def interpret (program, namespace):
"""
executes the program
"""
for command in program:
if command.parseinfo.rule == "assignment":
value = command.value
error = command.error
if value is not None:
value = parse_expr(value, namespace)
if error is not None:
error = parse_expr(error, namespace)
namespace[command.name] = assign (value, error, command.unit, command.name, command.longname)
elif command.parseinfo.rule == "multi_assignment":
# collect columns:
columns = {}
for columnIndex in range(len(command.header)):
values = []
for row in command.rows:
value = parse_expr(row[columnIndex], namespace)
values.append(value)
columns[command.header[columnIndex].name] = {
"header": command.header[columnIndex],
"values": values
}
# pair value columns with err-columns:
for name in columns:
if name.endswith("_err"):
continue
header = columns[name]["header"]
values = columns[name]["values"]
if name + "_err" in columns:
errorColumn = columns[name + "_err"]
if errorColumn["header"].error is not None:
raise RuntimeError("Variables with _err notation cannot use the <...> notation: %s"%errorColumn["header"].name)
if errorColumn["header"].longname is not None:
raise RuntimeError("Variables with _err notation cannot have a long name: %s"%errorColumn["header"].longname)
if header.error is not None:
raise RuntimeError("Variables with a corresponding _err column cannot have a general error specified: %s"%header.name)
namespace[name] = assign (values, errorColumn["values"], header.unit, name, header.longname, None, errorColumn["header"].unit)
else:
namespace[header.name] = assign (values, header.error, header.unit, header.name, header.longname)
elif command.parseinfo.rule == "python_code":
code = '\n'.join(command.code)
exec (code, None, namespace)
else:
raise RuntimeError("Unknown syntactic command type")
return namespace
def mean_value(self, quantity_to_assign, *quants, weighted=None, longname=None):
""" Calculates mean value of quantities and assigns it to new quantity
Args:
quantity_to_assign: name or quantity object of new mean value
quantities: one or more quantities names or objects of which mean value shall be calculated
weighted: if True, will weight mean value by errors (returns error if not possible)
if False, will not weight mean value by errors
if None, will try to weight mean value, but if at least one error is not given, will not weight it
longname: description for mean value quantity
"""
# get quantities
quantities_obj = []
for q in quants:
q_obj = quantities.parse_expr(q, self.data)
assert isinstance(q_obj, quantities.Quantity)
quantities_obj.append(q_obj)
if isinstance(quantity_to_assign, str):
name = quantity_to_assign
elif isinstance(quantity_to_assign, quantities.Quantity):
name = quantity_to_assign.name
quantity_to_assign = quantities.Quantity(name, longname)
self.data[name] = quantity_to_assign
# standard behaviour for "weighted"
if weighted is True:
force_weighted = True
else:
force_weighted = False
if weighted is None:
weighted = True
mean_value.mean_value(quantity_to_assign, quantities_obj, weighted=weighted, force_weighted=force_weighted)
def formula(self, quantity, adjust=True):
""" returns error formula of quantity as latex code
Args:
quantity: name of quantity or Quantity object
adjust: if True, replaces "_err" suffix by "\sigma" function and adds equals sign in front
Return:
latex code string of error formula
"""
quantity = quantities.parse_expr(quantity, self.data)
assert isinstance(quantity, quantities.Quantity)
if quantity.error_formula is None:
raise ValueError("quantity '%s' doesn't have an error formula.")
formula = quantity.error_formula
if isinstance(formula, str):
return formula
else:
# replace "_err" by sigma function
if adjust:
sigma = Function("\sigma")
for var in formula.free_symbols:
if var.name[-4:] == "_err":
formula = formula.subs(
var,
sigma(Symbol(var.name[:-4], **var._assumptions)))
return latex(sigma(quantity)) + " = " + latex(formula)
return formula
def formula(self, quantity, adjust=True):
""" returns error formula of quantity as latex code
Args:
quantity: name of quantity or Quantity object
adjust: if True, replaces "_err" suffix by "\sigma" function and adds equals sign in front
Return:
latex code string of error formula
"""
quantity = quantities.parse_expr(quantity, self.data)
assert isinstance(quantity, quantities.Quantity)
if quantity.error_formula is None:
raise ValueError("quantity '%s' doesn't have an error formula.")
formula = quantity.error_formula
if isinstance(formula,str):
return formula
else:
# replace "_err" by sigma function
if adjust:
sigma = Function("\sigma")
for var in formula.free_symbols:
if var.name[-4:] == "_err":
formula = formula.subs(var, sigma( Symbol(var.name[:-4], **var._assumptions)))
return latex(sigma(quantity)) + " = " + latex(formula)
return formula
def concat(self, new_name, *quants, longname=""):
""" concatenates quantities
Args:
new_name: name of new quantity
quants: quantities to be concatenated
"""
values = []
errors = []
dim = None
for q_str in quants:
q = quantities.parse_expr(q_str, self.data)
# check dimension
if dim is None:
dim = q.dim
else:
if not dim == q.dim:
raise RuntimeError("dimension mismatch\n%s != %s" %
(dim, q.dim))
# check if values or errors are None
if not values is None:
if q.value is None:
values = None
else:
v = q.value
if not isinstance(q.value, np.ndarray):
v = v.reshape((1))
values.append(v)
if not errors is None:
if q.error is None:
errors = None
else:
u = q.error
if not isinstance(q.error, np.ndarray):
u = u.reshape((1))
errors.append(u)
# concatenate
new_value = None
new_error = None
if not values is None:
new_value = np.concatenate(values)
if not errors is None:
new_error = np.concatenate(errors)
if new_value is None and new_error is None:
raise RuntimeError(
"Could not concatenate. At least one value and one error are None."
)
new_q = quantities.Quantity(new_name, longname)
new_q.value = new_value
new_q.error = new_error
new_q.dim = dim
self.data[new_name] = new_q
def concat(self, new_name, *quants, longname=""):
""" concatenates quantities
Args:
new_name: name of new quantity
quants: quantities to be concatenated
"""
values=[]
errors=[]
dim = None
for q_str in quants:
q = quantities.parse_expr(q_str, self.data)
# check dimension
if dim is None:
dim = q.dim
else:
if not dim==q.dim:
raise RuntimeError("dimension mismatch\n%s != %s" % (dim,q.dim))
# check if values or errors are None
if not values is None:
if q.value is None:
values = None
else:
v= q.value
if not isinstance(q.value,np.ndarray):
v = v.reshape((1))
values.append(v)
if not errors is None:
if q.error is None:
errors = None
else:
u = q.error
if not isinstance(q.error, np.ndarray):
u = u.reshape((1))
errors.append(u)
# concatenate
new_value = None
new_error = None
if not values is None:
new_value = np.concatenate(values)
if not errors is None:
new_error = np.concatenate(errors)
if new_value is None and new_error is None:
raise RuntimeError("Could not concatenate. At least one value and one error are None.")
new_q = quantities.Quantity(new_name, longname)
new_q.value = new_value
new_q.error = new_error
new_q.dim = dim
self.data[new_name] = new_q
def plot(self, *expr_pairs, save=None, xunit=None, yunit=None, xrange=None, yrange=None, ignore_dim=False):
""" Plots data or functions
Args:
expr_pairs: one or more pair of quantity on x-axis and on y-axis. e.g. ["p","V"]
y-axis can also be a function. e.g. ["t", "7*exp(t/t0)"]
save: string of file name without extension. if specified, plot will be saved to '<save>.png'
xunit: unit on x-axis. if not given, will find unit on its own
yunit: unit on y-axis. if not given, will find unit on its own
xrange: pair of x-axis range, e.g. [-5,10]
yrange: pair of y-axis range
ignore_dim: if True, will skip dimension check
"""
if len(expr_pairs) == 0:#
raise ValueError("nothing to plot specified.")
expr_pairs_obj = []
for expr_pair in expr_pairs:
# parse expressions
expr_pairs_obj.append( (quantities.parse_expr(expr_pair[0], self.data), quantities.parse_expr(expr_pair[1], self.data)) )
if not xunit is None:
xunit = units.parse_unit(xunit)[2]
if not yunit is None:
yunit = units.parse_unit(yunit)[2]
if not xrange is None:
xrange = [quantities.get_value(quantities.parse_expr(xrange[0], self.data)),
quantities.get_value(quantities.parse_expr(xrange[1], self.data))]
if not yrange is None:
yrange = [quantities.get_value(quantities.parse_expr(yrange[0], self.data)),
quantities.get_value(quantities.parse_expr(yrange[1], self.data))]
return plotting.plot(expr_pairs_obj, self.config, save=save, xunit=xunit, yunit=yunit, xrange=xrange, yrange=yrange, ignore_dim=ignore_dim)
def plot(self,
*expr_pairs,
save=None,
xunit=None,
yunit=None,
xrange=None,
yrange=None,
ignore_dim=False):
""" Plots data or functions
Args:
expr_pairs: one or more pair of quantity on x-axis and on y-axis. e.g. ["p","V"]
y-axis can also be a function. e.g. ["t", "7*exp(t/t0)"]
save: string of file name without extension. if specified, plot will be saved to '<save>.png'
xunit: unit on x-axis. if not given, will find unit on its own
yunit: unit on y-axis. if not given, will find unit on its own
xrange: pair of x-axis range, e.g. [-5,10]
yrange: pair of y-axis range
ignore_dim: if True, will skip dimension check
"""
if len(expr_pairs) == 0: #
raise ValueError("nothing to plot specified.")
expr_pairs_obj = []
for expr_pair in expr_pairs:
# parse expressions
expr_pairs_obj.append(
(quantities.parse_expr(expr_pair[0], self.data),
quantities.parse_expr(expr_pair[1], self.data)))
if not xunit is None:
xunit = units.parse_unit(xunit)[2]
if not yunit is None:
yunit = units.parse_unit(yunit)[2]
if not xrange is None:
xrange = [
quantities.get_value(
quantities.parse_expr(xrange[0], self.data)),
quantities.get_value(
quantities.parse_expr(xrange[1], self.data))
]
if not yrange is None:
yrange = [
quantities.get_value(
quantities.parse_expr(yrange[0], self.data)),
quantities.get_value(
quantities.parse_expr(yrange[1], self.data))
]
return plotting.plot(expr_pairs_obj,
self.config,
save=save,
xunit=xunit,
yunit=yunit,
xrange=xrange,
yrange=yrange,
ignore_dim=ignore_dim)
def mean_value(self,
quantity_to_assign,
*quants,
weighted=None,
longname=None):
""" Calculates mean value of quantities and assigns it to new quantity
Args:
quantity_to_assign: name or quantity object of new mean value
quantities: one or more quantities names or objects of which mean value shall be calculated
weighted: if True, will weight mean value by errors (returns error if not possible)
if False, will not weight mean value by errors
if None, will try to weight mean value, but if at least one error is not given, will not weight it
longname: description for mean value quantity
"""
# get quantities
quantities_obj = []
for q in quants:
q_obj = quantities.parse_expr(q, self.data)
assert isinstance(q_obj, quantities.Quantity)
quantities_obj.append(q_obj)
if isinstance(quantity_to_assign, str):
name = quantity_to_assign
elif isinstance(quantity_to_assign, quantities.Quantity):
name = quantity_to_assign.name
quantity_to_assign = quantities.Quantity(name, longname)
self.data[name] = quantity_to_assign
# standard behaviour for "weighted"
if weighted is True:
force_weighted = True
else:
force_weighted = False
if weighted is None:
weighted = True
mean_value.mean_value(quantity_to_assign,
quantities_obj,
weighted=weighted,
force_weighted=force_weighted)
def slice(self, new_name, quantity, start=0, end=None, longname=""):
""" creates new quantity from data set that only contains values from start to end
Args:
new_name: name of new quantity
quantity: name of quantity to be sliced
start: number of value in data set where new quantity is supposed to start
first value is 0
end: number of value to be the first one not taken into the new quantity
None to get all values until the end
longname: long name of new quantity
"""
q = quantities.parse_expr(quantity, self.data)
new_value = None
new_uncert = None
# check if values or uncerts are None
if not q.value is None:
if not isinstance(q.value, np.ndarray):
raise RuntimeError(
"Could not slice '%s'. It's not a data set." % quantity)
if end is None:
new_value = q.value[start:]
else:
new_value = q.value[start:end]
if not q.uncert is None:
if not isinstance(q.value, np.ndarray):
raise RuntimeError(
"Could not slice '%s'. Uncertainty is not an array." %
quantity)
if end is None:
new_uncert = q.uncert[start:]
else:
new_uncert = q.uncert[start:end]
new_q = quantities.Quantity(new_name, longname)
new_q.value = new_value
new_q.uncert = new_uncert
new_q.dim = q.dim
self.data[new_name] = new_q
def slice(self, new_name, quantity, start=0, end=None, longname=""):
""" creates new quantity from data set that only contains values from start to end
Args:
new_name: name of new quantity
quantity: name of quantity to be sliced
start: number of value in data set where new quantity is supposed to start
first value is 0
end: number of value to be the first one not taken into the new quantity
None to get all values until the end
longname: long name of new quantity
"""
q = quantities.parse_expr(quantity, self.data)
new_value = None
new_uncert = None
# check if values or uncerts are None
if not q.value is None:
if not isinstance(q.value, np.ndarray):
raise RuntimeError("Could not slice '%s'. It's not a data set." % quantity)
if end is None:
new_value = q.value[start:]
else:
new_value = q.value[start:end]
if not q.uncert is None:
if not isinstance(q.value, np.ndarray):
raise RuntimeError("Could not slice '%s'. Uncertainty is not an array." % quantity)
if end is None:
new_uncert = q.uncert[start:]
else:
new_uncert = q.uncert[start:end]
new_q = quantities.Quantity(new_name, longname)
new_q.value = new_value
new_q.uncert = new_uncert
new_q.dim = q.dim
self.data[new_name] = new_q
def interpret(program, namespace):
"""
executes the program
"""
for command in program:
if command.parseinfo.rule == "assignment":
value = command.value
error = command.error
if value is not None:
value = parse_expr(value, namespace)
if error is not None:
error = parse_expr(error, namespace)
namespace[command.name] = assign(value, error, command.unit,
command.name, command.longname)
elif command.parseinfo.rule == "multi_assignment":
# collect columns:
columns = {}
for columnIndex in range(len(command.header)):
values = []
for row in command.rows:
value = parse_expr(row[columnIndex], namespace)
values.append(value)
columns[command.header[columnIndex].name] = {
"header": command.header[columnIndex],
"values": values
}
# pair value columns with err-columns:
for name in columns:
if name.endswith("_err"):
continue
header = columns[name]["header"]
values = columns[name]["values"]
if name + "_err" in columns:
errorColumn = columns[name + "_err"]
if errorColumn["header"].error is not None:
raise RuntimeError(
"Variables with _err notation cannot use the <...> notation: %s"
% errorColumn["header"].name)
if errorColumn["header"].longname is not None:
raise RuntimeError(
"Variables with _err notation cannot have a long name: %s"
% errorColumn["header"].longname)
if header.error is not None:
raise RuntimeError(
"Variables with a corresponding _err column cannot have a general error specified: %s"
% header.name)
namespace[name] = assign(values, errorColumn["values"],
header.unit, name,
header.longname, None,
errorColumn["header"].unit)
else:
namespace[header.name] = assign(values, header.error,
header.unit, header.name,
header.longname)
elif command.parseinfo.rule == "python_code":
code = '\n'.join(command.code)
exec(code, None, namespace)
else:
raise RuntimeError("Unknown syntactic command type")
return namespace
def __getitem__(self, qname):
return quantities.parse_expr(qname, self.data)
def assign(self, name, value=None, error=None, unit=None, longname=None, value_unit=None, error_unit=None, replace=False, ignore_dim=False):
""" Assigns value and/or error to quantity
Args:
name: quantity name
longname: description of quantity
value: value to assign, can be expression, string, list or number
error: error to assign, can be expression, string, list or number, but mustn't depend on other quantities
unit: unit of both value and error, replaces 'value_unit' and 'error_unit' if given
value_unit: value unit expression or string
error_unit: error unit expression or string
replace: if True, will replace quantity instead of trying to keep data
ignore_dim: if True, will ignore calculated dimension and use given unit instead
"""
if not unit is None:
value_unit = unit
error_unit = unit
if value is None and error is None:
raise ValueError("At least either value or error must be specified.")
value_len = None
value_dim = None
value_formula = None
error_len = None
error_dim = None
error_formula = None
# if value is given
if not value is None:
# parse unit if given
if not value_unit is None:
factor, value_dim, value_unit = units.parse_unit(value_unit)
# parse value
if isinstance(value, list) or isinstance(value, tuple):
# if it's a list, parse each element
parsed_list = []
for v in value:
parsed_list.append(quantities.parse_expr(v, self.data))
elif isinstance(value, str) or isinstance(value, Expr):
# if it's not a list, parse once
value = quantities.parse_expr(value, self.data)
# if it's a calculation
if isinstance(value, Expr) and not value.is_number:
# calculate value from dependency
value_formula = value
value = quantities.get_value(value_formula)
# calculate dimension from dependency
if not ignore_dim:
calculated_dim = quantities.get_dimension(value_formula)
if not value_dim is None and not calculated_dim == value_dim:
raise RuntimeError("dimension mismatch for '%s'\n%s != %s" % (name, value_dim, calculated_dim))
elif value_dim is None:
value_dim = calculated_dim
else:
# if ignore_dim is True and there's no unit given -> dimensionless
if value_dim is None:
factor=1
value_dim = Dimension()
value_unit = S.One
# calculated value must be converted to given unit (ignore_dim=True)
value = np.float_(factor)*value
# if it's a number
else:
# if no unit given, set dimensionless
if value_unit is None:
factor = 1
value_dim = Dimension()
value_unit = S.One
value=np.float_(factor)*np.float_(value)
# calculate value length
if isinstance(value,np.ndarray):
value_len = len(value)
else:
value_len = 1
# if error is given
if not error is None:
# parse unit if given
if not error_unit is None:
factor, error_dim, error_unit = units.parse_unit(error_unit)
# parse value
if isinstance(error, list) or isinstance(error, tuple):
# if it's a list, parse each element
parsed_list = []
for u in error:
parsed_list.append(quantities.parse_expr(u, self.data))
elif isinstance(error, str) or isinstance(error, Expr):
# if it's not a list, parse once
error = quantities.parse_expr(error, self.data)
# make sure error is a number
if isinstance(error, Expr) and not error.is_number:
raise RuntimeError("error '%s' is not a number" % error)
# if no unit given, set dimensionless
if error_unit is None:
factor = 1
error_dim = Dimension()
error_unit = S.One
error=np.float_(factor)*np.float_(error)
# calculate error length, ignore len(error)==1 because it can be duplicated to fit any value length
if isinstance(error,np.ndarray):
error_len = len(error)
# if error can be calculated
elif not value_formula is None:
error, error_formula = quantities.get_error(value_formula)
# merge dimensions
dim = value_dim
if not dim is None and not error_dim is None and not dim == error_dim:
raise RuntimeError("value dimension and error dimension are not the same\n%s != %s" % (dim, error_dim))
if not error_dim is None:
dim = error_dim
# merge lengths
new_len = value_len
if not new_len is None and not error_len is None and not new_len == error_len:
raise RuntimeError("value length doesn't fit error length for '%s':\n%s != %s" % (name, new_len, error_len))
if not error_len is None:
new_len = error_len
# if quantity didn't exist
if not name in self.data or replace:
self.data[name] = quantities.Quantity(name)
# if it did exist
else:
# get old length, len(error)=1 is not a length, because it can be duplicated to fit any value length
old_len = None
if not self.data[name].value is None:
if isinstance(self.data[name].value, np.ndarray):
old_len = len(self.data[name].value)
else:
old_len = 1
if not self.data[name].error is None and isinstance(self.data[name].error, np.ndarray):
old_len = len(self.data[name].error)
# if new dimension or new length, create new quantity
if (not self.data[name].dim == dim or
(not old_len is None and not new_len is None and not old_len == new_len)):
self.data[name] = quantities.Quantity(name)
# save stuff
if not longname is None:
self.data[name].longname = longname
if not value is None:
self.data[name].value = value
self.data[name].value_formula = value_formula
if not value_unit is None:
self.data[name].prefer_unit = value_unit
elif not error_unit is None:
self.data[name].prefer_unit = error_unit
if not error is None:
self.data[name].error = error
self.data[name].error_formula = error_formula
self.data[name].dim = dim
# check if error must be duplicated to adjust to value length
if isinstance(self.data[name].value, np.ndarray) and isinstance(self.data[name].error, np.float_):
error_arr = np.full(len(self.data[name].value),self.data[name].error)
self.data[name].error = error_arr
def fit(self, fit_function, xydata, parameters, weighted=None, plot=False, ignore_dim=False):
""" fits function to data
Args:
fit_function: function to fit, e.g. "n*t**2 + m*t + b"
xydata: pair of x-quantity and y-quantity of data to fit to, e.g. ["t","U"]
parameters: list of parameters in fit function, e.g. ["n","m","b"]
weighted: if True, will weight fit by errors (returns error if not possible)
if False, will not weight fit by errors
if None, will try to weight fit, but if at least one error is not given, will not weight it
plot: Bool, if data and fit function should be plotted
ignore_dim: if True, will ignore dimensions and just calculate in base units instead
"""
if self.config["fit_module"] == "scipy":
import errorpro.fit_scipy as fit_module
else:
raise ValueError("no fit module called '%s'." % self.config["fit_module"])
# get parameter quantities
parameters_obj = []
for p in parameters:
if isinstance(p, str):
if not p in self.data:
self.data[p] = quantities.Quantity(p)
self.data[p].dim = Dimension()
parameters_obj.append(self.data[p])
elif isinstance(p, quantities.Quantity):
parameters_obj.append(p)
else:
raise TypeError("parameters can only be strings or Quantity objects")
# parse fit function
fit_function = quantities.parse_expr(fit_function, self.data)
# get data quantities
x_data = quantities.parse_expr(xydata[0], self.data)
# if x-data is an expression
if not isinstance(x_data, quantities.Quantity):
dummy = quantities.Quantity()
fit_function = fit_function.subs(x_data,dummy)
dummy.value = quantities.get_value(x_data)
dummy.error = quantities.get_error(x_data)[0]
dummy.dim = quantities.get_dimension(x_data)
x_data = dummy
y_data = quantities.parse_expr(xydata[1], self.data)
# if y-data is an expression
if not isinstance(y_data, quantities.Quantity):
dummy = quantities.Quantity()
dummy.value = quantities.get_value(y_data)
dummy.error = quantities.get_error(y_data)[0]
dummy.dim = quantities.get_dimension(y_data)
y_data = dummy
# check if dimension fits
if not ignore_dim:
try:
dim_func = quantities.get_dimension(fit_function)
except ValueError:
dim_func = None
if not dim_func == y_data.dim:
# try to solve for dimensionless parameters
known_dimensions = {x_data.name: x_data.dim}
known_dimensions = dim_solve(fit_function, y_data.dim, known_dimensions)
for q_name in known_dimensions:
if q_name in self.data:
if not self.data[q_name].dim == known_dimensions[q_name]:
self.data[q_name].dim = known_dimensions[q_name]
self.data[q_name].prefer_unit = None
dim_func = quantities.get_dimension(fit_function)
# if it still doesn't work, raise error
if not dim_func == y_data.dim:
raise RuntimeError("Finding dimensions of fit parameters was not sucessful.\n"\
"Check fit function or specify parameter units manually.\n"\
"This error will occur until dimensions are right.")
# fit
values, errors = fit_module.fit(x_data, y_data, fit_function, parameters_obj, weighted)
# save results
i = 0
for p in parameters_obj:
p.value = values[i]
p.value_formula = "fit"
p.error = errors[i]
p.error_formula = "fit"
i += 1
# plot
if plot:
return plotting.plot([(x_data, y_data), (x_data, fit_function)], self.config, ignore_dim=ignore_dim)
else:
return self.table(*parameters_obj)
def __getitem__(self, qname):
return quantities.parse_expr(qname, self.data)
def assign(self,
name,
value=None,
error=None,
unit=None,
longname=None,
value_unit=None,
error_unit=None,
replace=False,
ignore_dim=False):
""" Assigns value and/or error to quantity
Args:
name: quantity name
longname: description of quantity
value: value to assign, can be expression, string, list or number
error: error to assign, can be expression, string, list or number, but mustn't depend on other quantities
unit: unit of both value and error, replaces 'value_unit' and 'error_unit' if given
value_unit: value unit expression or string
error_unit: error unit expression or string
replace: if True, will replace quantity instead of trying to keep data
ignore_dim: if True, will ignore calculated dimension and use given unit instead
"""
if not unit is None:
value_unit = unit
error_unit = unit
if value is None and error is None:
raise ValueError(
"At least either value or error must be specified.")
value_len = None
value_dim = None
value_formula = None
error_len = None
error_dim = None
error_formula = None
# if value is given
if not value is None:
# parse unit if given
if not value_unit is None:
factor, value_dim, value_unit = units.parse_unit(value_unit)
# parse value
if isinstance(value, list) or isinstance(value, tuple):
# if it's a list, parse each element
parsed_list = []
for v in value:
parsed_list.append(quantities.parse_expr(v, self.data))
elif isinstance(value, str) or isinstance(value, Expr):
# if it's not a list, parse once
value = quantities.parse_expr(value, self.data)
# if it's a calculation
if isinstance(value, Expr) and not value.is_number:
# calculate value from dependency
value_formula = value
value = quantities.get_value(value_formula)
# calculate dimension from dependency
if not ignore_dim:
calculated_dim = quantities.get_dimension(value_formula)
if not value_dim is None and not calculated_dim == value_dim:
raise RuntimeError(
"dimension mismatch for '%s'\n%s != %s" %
(name, value_dim, calculated_dim))
elif value_dim is None:
value_dim = calculated_dim
else:
# if ignore_dim is True and there's no unit given -> dimensionless
if value_dim is None:
factor = 1
value_dim = Dimension()
value_unit = S.One
# calculated value must be converted to given unit (ignore_dim=True)
value = np.float_(factor) * value
# if it's a number
else:
# if no unit given, set dimensionless
if value_unit is None:
factor = 1
value_dim = Dimension()
value_unit = S.One
value = np.float_(factor) * np.float_(value)
# calculate value length
if isinstance(value, np.ndarray):
value_len = len(value)
else:
value_len = 1
# if error is given
if not error is None:
# parse unit if given
if not error_unit is None:
factor, error_dim, error_unit = units.parse_unit(error_unit)
# parse value
if isinstance(error, list) or isinstance(error, tuple):
# if it's a list, parse each element
parsed_list = []
for u in error:
parsed_list.append(quantities.parse_expr(u, self.data))
elif isinstance(error, str) or isinstance(error, Expr):
# if it's not a list, parse once
error = quantities.parse_expr(error, self.data)
# make sure error is a number
if isinstance(error, Expr) and not error.is_number:
raise RuntimeError("error '%s' is not a number" % error)
# if no unit given, set dimensionless
if error_unit is None:
factor = 1
error_dim = Dimension()
error_unit = S.One
error = np.float_(factor) * np.float_(error)
# calculate error length, ignore len(error)==1 because it can be duplicated to fit any value length
if isinstance(error, np.ndarray):
error_len = len(error)
# if error can be calculated
elif not value_formula is None:
error, error_formula = quantities.get_error(value_formula)
# merge dimensions
dim = value_dim
if not dim is None and not error_dim is None and not dim == error_dim:
raise RuntimeError(
"value dimension and error dimension are not the same\n%s != %s"
% (dim, error_dim))
if not error_dim is None:
dim = error_dim
# merge lengths
new_len = value_len
if not new_len is None and not error_len is None and not new_len == error_len:
raise RuntimeError(
"value length doesn't fit error length for '%s':\n%s != %s" %
(name, new_len, error_len))
if not error_len is None:
new_len = error_len
# if quantity didn't exist
if not name in self.data or replace:
self.data[name] = quantities.Quantity(name)
# if it did exist
else:
# get old length, len(error)=1 is not a length, because it can be duplicated to fit any value length
old_len = None
if not self.data[name].value is None:
if isinstance(self.data[name].value, np.ndarray):
old_len = len(self.data[name].value)
else:
old_len = 1
if not self.data[name].error is None and isinstance(
self.data[name].error, np.ndarray):
old_len = len(self.data[name].error)
# if new dimension or new length, create new quantity
if (not self.data[name].dim == dim
or (not old_len is None and not new_len is None
and not old_len == new_len)):
self.data[name] = quantities.Quantity(name)
# save stuff
if not longname is None:
self.data[name].longname = longname
if not value is None:
self.data[name].value = value
self.data[name].value_formula = value_formula
if not value_unit is None:
self.data[name].prefer_unit = value_unit
elif not error_unit is None:
self.data[name].prefer_unit = error_unit
if not error is None:
self.data[name].error = error
self.data[name].error_formula = error_formula
self.data[name].dim = dim
# check if error must be duplicated to adjust to value length
if isinstance(self.data[name].value, np.ndarray) and isinstance(
self.data[name].error, np.float_):
error_arr = np.full(len(self.data[name].value),
self.data[name].error)
self.data[name].error = error_arr
def fit(self,
fit_function,
xydata,
parameters,
weighted=None,
plot=False,
ignore_dim=False):
""" fits function to data
Args:
fit_function: function to fit, e.g. "n*t**2 + m*t + b"
xydata: pair of x-quantity and y-quantity of data to fit to, e.g. ["t","U"]
parameters: list of parameters in fit function, e.g. ["n","m","b"]
weighted: if True, will weight fit by errors (returns error if not possible)
if False, will not weight fit by errors
if None, will try to weight fit, but if at least one error is not given, will not weight it
plot: Bool, if data and fit function should be plotted
ignore_dim: if True, will ignore dimensions and just calculate in base units instead
"""
if self.config["fit_module"] == "scipy":
import errorpro.fit_scipy as fit_module
else:
raise ValueError("no fit module called '%s'." %
self.config["fit_module"])
# get parameter quantities
parameters_obj = []
for p in parameters:
if isinstance(p, str):
if not p in self.data:
self.data[p] = quantities.Quantity(p)
self.data[p].dim = Dimension()
parameters_obj.append(self.data[p])
elif isinstance(p, quantities.Quantity):
parameters_obj.append(p)
else:
raise TypeError(
"parameters can only be strings or Quantity objects")
# parse fit function
fit_function = quantities.parse_expr(fit_function, self.data)
# get data quantities
x_data = quantities.parse_expr(xydata[0], self.data)
# if x-data is an expression
if not isinstance(x_data, quantities.Quantity):
dummy = quantities.Quantity()
fit_function = fit_function.subs(x_data, dummy)
dummy.value = quantities.get_value(x_data)
dummy.error = quantities.get_error(x_data)[0]
dummy.dim = quantities.get_dimension(x_data)
x_data = dummy
y_data = quantities.parse_expr(xydata[1], self.data)
# if y-data is an expression
if not isinstance(y_data, quantities.Quantity):
dummy = quantities.Quantity()
dummy.value = quantities.get_value(y_data)
dummy.error = quantities.get_error(y_data)[0]
dummy.dim = quantities.get_dimension(y_data)
y_data = dummy
# check if dimension fits
if not ignore_dim:
try:
dim_func = quantities.get_dimension(fit_function)
except ValueError:
dim_func = None
if not dim_func == y_data.dim:
# try to solve for dimensionless parameters
known_dimensions = {x_data.name: x_data.dim}
known_dimensions = dim_solve(fit_function, y_data.dim,
known_dimensions)
for q_name in known_dimensions:
if q_name in self.data:
if not self.data[q_name].dim == known_dimensions[
q_name]:
self.data[q_name].dim = known_dimensions[q_name]
self.data[q_name].prefer_unit = None
dim_func = quantities.get_dimension(fit_function)
# if it still doesn't work, raise error
if not dim_func == y_data.dim:
raise RuntimeError("Finding dimensions of fit parameters was not sucessful.\n"\
"Check fit function or specify parameter units manually.\n"\
"This error will occur until dimensions are right.")
# fit
values, errors = fit_module.fit(x_data, y_data, fit_function,
parameters_obj, weighted)
# save results
i = 0
for p in parameters_obj:
p.value = values[i]
p.value_formula = "fit"
p.error = errors[i]
p.error_formula = "fit"
i += 1
# plot
if plot:
return plotting.plot([(x_data, y_data), (x_data, fit_function)],
self.config,
ignore_dim=ignore_dim)
else:
return self.table(*parameters_obj)