name = "QCP_CUDA_MEM_CALCULATOR"
else:
print("Wrong values to pyRMSD ! Please Fix")
exit()
if symm_groups:
print("We have ambiguity.")
s1 = parse_symm_groups_for_pyrmsd(symm_groups)
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
calculator_name,
fittingCoordsets=conforms,
calcSymmetryGroups=s1,
fitSymmetryGroups=s1)
else:
calculator = pyRMSD.RMSDCalculator.RMSDCalculator(
calculator_name, conforms)
# additionally set number of cores for parallel calculator
if mode == "cpu_omp":
calculator.setNumberOfOpenMPThreads(int(cores))
if not symm_groups:
rmsd = calculator.pairwiseRMSDMatrix()
else:
rmsd = []
for i in range(len(conforms) - 1):
rmsd += list(calculator.oneVsFollowing(i))
rmsd_matrix = CondensedMatrix(rmsd)
inner_data = rmsd_matrix.get_data()
np.save("Distances_Matrix.data", inner_data)
return inner_data
