Adding a new observable

An observable is an experimentally measurable and theoretically computable quantity that can optionally depend on one or more parameters (e.g. kinematic quantities like $q^2$). Defining a new observable in flavio requires

• defining a function that returns the theory prediction as a function of parameters and Wilson coefficients
• creating an instance of Observable (naming the observable) and Prediction (linking the name to the funtion.)

Toy example

For illustration, imagine we want to define an observable that just coincides with the $B^0$ meson mass (which is set by the parameter m_B0).

from flavio.classes import Observable, Prediction

def my_fct(wc_obj, par):
    return par['m_B0']

Observable('m_B0-obs')
Prediction('m_B0-obs', my_fct)

where m_B0-obs is the string name we gave to the observable. Note how the function always has at least two arguments: the first is an instance of flavio.WilsonCoefficients, the second a parameter dictionary. Of course this example is not very meaningful. In practice, the function will contain all the (possibly complicated) physics and might be split to a separate file (or several files).

Adding metadata

While the code will work without it, metadata like a TeX name and description should be added to ease documentation. Extending the toy example above, this could be done via

my_obs = Observable('m_B0-obs')
my_obs.tex = r"$m_{B^0}$"
my_obs.description = r"Mass of the $B^0$ meson."

Note that when using TeX code in strings, you should enclose the string in r"...", so backslashes and other characters are interpreted literally and not as escape characters.

Observables depending on arguments

When the observable depends on arguments (rather than being just a single number), these arguments must appear in the prediction function and they have to be passed to the observable instance. A toy example with an observable depending on the arguments a and b:

def my_fct(wc_obj, par, a, b):
    ...
    return ...

Observable(name='my_observable', arguments=['a', 'b'])
Prediction('my_observable', my_fct)