Easy conversions between different styles of expressions. Formulate currently supports converting between ROOT and numexpr style expressions.
Install formulate like any other Python package:
pip install --user formulateor similar (use `sudo, `virtualenv, or `conda` if you wish).
$ python -m formulate --from-root '(A && B) || TMath::Sqrt(A)' --to-numexpr
(A & B) | sqrt(A)
$ python -m formulate --from-numexpr '(A & B) | sqrt(A)' --to-root
(A && B) || TMath::Sqrt(A)
$ python -m formulate --from-root '(A && B) || TMath::Sqrt(1.23) * e**1.2 + 5*pi' --variables
A
B
$ python -m formulate --from-root '(A && B) || TMath::Sqrt(1.23) * e**1.2 + 5*pi' --named-constants
E
PI
$ python -m formulate --from-root '(A && B) || TMath::Sqrt(1.23) * e**1.2 + 5*pi' --unnamed-constants
1.2
1.23
5The most basic usage involves calling from_$BACKEND and then to_$BACKEND, for example when starting with a ROOT style expression:
>>> import formulate
>>> momentum = formulate.from_root('TMath::Sqrt(X_PX**2 + X_PY**2 + X_PZ**2)')
>>> momentum
Expression<SQRT>(Expression<ADD>(Expression<POW>(Variable(X_PX), UnnamedConstant(2)), Expression<POW>(Variable(X_PY), UnnamedConstant(2)), Expression<POW>(Variable(X_PZ), UnnamedConstant(2))))
>>> momentum.to_numexpr()
'sqrt(((X_PX ** 2) + (X_PY ** 2) + (X_PZ ** 2)))'
>>> momentum.to_root()
'TMath::Sqrt(((X_PX ** 2) + (X_PY ** 2) + (X_PZ ** 2)))'Similarly, when starting with a numexpr style expression:
>>> my_selection = formulate.from_numexpr('X_PT > 5 & (Mu_NHits > 3 | Mu_PT > 10)')
>>> my_selection.to_root()
'(X_PT > 5) && ((Mu_NHits > 3) || (Mu_PT > 10))'
>>> my_selection.to_numexpr()
'(X_PT > 5) & ((Mu_NHits > 3) | (Mu_PT > 10))'If the the type of expression isn't known in advance formulate can also auto detect it:
>>> my_sum = formulate.from_auto('True + False')
>>> my_sum.to_root()
'true + false'
>>> my_sum.to_numexpr()
'True + False'When calling from_* the returned object is derived from formulate.ExpressionComponent. From this object you can inspect the expression to find it's dependencies:
>>> my_check = formulate.from_auto('(X_THETA*TMath::DegToRad() > pi/4) && D_PE > 9.2')
>>> my_check.variables
{'D_PE', 'X_THETA'}
>>> my_check.named_constants
{'DEG2RAD', 'PI'}
>>> my_check.unnamed_constants
{'4', '9.2'}Additionally ExpressionComponent s can be combined using both operators and numpy functions:
>>> new_selection = (momentum > 100) and (my_check or (numpy.sqrt(my_sum) < 1))
>>> new_selection.to_numexpr()
'((X_THETA * 0.017453292519943295) > (3.141592653589793 / 4)) & (D_PE > 9.2)'As the == operator returns a new expression, it can't be used to check for equality. Instead the .equivalent method should be used:
TODO: Implement this using expression.equivalent !
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent