[Feature] Add 'Support' as an attribute.

pasqal-io / qadence2-expressions

Qadence 2 expressions consisting of symbolic expressions and blocks, i.e. quantum digital, analog, digital-analog and composite gates, together with their logic and engine.

https://pasqal-io.github.io/qadence2-expressions/latest/

1 stars 0 forks source link

[Feature] Add 'Support' as an attribute. #31

Closed RolandMacDoland closed 4 hours ago

RolandMacDoland commented 4 hours ago

Support is currently an arg in the quantum op expression creation and is accessed through indexing self. This is fragile as it presupposes the user to pass arguments in an defined order which is not guaranteed nor enforced in our parameters list. Possible solutions:

[ ] Make Support a positional argument
[ ] Make Support an attribute that can be key accessed
- [ ] Be a bit defensive in checking the presence of that attribute and raise an error if absent.

RolandMacDoland commented 4 hours ago

@kaosmicadei wdyt ?

kaosmicadei commented 4 hours ago

The position of the arguments is only for internal use. The idea is to use the class method constructor Expression.quantum_operator, which already implements the support as a positional argument.

But this is an advanced user feature used to define new operators. Regular users are not required to go that deep.

About the indexation of the arguments, it’s the standard procedure with S-expression and AST.

Moving the support to the attribute also changes the mean of it. The attributes were designed to hold metadata related to symbolic simplification.

This can be rethought to become more pythonic but can potentially make the symbolic manipulation more complex.

RolandMacDoland commented 4 hours ago

The position of the arguments is only for internal use. The idea is to use the class method constructor Expression.quantum_operator, which already implements the support as a positional argument.

But this is an advanced user feature used to define new operators. Regular users are not required to go that deep.

About the indexation of the arguments, it’s the standard procedure with S-expression and AST.

Moving the support to the attribute also changes the mean of it. The attributes were designed to hold metadata related to symbolic simplification.

This can be rethought to become more pythonic but can potentially make the symbolic manipulation more complex.

Yeah OK I see. It simply felt a bit fragile to see that indexing but effectively this is enforced by the Expression constructors.