def test_SingleScaler_update_for_minimisation(): """Test the update_for_minimisation method of the singlescaler.""" # test_params.scaling_options.nproc = 1 p, e, r = (generated_param(), generated_exp(), generated_refl_2()) exp = create_scaling_model(p, e, r) p.reflection_selection.method = "use_all" single_scaler = SingleScaler(p, exp[0], r) apm_fac = create_apm_factory(single_scaler) single_scaler.components["scale"].parameters /= 2.0 apm = apm_fac.make_next_apm() Ih_table = single_scaler.Ih_table.blocked_data_list[0] Ih_table.calc_Ih() assert list(Ih_table.inverse_scale_factors) == [1.0, 1.0] assert list(Ih_table.Ih_values) == [10.0, 1.0] single_scaler.update_for_minimisation(apm, 0) # Should set new scale factors, and calculate Ih and weights. bf = basis_function().calculate_scales_and_derivatives(apm.apm_list[0], 0) assert list(Ih_table.inverse_scale_factors) == list(bf[0]) assert list(Ih_table.Ih_values) != [1.0, 10.0] assert approx_equal(list(Ih_table.Ih_values), list(Ih_table.intensities / bf[0])) for i in range(Ih_table.derivatives.n_rows): for j in range(Ih_table.derivatives.n_cols): assert approx_equal(Ih_table.derivatives[i, j], bf[1][i, j]) assert Ih_table.derivatives.non_zeroes == bf[1].non_zeroes
def test_SingleScaler_update_for_minimisation(): """Test the update_for_minimisation method of the singlescaler.""" # test_params.scaling_options.nproc = 1 p, e, r = (generated_param(), generated_exp(), generated_refl_2()) exp = create_scaling_model(p, e, r) p.reflection_selection.method = "use_all" single_scaler = SingleScaler(p, exp[0], r) pmg = ScalingParameterManagerGenerator( single_scaler.active_scalers, ScalingTarget(), single_scaler.params.scaling_refinery.refinement_order, ) single_scaler.components["scale"].parameters /= 2.0 apm = pmg.parameter_managers()[0] Ih_table = single_scaler.Ih_table.blocked_data_list[0] Ih_table.calc_Ih() assert list(Ih_table.inverse_scale_factors) == [1.0, 1.0] assert list(Ih_table.Ih_values) == [10.0, 1.0] single_scaler.update_for_minimisation(apm, 0) # Should set new scale factors, and calculate Ih and weights. bf = RefinerCalculator.calculate_scales_and_derivatives(apm.apm_list[0], 0) assert list(Ih_table.inverse_scale_factors) == list(bf[0]) assert list(Ih_table.Ih_values) != [1.0, 10.0] assert list(Ih_table.Ih_values) == pytest.approx( list(Ih_table.intensities / bf[0])) for i in range(Ih_table.derivatives.n_rows): for j in range(Ih_table.derivatives.n_cols): assert Ih_table.derivatives[i, j] == pytest.approx(bf[1][i, j]) assert Ih_table.derivatives.non_zeroes == bf[1].non_zeroes
def test_target_gradient_calculation_finite_difference(small_reflection_table, single_exp, physical_param): """Test the calculated gradients against a finite difference calculation.""" model = PhysicalScalingModel.from_data(physical_param, single_exp, small_reflection_table) # need to 'add_data' model.configure_components(small_reflection_table, single_exp, physical_param) model.components["scale"].update_reflection_data() model.components["decay"].update_reflection_data() apm = multi_active_parameter_manager( ScalingTarget(), [model.components], [["scale", "decay"]], scaling_active_parameter_manager, ) model.components["scale"].inverse_scales = flex.double([2.0, 1.0, 2.0]) model.components["decay"].inverse_scales = flex.double([1.0, 1.0, 0.4]) Ih_table = IhTable([small_reflection_table], single_exp.crystal.get_space_group()) with patch.object(SingleScaler, "__init__", lambda x, y, z, k: None): scaler = SingleScaler(None, None, None) scaler._Ih_table = Ih_table # Now do finite difference check. target = ScalingTarget() scaler.update_for_minimisation(apm, 0) grad = target.calculate_gradients(scaler.Ih_table.blocked_data_list[0]) res = target.calculate_residuals(scaler.Ih_table.blocked_data_list[0]) assert (res > 1e-8), """residual should not be zero, or the gradient test below will not really be working!""" # Now compare to finite difference f_d_grad = calculate_gradient_fd(target, scaler, apm) print(list(f_d_grad)) print(list(grad)) assert list(grad) == pytest.approx(list(f_d_grad)) sel = f_d_grad > 1e-8 assert sel, """assert sel has some elements, as finite difference grad should
def test_target_jacobian_calculation_finite_difference(physical_param, single_exp, large_reflection_table): """Test the calculated jacobian against a finite difference calculation.""" physical_param.physical.decay_correction = False model = PhysicalScalingModel.from_data(physical_param, single_exp, large_reflection_table) # need to 'add_data' model.configure_components(large_reflection_table, single_exp, physical_param) model.components["scale"].update_reflection_data() apm = multi_active_parameter_manager( ScalingTarget(), [model.components], [["scale"]], scaling_active_parameter_manager, ) Ih_table = IhTable([large_reflection_table], single_exp.crystal.get_space_group()) with patch.object(SingleScaler, "__init__", lambda x, y, z, k: None): scaler = SingleScaler(None, None, None) scaler._Ih_table = Ih_table target = ScalingTarget() scaler.update_for_minimisation(apm, 0) fd_jacobian = calculate_jacobian_fd(target, scaler, apm) r, jacobian, w = target.compute_residuals_and_gradients( scaler.Ih_table.blocked_data_list[0]) assert r == pytest.approx( [-50.0 / 3.0, 70.0 / 3.0, -20.0 / 3.0, 12.5, -2.5] + [-25.0, 0.0, -75.0, 0.0, 200.0]) assert w == pytest.approx( [0.1, 0.1, 0.1, 0.02, 0.1, 0.02, 0.01, 0.02, 0.01, 0.01]) n_rows = jacobian.n_rows n_cols = jacobian.n_cols print(jacobian) print(fd_jacobian) for i in range(0, n_rows): for j in range(0, n_cols): assert jacobian[i, j] == pytest.approx(fd_jacobian[i, j], abs=1e-4)
def test_update_error_model(mock_errormodel, mock_errormodel2): """Test the update_error_model method""" p, e, r = (generated_param(), generated_exp(), generated_refl()) exp = create_scaling_model(p, e, r) p.reflection_selection.method = "use_all" # test initialised correctly scaler = SingleScaler(p, exp[0], r) block = scaler.global_Ih_table.blocked_data_list[0] original_vars = block.variances # test update error model - should update weights in global Ih # as will be setting different things in Ih_table and reflection table, split # up the test to use two different error models. scaler._update_error_model(mock_errormodel) assert list(block.variances) == list(original_vars) newvars = flex.double(range(1, 8)) assert list(block.block_selections[0]) == [2, 0, 4, 5, 6, 1, 3] assert list(block.weights) == list(1.0 / newvars) assert scaler.experiment.scaling_model.error_model is mock_errormodel # now test for updating of reflection table # do again with second errormodel scaler.global_Ih_table.reset_error_model() scaler._update_error_model(mock_errormodel2) assert list(block.variances) == list(original_vars) newvars = flex.double(range(1, 9)) assert list(block.block_selections[0]) == [2, 0, 4, 5, 6, 1, 3] # [2, 3, 4, 5, 6, 7, 8] < set these in ^ these positions (taking into account # the one non-suitable refl at index 5) assert list(block.weights) == list(1.0 / newvars)[:-1] assert scaler.experiment.scaling_model.error_model is mock_errormodel2
def test_SingleScaler_combine_intensities(): """test combine intensities method""" p, e, r = (generated_param(), generated_exp(), generated_refl_for_comb()) exp = create_scaling_model(p, e, r) p.reflection_selection.method = "use_all" scaler = SingleScaler(p, exp[0], r) scaler.combine_intensities() # The input makes the profile intensities best - so check these are set in the # reflection table and global_Ih_table assert list(scaler.reflection_table["intensity"]) == list( r["intensity.prf.value"]) assert list(scaler.reflection_table["variance"]) == list( r["intensity.prf.variance"]) block = scaler.global_Ih_table.blocked_data_list[0] block_sel = block.block_selections[0] suitable = scaler.suitable_refl_for_scaling_sel assert list(block.intensities) == list( scaler.reflection_table["intensity"].select(suitable).select( block_sel)) assert list(block.variances) == list( scaler.reflection_table["variance"].select(suitable).select(block_sel))
def test_SingleScaler_expand_scales_to_all_reflections(mock_apm): p, e, r = (generated_param(), generated_exp(), generated_refl()) exp = create_scaling_model(p, e, r) p.reflection_selection.method = "use_all" scaler = SingleScaler(p, exp[0], r) # test expand to all reflections method. First check scales are all 1, then # update a component to simulate a minimisation result, then check that # scales are set only in all suitable reflections (as it may not be possible # to calculate scales for unsuitable reflections!) # Must also update the scales in the global_Ih_table assert list(scaler.reflection_table["inverse_scale_factor"]) == [1.0] * 8 scaler.experiment.scaling_model.components[ "scale"].parameters = flex.double([2.0]) scaler.expand_scales_to_all_reflections(calc_cov=False) assert (list( scaler.reflection_table["inverse_scale_factor"]) == [2.0] * 5 + [1.0] + [2.0] * 2) assert (list( scaler.global_Ih_table.blocked_data_list[0].inverse_scale_factors) == [2.0] * 7) assert list( scaler.reflection_table["inverse_scale_factor_variance"]) == [0.0] * 8 # now try again apm = Mock() apm.n_active_params = 2 var_list = [1.0, 0.1, 0.1, 0.5] apm.var_cov_matrix = flex.double(var_list) apm.var_cov_matrix.reshape(flex.grid(2, 2)) scaler.update_var_cov(apm) assert scaler.var_cov_matrix[0, 0] == var_list[0] assert scaler.var_cov_matrix[0, 1] == var_list[1] assert scaler.var_cov_matrix[1, 0] == var_list[2] assert scaler.var_cov_matrix[1, 1] == var_list[3] assert scaler.var_cov_matrix.non_zeroes == 4 scaler.expand_scales_to_all_reflections(calc_cov=True) assert list( scaler.reflection_table["inverse_scale_factor_variance"] ) == pytest.approx( [2.53320, 1.07106, 1.08125, 1.23219, 1.15442, 0.0, 1.0448, 1.0448], 1e-4) # Second case - when var_cov_matrix is only part of full matrix. p, e, r = (generated_param(), generated_exp(), generated_refl()) exp = create_scaling_model(p, e, r) p.reflection_selection.method = "use_all" scaler = SingleScaler(p, exp[0], r) apm = mock_apm scaler.update_var_cov(apm) assert scaler.var_cov_matrix.non_zeroes == 1 assert scaler.var_cov_matrix[0, 0] == 2.0 assert scaler.var_cov_matrix.n_cols == 2 assert scaler.var_cov_matrix.n_rows == 2 assert scaler.var_cov_matrix.non_zeroes == 1
def test_targetscaler_initialisation(): """Unit tests for the MultiScalerBase class.""" p, e = (generated_param(), generated_exp(2)) r1 = generated_refl(id_=0) p.reflection_selection.method = "intensity_ranges" r1["intensity.sum.value"] = r1["intensity"] r1["intensity.sum.variance"] = r1["variance"] r2 = generated_refl(id_=1) r2["intensity.sum.value"] = r2["intensity"] r2["intensity.sum.variance"] = r2["variance"] exp = create_scaling_model(p, e, [r1, r2]) singlescaler1 = SingleScaler(p, exp[0], r1, for_multi=True) singlescaler2 = SingleScaler(p, exp[1], r2, for_multi=True) # singlescaler2.experiments.scaling_model.set_scaling_model_as_scaled() targetscaler = TargetScaler(scaled_scalers=[singlescaler1], unscaled_scalers=[singlescaler2]) # check initialisation assert len(targetscaler.active_scalers) == 1 assert len(targetscaler.single_scalers) == 1 assert targetscaler.active_scalers[0] == singlescaler2 assert targetscaler.single_scalers[0] == singlescaler1 # check for correct setup of global Ih table assert targetscaler.global_Ih_table.size == 7 # only for active scalers assert list( targetscaler.global_Ih_table.blocked_data_list[0].intensities) == [ 3.0, 1.0, 500.0, 2.0, 2.0, 2.0, 4.0, ] block_selections = targetscaler.global_Ih_table.blocked_data_list[ 0].block_selections assert list(block_selections[0]) == [2, 0, 4, 5, 6, 1, 3] # check for correct setup of Ih_table assert targetscaler.Ih_table.size == 6 assert list(targetscaler.Ih_table.blocked_data_list[0].intensities) == [ 3.0, 1.0, 2.0, 2.0, 2.0, 4.0, ] block_selections = targetscaler.Ih_table.blocked_data_list[ 0].block_selections assert list(block_selections[0]) == [2, 0, 5, 6, 1, 3] # check for correct setup of target Ih_Table assert targetscaler.target_Ih_table.size == 6 assert list( targetscaler.target_Ih_table.blocked_data_list[0].intensities) == [ 3.0, 1.0, 2.0, 2.0, 2.0, 4.0, ] block_selections = targetscaler.target_Ih_table.blocked_data_list[ 0].block_selections assert list(block_selections[0]) == [ 2, 0, 4, 5, 1, 3, ] # different as taget_Ih_table # not created with indices lists. block_selections = targetscaler.Ih_table.blocked_data_list[ 0].block_selections # check for correct data/d_values in components for i, scaler in enumerate(targetscaler.active_scalers): d_suitable = scaler.reflection_table["d"].select( scaler.suitable_refl_for_scaling_sel) decay = scaler.experiment.scaling_model.components["decay"] # first check 'data' contains all suitable reflections assert list(decay.data["d"]) == list(d_suitable) # Now check 'd_values' (which will be used for minim.) matches Ih_table data assert list(decay.d_values[0]) == list( d_suitable.select(block_selections[i])) # but shouldn't have updated other assert (targetscaler.single_scalers[0].experiment.scaling_model. components["decay"].d_values == [])
def test_SingleScaler_initialisation(): """Test that all attributes are correctly set upon initialisation""" p, e, r = (generated_param(), generated_exp(), generated_refl()) exp = create_scaling_model(p, e, r) p.reflection_selection.method = "use_all" # test initialised correctly scaler = SingleScaler(p, exp[0], r) assert (list(scaler.suitable_refl_for_scaling_sel) == [True] * 5 + [False] + [True] * 2) # all 7 of the suitable should be within the scaling_subset assert list(scaler.scaling_subset_sel) == [True] * 7 # one of these is not in the scaling selection due to being an outlier. assert list(scaler.scaling_selection) == [True] * 4 + [False] + [True] * 2 assert list(scaler.outliers) == [False] * 4 + [True] + [False] * 2 assert scaler.n_suitable_refl == 7 # check for correct setup of global_Ih_table # block selection is order to extract out from suitable_reflections assert scaler.global_Ih_table.size == 7 assert list(scaler.global_Ih_table.blocked_data_list[0].intensities) == [ 3.0, 1.0, 500.0, 2.0, 2.0, 2.0, 4.0, ] block_selection = scaler.global_Ih_table.blocked_data_list[ 0].block_selections[0] assert list(block_selection) == [2, 0, 4, 5, 6, 1, 3] # check for correct setup of Ih_table assert scaler.Ih_table.size == 6 assert list(scaler.Ih_table.blocked_data_list[0].intensities) == [ 3.0, 1.0, 2.0, 2.0, 2.0, 4.0, ] block_selection = scaler.Ih_table.blocked_data_list[0].block_selections[0] assert list(block_selection) == [2, 0, 5, 6, 1, 3] # check for correct data/d_values in components d_suitable = r["d"].select(scaler.suitable_refl_for_scaling_sel) decay = scaler.experiment.scaling_model.components["decay"] # first check 'data' contains all suitable reflections assert list(decay.data["d"]) == list(d_suitable) # Now check 'd_values' (which will be used for minim.) matches Ih_table data assert list(decay.d_values[0]) == list(d_suitable.select(block_selection)) # test make ready for scaling method # set some new outliers and check for updated datastructures outlier_list = [False] * 3 + [True] * 2 + [False] * 2 scaler.outliers = flex.bool(outlier_list) scaler.make_ready_for_scaling(outlier=True) assert scaler.Ih_table.size == 5 assert list(scaler.Ih_table.blocked_data_list[0].intensities) == [ 3.0, 1.0, 2.0, 2.0, 2.0, ] block_selection = scaler.Ih_table.blocked_data_list[0].block_selections[0] assert list(block_selection) == [2, 0, 5, 6, 1] assert list(decay.d_values[0]) == list(d_suitable.select(block_selection)) # test set outliers assert list(r.get_flags(r.flags.outlier_in_scaling)) == [False] * 8 scaler._set_outliers() assert list(r.get_flags( r.flags.outlier_in_scaling)) == outlier_list + [False]
def create(cls, params, experiment, reflection_table, for_multi=False): """Perform reflection_table preprocessing and create a SingleScaler.""" cls.ensure_experiment_identifier(params, experiment, reflection_table) logger.info( "Preprocessing data for scaling. The id assigned to this \n" "dataset is %s, and the scaling model type being applied is %s. \n", list(reflection_table.experiment_identifiers().values())[0], experiment.scaling_model.id_, ) reflection_table, reasons = cls.filter_bad_reflections( reflection_table) if "inverse_scale_factor" not in reflection_table: reflection_table["inverse_scale_factor"] = flex.double( reflection_table.size(), 1.0) elif (reflection_table["inverse_scale_factor"].count(0.0) == reflection_table.size()): reflection_table["inverse_scale_factor"] = flex.double( reflection_table.size(), 1.0) reflection_table = choose_scaling_intensities( reflection_table, params.reflection_selection.intensity_choice) excluded_for_scaling = reflection_table.get_flags( reflection_table.flags.excluded_for_scaling) user_excluded = reflection_table.get_flags( reflection_table.flags.user_excluded_in_scaling) reasons.add_reason("user excluded", user_excluded.count(True)) reasons.add_reason("excluded for scaling", excluded_for_scaling.count(True)) n_excluded = (excluded_for_scaling | user_excluded).count(True) if n_excluded == reflection_table.size(): logger.info( "All reflections were determined to be unsuitable for scaling." ) logger.info(reasons) raise BadDatasetForScalingException( """Unable to use this dataset for scaling""") else: logger.info( "%s/%s reflections not suitable for scaling\n%s", n_excluded, reflection_table.size(), reasons, ) if not for_multi: determine_reflection_selection_parameters(params, [experiment], [reflection_table]) if params.reflection_selection.method == "intensity_ranges": reflection_table = quasi_normalisation(reflection_table, experiment) if (params.reflection_selection.method in (None, Auto, "auto", "quasi_random")) or ( experiment.scaling_model.id_ == "physical" and "absorption" in experiment.scaling_model.components): if experiment.scan: # calc theta and phi cryst reflection_table["phi"] = ( reflection_table["xyzobs.px.value"].parts()[2] * experiment.scan.get_oscillation()[1]) reflection_table = calc_crystal_frame_vectors( reflection_table, experiment) return SingleScaler(params, experiment, reflection_table, for_multi)
def create(cls, params, experiment, reflection_table, for_multi=False): """Perform reflection_table preprocessing and create a SingleScaler.""" cls.ensure_experiment_identifier(experiment, reflection_table) logger.info( "The scaling model type being applied is %s. \n", experiment.scaling_model.id_, ) try: reflection_table = cls.filter_bad_reflections( reflection_table, partiality_cutoff=params.cut_data.partiality_cutoff, min_isigi=params.cut_data.min_isigi, intensity_choice=params.reflection_selection.intensity_choice, ) except ValueError: raise BadDatasetForScalingException # combine partial measurements of same reflection, to handle those reflections # that were split by dials.integrate - changes size of reflection table. reflection_table = sum_partial_reflections(reflection_table) if "inverse_scale_factor" not in reflection_table: reflection_table["inverse_scale_factor"] = flex.double( reflection_table.size(), 1.0) elif (reflection_table["inverse_scale_factor"].count(0.0) == reflection_table.size()): reflection_table["inverse_scale_factor"] = flex.double( reflection_table.size(), 1.0) reflection_table = choose_initial_scaling_intensities( reflection_table, params.reflection_selection.intensity_choice) excluded_for_scaling = reflection_table.get_flags( reflection_table.flags.excluded_for_scaling) user_excluded = reflection_table.get_flags( reflection_table.flags.user_excluded_in_scaling) reasons = Reasons() reasons.add_reason("user excluded", user_excluded.count(True)) reasons.add_reason("excluded for scaling", excluded_for_scaling.count(True)) n_excluded = (excluded_for_scaling | user_excluded).count(True) if n_excluded == reflection_table.size(): logger.info( "All reflections were determined to be unsuitable for scaling." ) logger.info(reasons) raise BadDatasetForScalingException( """Unable to use this dataset for scaling""") else: logger.info( "Excluding %s/%s reflections\n%s", n_excluded, reflection_table.size(), reasons, ) if params.reflection_selection.method == "intensity_ranges": reflection_table = quasi_normalisation(reflection_table, experiment) if (params.reflection_selection.method in (None, Auto, "auto", "quasi_random")) or ( experiment.scaling_model.id_ == "physical" and "absorption" in experiment.scaling_model.components): if experiment.scan: reflection_table = calc_crystal_frame_vectors( reflection_table, experiment) alignment_axis = (1.0, 0.0, 0.0) reflection_table["s0c"] = align_axis_along_z( alignment_axis, reflection_table["s0c"]) reflection_table["s1c"] = align_axis_along_z( alignment_axis, reflection_table["s1c"]) try: scaler = SingleScaler(params, experiment, reflection_table, for_multi) except BadDatasetForScalingException as e: raise ValueError(e) else: return scaler