def assign(value, error=None, unit=None, name=None, longname=None, value_unit=None, error_unit=None, ignore_dim=False):
""" function to create a new quantity
Args:
value: number or string that can be parsed by numpy, or sympy
expression. If it's a sympy expression containing quantities, it
will perform the calculation, otherwise it just saves the value.
error: number that is saved as the value's uncertainty. this will replace
any error coming from a calculation.
unit: sympy expression of Unit objects. This is used to convert and save
value and error in base units. Replaces value_unit and error_unit if
specified.
name: short name of the quantity (usually one letter). If not specified,
quantity will get a dummy name.
longname: optional additional description of the quantity
value_unit: unit of value. Use this if value and error have different units.
error_unit: unit of error.
ignore_dim: bool. Keeps function from raising an error even if calculated
and given unit don't match. Then given unit is used instead.
"""
value_formula = None
value_factor = 1
value_dim = Dimension()
error_formula = None
error_factor = 1
error_dim = Dimension()
# parse units
if unit is not None:
# if one general unit is given
value_factor, value_dim, value_unit = parse_unit(unit)
error_factor = value_factor
error_dim = value_dim
error_unit = value_unit
else:
# if value unit is given
if value_unit is not None:
value_factor, value_dim, value_unit = parse_unit(value_unit)
# if error unit is given
if error_unit is not None:
error_factor, error_dim, error_unit = parse_unit(error_unit)
# check dimension consistency between value_dim and error_dim
if value_unit is not None and not value_dim == error_dim:
raise RuntimeError("dimension mismatch\n%s != %s" % (value_dim, error_dim))
# process value
# if it's a calculation
if isinstance(value, Expr) and not value.is_number:
value_formula = value
value = get_value(value_formula)
if ignore_dim:
# with ignore_dim=True, calculated value is converted to given unit
value = np.float_(value_factor)*np.float_(value)
else:
# calculate dimension from dependency
calculated_dim = get_dimension(value_formula)
if value_unit is None:
value_dim = calculated_dim
else:
if not calculated_dim == value_dim:
raise RuntimeError("dimension mismatch \n%s != %s" % (value_dim, calculated_dim))
# if it's a number
else:
value=np.float_(value_factor)*np.float_(value)
# process error
if error is not None:
error=np.float_(error_factor)*np.float_(error)
# check value and error shapes and duplicate error in case
if error.shape == () or value.shape[-len(error.shape):] == error.shape:
error = np.resize(error, value.shape)
else:
raise RuntimeError("length of value and error don't match and "\
"can't be adjusted by duplicating.\n"\
"%s and %s" % (value.shape, error.shape))
# if error can be calculated
elif value_formula is not None:
error, error_formula = get_error(value_formula)
if ignore_dim:
# with ignore_dim=True, calculated error is converted to given unit
error = np.float_(error_factor)*np.float_(error)
q = Quantity(name, longname)
q.value = value
q.value_formula = value_formula
q.error = error
q.error_formula = error_formula
if value_unit is not None:
q.prefer_unit = value_unit
else:
q.prefer_unit = error_unit
q.dim = value_dim
return q
