def test_reporter_from_dantzig(test_mp, test_data_path):
scen = make_dantzig(test_mp, solve=test_data_path)
# Reporter.from_scenario can handle the Dantzig problem
rep = Reporter.from_scenario(scen)
# Partial sums are available automatically (d is defined over i and j)
d_i = rep.get('d:i')
# Units pass through summation
assert d_i.attrs['_unit'] == UNITS.parse_units('km')
# Summation across all dimensions results a 1-element Quantity
d = rep.get('d:')
assert len(d) == 1
assert np.isclose(d.iloc[0], 11.7)
# Weighted sum
weights = Quantity(
xr.DataArray([1, 2, 3],
coords=['chicago new-york topeka'.split()],
dims=['j']))
new_key = rep.aggregate('d:i-j', 'weighted', 'j', weights)
# ...produces the expected new key with the summed dimension removed and
# tag added
assert new_key == 'd:i:weighted'
# ...produces the expected new value
obs = rep.get(new_key)
exp = (rep.get('d:i-j') * weights).sum(dim=['j']) / weights.sum(dim=['j'])
# TODO: attrs has to be explicitly copied here because math is done which
# returns a pd.Series
exp = Quantity(exp, attrs=rep.get('d:i-j').attrs)
assert_series_equal(obs.sort_index(), exp.sort_index())
# Disaggregation with explicit data
# (cases of canned food 'p'acked in oil or water)
shares = xr.DataArray([0.8, 0.2], coords=[['oil', 'water']], dims=['p'])
new_key = rep.disaggregate('b:j', 'p', args=[Quantity(shares)])
# ...produces the expected key with new dimension added
assert new_key == 'b:j-p'
b_jp = rep.get('b:j-p')
# Units pass through disaggregation
assert b_jp.attrs['_unit'] == 'cases'
# Set elements are available
assert rep.get('j') == ['new-york', 'chicago', 'topeka']
# 'all' key retrieves all quantities
obs = {da.name for da in rep.get('all')}
exp = set('a b d f demand demand-margin z x'.split())
assert obs == exp
# Shorthand for retrieving a full key name
assert rep.full_key('d') == 'd:i-j' and isinstance(rep.full_key('d'), Key)
def test_reporting_units():
"""Test handling of units within Reporter computations."""
r = Reporter()
# Create some dummy data
dims = dict(coords=['a b c'.split()], dims=['x'])
r.add('energy:x',
Quantity(xr.DataArray([1., 3, 8], attrs={'_unit': 'MJ'}, **dims)))
r.add('time',
Quantity(xr.DataArray([5., 6, 8], attrs={'_unit': 'hour'}, **dims)))
r.add('efficiency', Quantity(xr.DataArray([0.9, 0.8, 0.95], **dims)))
# Aggregation preserves units
r.add('energy', (computations.sum, 'energy:x', None, ['x']))
assert r.get('energy').attrs['_unit'] == UNITS.parse_units('MJ')
# Units are derived for a ratio of two quantities
r.add('power', (computations.ratio, 'energy:x', 'time'))
assert r.get('power').attrs['_unit'] == UNITS.parse_units('MJ/hour')
# Product of dimensioned and dimensionless quantities keeps the former
r.add('energy2', (computations.product, 'energy:x', 'efficiency'))
assert r.get('energy2').attrs['_unit'] == UNITS.parse_units('MJ')
def test_reporter_add_product(test_mp):
scen = ixmp.Scenario(test_mp, 'reporter_add_product',
'reporter_add_product', 'new')
*_, x = add_test_data(scen)
rep = Reporter.from_scenario(scen)
# add_product() works
key = rep.add_product('x squared', 'x', 'x', sums=True)
# Product has the expected dimensions
assert key == 'x squared:t-y'
# Product has the expected value
exp = as_quantity(x * x)
exp.attrs['_unit'] = UNITS('kilogram ** 2').units
assert_qty_equal(exp, rep.get(key))