def test_same_distribution_after_household_subtraction(self):
        """Using the control_totals and no marginal characteristics,
        subtract households and ensure that the distribution within each group stays the same
        """
        annual_household_control_totals_data = {
            "year": array([2000]),
            "total_number_of_households": array([20000])
            }

        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name='hh_set', table_data=self.households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name='hct_set', table_data=annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what="household", id_name="year")

        storage.write_table(table_name='hc_set', table_data=self.household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        model = HouseholdTransitionModel()
        model.run(year=2000, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        #check that there are indeed 20000 total households after running the model
        results = hh_set.size()
        should_be = [20000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the distribution of households in each group is the same as before running the model
        results = self.get_count_all_groups(hh_set)
        should_be = [6000.0/33000.0*20000.0, 2000.0/33000.0*20000.0, 3000.0/33000.0*20000.0, 4000.0/33000.0*20000.0,
                     2000.0/33000.0*20000.0, 5000.0/33000.0*20000.0, 3000.0/33000.0*20000.0, 8000.0/33000.0*20000.0]
        self.assertEqual(ma.allclose(results, should_be, rtol=0.05),
                         True, "Error, should_be: %s,\n but result: %s" % (should_be, results))

    def test_controlling_with_three_marginal_characteristics(self):
        """Controlling with all three possible marginal characteristics in this example, age_of_head, income, and persons,
        this would partition the 8 groups into the same 8 groups, and with a control total specified for each group, we must
        ensure that the control totals for each group exactly meet the specifications.
        """

        #IMPORTANT: marginal characteristics grouping indices have to start at 0!
        annual_household_control_totals_data = {
            "year": array(8*[2000]),
            #"age_of_head": array(4*[0] + 4*[1]),
            "age_of_head_min": array([ 0, 0, 0, 0, 50, 50, 50, 50]),
            "age_of_head_max": array([49,49,49,49,100,100,100,100]),
            #"income": array(2*[0] + 2*[1] + 2*[0] + 2*[1]),
            "income_min": array([    0,    0,40000,40000,    0,    0,40000,40000]),
            "income_max": array([39999,39999,   -1,   -1,39999,39999,   -1,   -1]),
            #"persons": array([0,1,0,1,0,1,0,1]),
            "persons_min": array([0, 3,0, 3,0, 3,0, 3]),
            "persons_max": array([2,-1,2,-1,2,-1,2,-1]),
            "total_number_of_households": array([4000, 5000, 1000, 3000, 0, 6000, 3000, 8000])
            }
        ##size of columns was not even, removed last element of min and max
        #household_characteristics_for_ht_data = {
            #"characteristic": array(2*['age_of_head'] + 2*['income'] + 2*['persons']),
            #"min": array([0, 50, 0, 40000, 0, 3]),
            #"max": array([49, 100, 39999, -1, 2, -1]) 
            #}
        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name='hh_set', table_data=self.households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name='hct_set', table_data=annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what='household', id_name=[])

        #storage.write_table(table_name='hc_set', table_data=household_characteristics_for_ht_data)
        #hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        # unplace some households
        where10 = where(hh_set.get_attribute("grid_id")<>10)[0]
        hh_set.modify_attribute(name="grid_id", data=zeros(where10.size), index=where10)

        model = TransitionModel(hh_set, control_total_dataset=hct_set)
        model.run(year=2000, target_attribute_name="total_number_of_households", reset_dataset_attribute_value={'grid_id':-1})

        #check that there are indeed 33000 total households after running the model
        results = hh_set.size()
        should_be = [30000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the number of households in each group exactly match the control totals specified
        results = self.get_count_all_groups(hh_set)
        should_be = [4000, 5000, 1000, 3000, 0, 6000, 3000, 8000]
        self.assertEqual(ma.allclose(results, should_be),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

def create_households_for_estimation(agent_set, dbcon):
        estimation_set = HouseholdDataset(in_storage=StorageFactory().get_storage('mysql_storage', storage_location=dbcon),
                in_table_name="households_for_estimation")
        agent_set.unload_primary_attributes()
        agent_set.load_dataset(attributes='*')
        estimation_set.load_dataset(attributes=agent_set.get_primary_attribute_names())
        for attr in agent_set.get_attribute_names():
            agent_set.attribute_boxes[attr].set_data(concatenate((estimation_set.attribute_boxes[attr].get_data(), agent_set.attribute_boxes[attr].get_data())))
        agent_set._update_id_mapping()
        agent_set.update_size()
        return (agent_set, arange(estimation_set.size()))

def run_ALCM(niter):
    nhhs = 100
    ngcs = 10
    ngcs_attr = ngcs/2
    ngcs_noattr = ngcs - ngcs_attr
    hh_grid_ids = array(nhhs*[-1])
    
    storage = StorageFactory().get_storage('dict_storage')

    households_table_name = 'households'        
    storage.write_table(
        table_name = households_table_name,
        table_data = {
            'household_id': arange(nhhs)+1, 
            'grid_id': hh_grid_ids
            }
        )
        
    gridcells_table_name = 'gridcells'        
    storage.write_table(
        table_name = gridcells_table_name,
        table_data = {
            'grid_id': arange(ngcs)+1, 
            'cost':array(ngcs_attr*[100]+ngcs_noattr*[1000])
            }
        )

    households = HouseholdDataset(in_storage=storage, in_table_name=households_table_name)
    gridcells = GridcellDataset(in_storage=storage, in_table_name=gridcells_table_name)
    
    # create coefficients and specification
    coefficients = Coefficients(names=('costcoef', ), values=(-0.001,))
    specification = EquationSpecification(variables=('gridcell.cost', ), coefficients=('costcoef', ))
    logger.be_quiet()
    result = zeros((niter,ngcs))
    for iter in range(niter):
        hlcm = HouseholdLocationChoiceModelCreator().get_model(location_set=gridcells, compute_capacity_flag=False, 
                choices = 'opus_core.random_choices_from_index', 
                sampler=None,
                #sample_size_locations = 30
                )
        hlcm.run(specification, coefficients, agent_set=households, debuglevel=1,
                  chunk_specification={'nchunks':1})
        
        # get results
        gridcells.compute_variables(['urbansim.gridcell.number_of_households'],
            resources=Resources({'household':households}))
        result_more_attractive = gridcells.get_attribute_by_id('number_of_households', arange(ngcs_attr)+1)
        result_less_attractive = gridcells.get_attribute_by_id('number_of_households', arange(ngcs_attr+1, ngcs+1))
        households.set_values_of_one_attribute(attribute='grid_id', values=hh_grid_ids)
        gridcells.delete_one_attribute('number_of_households')
        result[iter,:] = concatenate((result_more_attractive, result_less_attractive))
        #print result #, result_more_attractive.sum(), result_less_attractive.sum()
    return result

    def test_controlling_with_one_marginal_characteristic(self):
        """Using the age_of_head as a marginal characteristic, which would partition the 8 groups into two larger groups
        (those with age_of_head < 40 and >= 40), ensure that the control totals are met and that the distribution within
        each large group is the same before and after running the model
        """

        #IMPORTANT: marginal characteristics grouping indices have to start at 0!
        #i.e. below, there is one marg. char. "age_of_head". here we indicate that the first "large group" (groups 1-4),
        #consisting of those groups with age_of_head < 40 should total 25000 households after running this model for one year,
        #and the second large group, those groups with age_of_head > 40, should total 15000 households
        annual_household_control_totals_data = {
            "year": array([2000, 2000]),
            "age_of_head": array([0,1]),
            "total_number_of_households": array([25000, 15000])
            }

        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name='hh_set', table_data=self.households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name='hct_set', table_data=annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what='household', id_name=['year' ,'age_of_head'])

        storage.write_table(table_name='hc_set', table_data=self.household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        storage.write_table(table_name='prs_set', table_data=self.person_data)
        prs_set = PersonDataset(in_storage=storage, in_table_name='prs_set')
        
        model = HouseholdTransitionModel()
        model.run(year=2000, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        #check that there are indeed 40000 total households after running the model
        results = hh_set.size()
        should_be = [40000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the total number of households within first four groups increased by 10000
        #and that the total number of households within last four groups decreased by 3000
        results = self.get_count_all_groups(hh_set)
        should_be = [25000, 15000]
        self.assertEqual(ma.allclose([sum(results[0:4]), sum(results[4:8])], should_be, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the distribution of households within groups 1-4 and 5-8 are the same before and after
        #running the model, respectively

        should_be = [6000.0/15000.0*25000.0, 2000.0/15000.0*25000.0, 3000.0/15000.0*25000.0, 4000.0/15000.0*25000.0,
                     2000.0/18000.0*15000.0, 5000.0/18000.0*15000.0, 3000.0/18000.0*15000.0, 8000.0/18000.0*15000.0]
        self.assertEqual(ma.allclose(results, should_be, rtol=0.05),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

    def test_unplaced_agents_decrease_available_space(self):
        """Using the household location choice model, create a set of available spaces and
        2000 unplaced agents (along with 5000 placed agents). Run the model, and check that
        the unplaced agents were placed, and the number of available spaces has decreased"""
        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name='households',
            table_data = {
                'grid_id': array(2000*[0] + 5000*[1]),
                'household_id': arange(7000)+1
                }
            )

        storage.write_table(table_name='gridcells',
            table_data= {
                'residential_units':array(50*[10000]),
                'grid_id':  arange(50)+1
                }
            )

        households = HouseholdDataset(in_storage=storage, in_table_name='households')
        gridcells = GridcellDataset(in_storage=storage, in_table_name='gridcells')

        coefficients = Coefficients(names=("dummy",), values=(0.1,))
        specification = EquationSpecification(variables=("gridcell.residential_units",), coefficients=("dummy",))

        """need to specify to the household location choice model exactly which households are moving,
        because by default it assumes all current households want to move, but in this test,
        the 5000 households already in gridcell #1 shouldn't move.
        here, we specify that only the unplaced households should be moved."""
        agents_index = where(households.get_attribute("grid_id") == 0)[0]

        hlcm = HouseholdLocationChoiceModelCreator().get_model(location_set=gridcells,
               choices = "opus_core.random_choices_from_index", sample_size_locations = 30)
        hlcm.run(specification, coefficients, agent_set=households, agents_index=agents_index, debuglevel=1)

        gridcells.compute_variables(["urbansim.gridcell.vacant_residential_units"],
                                    resources=Resources({"household":households}))
        vacancies = gridcells.get_attribute("vacant_residential_units")

        """since there were 5000 households already in gridcell #1, and gridcell #1 has
        10000 residential units, there should be no more than 5000 vacant residential units
        in gridcell #1 after running this model"""
        self.assertEqual(vacancies[0] <= 5000,
                         True, "Error: %d" % (vacancies[0],))
        """there should be exactly 430000 vacant residential units after the model run,
        because there were originally 50 gridcells with 10000 residential units each,
        and a total of 7000 units are occupied after the run"""
        self.assertEqual(sum(vacancies) == 50 * 10000 - 7000,
                         True, "Error: %d" % (sum(vacancies)))

    def test_same_distribution_after_household_subtraction(self):
        """Using the control_totals and no marginal characteristics,
        subtract households and ensure that the distribution within each group stays the same
        """
        annual_household_control_totals_data = {
            "year": array([2000, 2000]),
            "total_number_of_households": array([8000, 12000]),
             "faz_id": array([1,2])
            }

        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name = 'hh_set', table_data = self.households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name = 'hct_set', table_data = annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what="household")

        storage.write_table(table_name = 'hc_set', table_data = self.household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')
#        storage.write_table(table_name='prs_set', table_data=self.person_data)
#        prs_set = PersonDataset(in_storage=storage, in_table_name='prs_set')
        model = SubareaHouseholdTransitionModel(subarea_id_name="faz_id")
#        model.run(year=2000, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)
        model.run(year=2000, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        #check that there are indeed 8000 (area 1) and 12000 (area 2) total households after running the model
        areas = hh_set.get_attribute("faz_id")
        results = array([0,0])
        for iarea in [0,1]:
            results[iarea] = where(areas == [1,2][iarea])[0].size
        should_be = [8000, 12000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the distribution of households in each group is the same as before running the model
        results = self.get_count_all_groups(hh_set)
        should_be = array([# area 1 
                     3000.0/16500.0*8000.0, 1000.0/16500.0*8000.0, 1500.0/16500.0*8000.0, 2000.0/16500.0*8000.0,
                     1000.0/16500.0*8000.0, 2500.0/16500.0*8000.0, 1500.0/16500.0*8000.0, 4000.0/16500.0*8000.0,
                     # area 2
                     3000.0/16500.0*12000.0, 1000.0/16500.0*12000.0, 1500.0/16500.0*12000.0, 2000.0/16500.0*12000.0,
                     1000.0/16500.0*12000.0, 2500.0/16500.0*12000.0, 1500.0/16500.0*12000.0, 4000.0/16500.0*12000.0])
        self.assertEqual(ma.allclose(results, should_be, rtol=0.1),
                         True, "Error, should_be: %s,\n but result: %s" % (should_be, results))

    def test_same_distribution_after_household_addition(self):
        """Using the control_totals and no marginal characteristics,
        add households and ensure that the distribution within each group stays the same
        """

        annual_household_control_totals_data = {
            "year": array([2000, 2000]),
            "total_number_of_households": array([20000, 30000]),
            "large_area_id": array([1,2])
            }

        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name = 'hh_set', table_data = self.households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')
        
        storage.write_table(table_name = 'hct_set', table_data = annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what="household")

        storage.write_table(table_name = 'hc_set', table_data = self.household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        model = RegionalHouseholdTransitionModel()
        model.run(year=2000, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        #check that there are 20000 (area 1) and 30000 (area 2) total households after running the model
        areas = hh_set.get_attribute("large_area_id")
        results = array([0,0])
        for iarea in [0,1]:
            results[iarea] = where(areas == [1,2][iarea])[0].size
        should_be = [20000, 30000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the number of unplaced households is exactly the number of new households created
        results = where(hh_set.get_attribute("grid_id")<=0)[0].size
        should_be = [17000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the distribution of households in each group and each area is the same as before running the model
        results = self.get_count_all_groups(hh_set)
        should_be = array([
                    # area 1 
                     3000.0/16500.0*20000.0, 1000.0/16500.0*20000.0, 1500.0/16500.0*20000.0, 2000.0/16500.0*20000.0,
                     1000.0/16500.0*20000.0, 2500.0/16500.0*20000.0, 1500.0/16500.0*20000.0, 4000.0/16500.0*20000.0,
                     # area 2
                     3000.0/16500.0*30000.0, 1000.0/16500.0*30000.0, 1500.0/16500.0*30000.0, 2000.0/16500.0*30000.0,
                     1000.0/16500.0*30000.0, 2500.0/16500.0*30000.0, 1500.0/16500.0*30000.0, 4000.0/16500.0*30000.0])
        self.assertEqual(ma.allclose(results, should_be, rtol=0.1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))
        # check the types of the attributes
        self.assertEqual(hh_set.get_attribute("age_of_head").dtype, int32,
                         "Error in data type of the new household set. Should be: int32, is: %s" % str(hh_set.get_attribute("age_of_head").dtype))
        self.assertEqual(hh_set.get_attribute("income").dtype, int32,
                         "Error in data type of the new household set. Should be: int32, is: %s" % str(hh_set.get_attribute("income").dtype))
        self.assertEqual(hh_set.get_attribute("persons").dtype, int8,
                         "Error in data type of the new household set. Should be: int8, is: %s" % str(hh_set.get_attribute("persons").dtype))

    def test_same_distribution_after_household_addition(self):
        """Using the control_totals and no marginal characteristics,
        add households and ensure that the distribution within each group stays the same
        """

        annual_household_control_totals_data = {
            "year": array([2000]),
            "total_number_of_households": array([50000])
            }

        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name='hh_set', table_data=self.households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name='prs_set', table_data=self.person_data)
        prs_set = PersonDataset(in_storage=storage, in_table_name='prs_set')
        
        storage.write_table(table_name='hct_set', table_data=annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what="household", id_name="year")

        storage.write_table(table_name='hc_set', table_data=self.household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        model = HouseholdTransitionModel()
        model.run(year=2000, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        #check that there are indeed 50000 total households after running the model
        results = hh_set.size()
        should_be = [50000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the number of unplaced households is exactly the number of new households created
        results = where(hh_set.get_attribute("building_id")<=0)[0].size
        should_be = [17000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the distribution of households in each group is the same as before running the model
        results = self.get_count_all_groups(hh_set)
        should_be = array([6000.0/33000.0*50000.0, 2000.0/33000.0*50000.0, 3000.0/33000.0*50000.0, 4000.0/33000.0*50000.0,
                     2000.0/33000.0*50000.0, 5000.0/33000.0*50000.0, 3000.0/33000.0*50000.0, 8000.0/33000.0*50000.0])
        self.assertEqual(ma.allclose(results, should_be, rtol=0.05),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))
        # check the types of the attributes
        self.assertEqual(hh_set.get_attribute("age_of_head").dtype, int32,
                         "Error in data type of the new household set. Should be: int32, is: %s" % str(hh_set.get_attribute("age_of_head").dtype))
        self.assertEqual(hh_set.get_attribute("income").dtype, int32,
                         "Error in data type of the new household set. Should be: int32, is: %s" % str(hh_set.get_attribute("income").dtype))
        self.assertEqual(hh_set.get_attribute("persons").dtype, int8,
                         "Error in data type of the new household set. Should be: int8, is: %s" % str(hh_set.get_attribute("persons").dtype))

    def test_controlling_with_one_marginal_characteristic(self):
        """Using the age_of_head as a marginal characteristic, which would partition the 8 groups into two larger groups
        (those with age_of_head < 40 and >= 40), ensure that the control totals are met and that the distribution within
        each large group is the same before and after running the model
        """

        #IMPORTANT: marginal characteristics grouping indices have to start at 0!
        #i.e. below, there is one marg. char. "age_of_head". here we indicate that the first "large group" (groups 1-4),
        #consisting of those groups with age_of_head < 40 should total 25000 households after running this model for one year,
        #and the second large group, those groups with age_of_head > 40, should total 15000 households
        annual_household_control_totals_data = {
            "year": array([2000, 2000, 2000, 2000]),
            "age_of_head": array([0, 1, 0, 1]),
            "total_number_of_households": array([20000, 10000, 5000, 5000]),
            "faz_id": array([1, 1, 2, 2] )                                   
            }

        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name = 'hh_set', table_data = self.households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name = 'hct_set', table_data = annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what='household')

#        storage.write_table(table_name='prs_set', table_data=self.person_data)
#        prs_set = PersonDataset(in_storage=storage, in_table_name='prs_set')

        storage.write_table(table_name = 'hc_set', table_data = self.household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        model = SubareaHouseholdTransitionModel(subarea_id_name="faz_id")
#        model.run(year=2000, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)
        model.run(year=2000, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        #check that there are indeed 40000 total households after running the model
        areas = hh_set.get_attribute("faz_id")
        results = array([0,0])
        for iarea in [0,1]:
            results[iarea] = where(areas == [1,2][iarea])[0].size
        should_be = [30000, 10000]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        #check that the number of households within the groups correspond to the control totals
        results = self.get_count_all_groups(hh_set)
        should_be = [20000, 10000, 5000, 5000]
        idx1 = arange(0,4)
        idx2 = arange(4,8)
        idx3 = arange(8,12)
        idx4 = arange(12,16)
        self.assertEqual(ma.allclose([results[idx1].sum(), results[idx2].sum(), results[idx3].sum(), results[idx4].sum()], should_be, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, 
                                 array([results[idx1].sum(), results[idx2].sum(), results[idx3].sum(), results[idx4].sum()])))

        #check that the distribution of households within the groups are the same before and after
        #running the model, respectively

        should_be = [# area 1 
                     3000.0/7500.0*20000.0, 1000.0/7500.0*20000.0, 1500.0/7500.0*20000.0, 2000.0/7500.0*20000.0,
                     1000.0/9000.0*10000.0, 2500.0/9000.0*10000.0, 1500.0/9000.0*10000.0, 4000.0/9000.0*10000.0,
                     # area 2
                     3000.0/7500.0*5000.0, 1000.0/7500.0*5000.0, 1500.0/7500.0*5000.0, 2000.0/7500.0*5000.0,
                     1000.0/9000.0*5000.0, 2500.0/9000.0*5000.0, 1500.0/9000.0*5000.0, 4000.0/9000.0*5000.0]
        self.assertEqual(ma.allclose(results, should_be, rtol=0.1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

    def test_controlling_age_of_head(self):
        """ Controls for one marginal characteristics, namely age_of_head.
        """
        annual_household_control_totals_data = {
            "year": array([2000, 2000, 2000, 2001, 2001, 2001, 2002, 2002, 2002]),
            "age_of_head": array([0,1,2,0,1,2, 0,1,2]),
            "total_number_of_households": array([25013, 21513, 18227,  # 2000
                                                 10055, 15003, 17999, # 2001
                                                 15678, 14001, 20432]) # 2002
            }

        household_characteristics_for_ht_data = {
            "characteristic": array(3*['age_of_head']),
            "min": array([0, 35, 65]),
            "max": array([34, 64, -1])
            }

        households_data = {
            "household_id":arange(15000)+1,
            "building_id": array(15000*[1]),
            "age_of_head": array(1000*[25] + 1000*[28] + 2000*[32] + 1000*[34] +
                            2000*[35] + 1000*[40] + 1000*[54]+ 1000*[62] +
                            1000*[65] + 1000*[68] + 2000*[71] + 1000*[98]),
            "persons": array(1000*[2] + 2000*[3] + 1000*[1] + 1000*[6] + 1000*[1] + 1000*[4] +
                                3000*[1]+ 5000*[5], dtype=int8)
            }
        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name='hh_set', table_data=households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name='hct_set', table_data=annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what='household',
                                      id_name=['year' ,'age_of_head'])

        storage.write_table(table_name='hc_set', table_data=household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        storage.write_table(table_name='prs_set', table_data=self.person_data)
        prs_set = PersonDataset(in_storage=storage, in_table_name='prs_set')
        
        model = HouseholdTransitionModel(debuglevel=3)
        # this run should add households in all four categories
        model.run(year=2000, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        results = hh_set.size()
        should_be = [(hct_set.get_attribute("total_number_of_households")[0:3]).sum()]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        results = zeros(hc_set.size(), dtype=int32)
        results[0] = where(hh_set.get_attribute('age_of_head') <= hc_set.get_attribute("max")[0], 1,0).sum()
        for i in range(1, hc_set.size()-1):
            results[i] = logical_and(where(hh_set.get_attribute('age_of_head') >= hc_set.get_attribute("min")[i], 1,0),
                                 where(hh_set.get_attribute('age_of_head') <= hc_set.get_attribute("max")[i], 1,0)).sum()
        results[hc_set.size()-1] = where(hh_set.get_attribute('age_of_head') >= hc_set.get_attribute("min")[hc_set.size()-1], 1,0).sum()
        should_be = hct_set.get_attribute("total_number_of_households")[0:3]
        self.assertEqual(ma.allclose(results, should_be, rtol=1e-6),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        # this run should remove households in all four categories
        model.run(year=2001, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)
        results = hh_set.size()
        should_be = [(hct_set.get_attribute("total_number_of_households")[3:6]).sum()]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        results = zeros(hc_set.size(), dtype=int32)
        results[0] = where(hh_set.get_attribute('age_of_head') <= hc_set.get_attribute("max")[0], 1,0).sum()
        for i in range(1, hc_set.size()-1):
            results[i] = logical_and(where(hh_set.get_attribute('age_of_head') >= hc_set.get_attribute("min")[i], 1,0),
                                 where(hh_set.get_attribute('age_of_head') <= hc_set.get_attribute("max")[i], 1,0)).sum()
        results[hc_set.size()-1] = where(hh_set.get_attribute('age_of_head') >= hc_set.get_attribute("min")[hc_set.size()-1], 1,0).sum()
        should_be = hct_set.get_attribute("total_number_of_households")[3:6]
        self.assertEqual(ma.allclose(results, should_be, rtol=1e-6),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        # this run should add and remove households
        model.run(year=2002, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)
        results = hh_set.size()
        should_be = [(hct_set.get_attribute("total_number_of_households")[6:9]).sum()]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        results = zeros(hc_set.size(), dtype=int32)
        results[0] = where(hh_set.get_attribute('age_of_head') <= hc_set.get_attribute("max")[0], 1,0).sum()
        for i in range(1, hc_set.size()-1):
            results[i] = logical_and(where(hh_set.get_attribute('age_of_head') >= hc_set.get_attribute("min")[i], 1,0),
                                 where(hh_set.get_attribute('age_of_head') <= hc_set.get_attribute("max")[i], 1,0)).sum()
        results[hc_set.size()-1] = where(hh_set.get_attribute('age_of_head') >= hc_set.get_attribute("min")[hc_set.size()-1], 1,0).sum()
        should_be = hct_set.get_attribute("total_number_of_households")[6:9]
        self.assertEqual(ma.allclose(results, should_be, rtol=1e-6),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

    def test_controlling_income(self):
        """ Controls for one marginal characteristics, namely income.
        """
        annual_household_control_totals_data = {
            "year": array([2000, 2000, 2000, 2000, 2001, 2001, 2001, 2001, 2002, 2002, 2002, 2002]),
            "income": array([0,1,2,3,0,1,2,3, 0,1,2,3]),
            "total_number_of_households": array([25013, 21513, 18227, 18493, # 2000
                                                 10055, 15003, 17999, 17654, # 2001
                                                 15678, 14001, 20432, 14500]) # 2002
            }

        household_characteristics_for_ht_data = {
            "characteristic": array(4*['income']),
            "min": array([0, 40000, 120000, 70000]), # category 120000 has index 3 and category 70000 has index 2 
            "max": array([39999, 69999, -1, 119999]) # (testing row invariance)
            }
        hc_sorted_index = array([0,1,3,2])
        households_data = {
            "household_id":arange(20000)+1,
            "building_id": array(19950*[1] + 50*[0]),
            "income": array(1000*[1000] + 1000*[10000] + 2000*[20000] + 1000*[35000] + 2000*[45000] +
                                1000*[50000] + 2000*[67000]+ 2000*[90000] + 1000*[100005] + 2000*[110003] +
                                1000*[120000] + 1000*[200000] + 2000*[500000] + 1000*[630000]),
            "persons": array(3000*[2] + 2000*[3] + 1000*[1] + 1000*[6] + 1000*[1] + 1000*[4] +
                                3000*[1]+ 8000*[5], dtype=int8)
            }
        storage = StorageFactory().get_storage('dict_storage')

        storage.write_table(table_name='hh_set', table_data=households_data)
        hh_set = HouseholdDataset(in_storage=storage, in_table_name='hh_set')

        storage.write_table(table_name='hct_set', table_data=annual_household_control_totals_data)
        hct_set = ControlTotalDataset(in_storage=storage, in_table_name='hct_set', what='household', id_name=['year' ,'income'])

        storage.write_table(table_name='hc_set', table_data=household_characteristics_for_ht_data)
        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name='hc_set')

        storage.write_table(table_name='prs_set', table_data=self.person_data)
        prs_set = PersonDataset(in_storage=storage, in_table_name='prs_set')
        
        model = HouseholdTransitionModel(debuglevel=3)
        # this run should add households in all four categories
        model.run(year=2000, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)

        results = hh_set.size()
        should_be = [83246]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        results = zeros(hc_set.size(), dtype=int32)
        results[0] = where(hh_set.get_attribute('income') <= 
                                            hc_set.get_attribute("max")[hc_sorted_index[0]], 1,0).sum()
        for i in range(1, hc_set.size()-1):
            results[i] = logical_and(where(hh_set.get_attribute('income') >= 
                                           hc_set.get_attribute("min")[hc_sorted_index[i]], 1,0),
                                     where(hh_set.get_attribute('income') <= 
                                           hc_set.get_attribute("max")[hc_sorted_index[i]], 1,0)).sum()
        results[-1] = where(hh_set.get_attribute('income') >= hc_set.get_attribute("min")[hc_sorted_index[-1]], 1,0).sum()
        should_be = hct_set.get_attribute("total_number_of_households")[0:4]
        self.assertEqual(ma.allclose(results, should_be, rtol=1e-6),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        # this run should remove households in all four categories
        model.run(year=2001, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)
        results = hh_set.size()
        should_be = [(hct_set.get_attribute("total_number_of_households")[4:8]).sum()]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        results = zeros(hc_set.size(), dtype=int32)
        results[0] = where(hh_set.get_attribute('income') <= 
                                            hc_set.get_attribute("max")[hc_sorted_index[0]], 1,0).sum()
        for i in range(1, hc_set.size()-1):
            results[i] = logical_and(where(hh_set.get_attribute('income') >= 
                                           hc_set.get_attribute("min")[hc_sorted_index[i]], 1,0),
                                     where(hh_set.get_attribute('income') <= 
                                           hc_set.get_attribute("max")[hc_sorted_index[i]], 1,0)).sum()
        results[-1] = where(hh_set.get_attribute('income') >= hc_set.get_attribute("min")[hc_sorted_index[-1]], 1,0).sum()
        should_be = hct_set.get_attribute("total_number_of_households")[4:8]
        self.assertEqual(ma.allclose(results, should_be, rtol=1e-6),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        # this run should add and remove households
        model.run(year=2002, person_set=prs_set, household_set=hh_set, control_totals=hct_set, characteristics=hc_set)
        results = hh_set.size()
        should_be = [(hct_set.get_attribute("total_number_of_households")[8:13]).sum()]
        self.assertEqual(ma.allclose(should_be, results, rtol=1e-1),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

        results = zeros(hc_set.size(), dtype=int32)
        results[0] = where(hh_set.get_attribute('income') <= hc_set.get_attribute("max")[hc_sorted_index[0]], 1,0).sum()
        for i in range(1, hc_set.size()-1):
            results[i] = logical_and(where(hh_set.get_attribute('income') >= 
                                           hc_set.get_attribute("min")[hc_sorted_index[i]], 1,0),
                                     where(hh_set.get_attribute('income') <= 
                                           hc_set.get_attribute("max")[hc_sorted_index[i]], 1,0)).sum()
        results[-1] = where(hh_set.get_attribute('income') >= hc_set.get_attribute("min")[hc_sorted_index[-1]], 1,0).sum()
        should_be = hct_set.get_attribute("total_number_of_households")[8:13]
        self.assertEqual(ma.allclose(results, should_be, rtol=1e-6),
                         True, "Error, should_be: %s, but result: %s" % (should_be, results))

def run_HTM(niter):
        nhhs = 5000
        ngroups = 4
        nhhsg = int(nhhs/ngroups)
        nhhslg = nhhs-(ngroups-1)*nhhsg
        should_nhhs = nhhs-2000

        storage = StorageFactory().get_storage('dict_storage')

        hc_set_table_name = 'hc_set'        
        storage.write_table(
            table_name = hc_set_table_name,
            table_data = {
                'characteristic': array(4*['income']+4*['age_of_head']), 
                'min':array([0,1001,5001, 10001, 0, 31, 41, 61]), 
                'max':array([1000, 5000, 10000,-1, 30, 40, 60, -1])
                },
            )
            
        hct_set_table_name = 'hct_set'        
        storage.write_table(
            table_name = hct_set_table_name,
            table_data = {
                'year':array([2000]), 
                'total_number_of_households':array([should_nhhs])
                },
            )
            
        households_table_name = 'households'        
        storage.write_table(
            table_name = households_table_name,
            table_data = {
                'age_of_head': array(nhhsg/2*[18]+(nhhsg-nhhsg/2)*[35] +
                    nhhsg/2*[30] + (nhhsg-nhhsg/2)*[40] +
                    nhhsg/2*[38] + (nhhsg-nhhsg/2)*[65] + 
                    nhhslg/2*[50] + (nhhslg-nhhslg/2)*[80]
                    ),
                'income': array(nhhsg*[500] + nhhsg*[2000] + 
                    nhhsg*[7000] + nhhslg*[15000]
                    ),
                'household_id':arange(nhhs)+1
                },
            )

        hc_set = HouseholdCharacteristicDataset(in_storage=storage, in_table_name=hc_set_table_name)
        hct_set = ControlTotalDataset(
            in_storage = storage, 
            in_table_name = hct_set_table_name, 
            what = 'household', 
            id_name = ['year']
            )
          
        logger.be_quiet()
        result = zeros((niter,4))
        for iter in range(niter):
            households = HouseholdDataset(in_storage=storage, in_table_nam