Create data classes used by the Azure physical costing model #356
Conversation
| algorithm_qubits = _PRETTY_FLOAT # Number of algorithm qubits $Q_{alg}$ | ||
| measurements = _PRETTY_FLOAT # Number of measurements $M_R$. | ||
| t_gates = _PRETTY_FLOAT # Number of T gates $M_T$. | ||
| toffoli_gates = _PRETTY_FLOAT # Number of Toffoli gates $M_{Tof}$. | ||
| rotation_gates = _PRETTY_FLOAT # Number of Rotations $M_R$. | ||
| rotation_circuit_depth = _PRETTY_FLOAT # Depth of rotation circuit $D_R$. |
There was a problem hiding this comment.
these comments are redundant with the attributes section. You can put the latex in the attributes docstring
|
|
||
|
|
||
| @frozen | ||
| class AlgorithmSpecs: |
There was a problem hiding this comment.
spec? that's short for specification. This doesn't really specify an algorithm. These are properties of an algorithm
There was a problem hiding this comment.
is this class redundant with magic_state_factory.MagicStateCount? Can you refactor to remove code duplication
| """AlgorithmSpecs is a summary of a quantum algorithm/circuit. | ||
|
|
||
| Attributes: | ||
| algorithm_qubits: Number of qubits used by the algorithm. |
There was a problem hiding this comment.
Each of these defaults to zero. Is this a sensible default? Please document this fact somewhere in the docstring since there is indirection in the attribute field type
| def physical_qubits(self, code_distance: int | np.ndarray) -> int | np.ndarray: | ||
| """Computes number of physical qubits""" |
There was a problem hiding this comment.
"""The number of physical qubits for the given code distance using this scheme.
also: this needs an abc.abstractmethod or NotImplementedError.
| def logical_time_step( | ||
| self, code_distance: int | np.ndarray, physical_parameters: PhysicalParameters | ||
| ) -> float: | ||
| """Computes the logical time step.""" |
|
|
||
|
|
||
| @frozen | ||
| class QuantumErrorCorrectionScheme: |
|
|
||
| @frozen | ||
| class QuantumErrorCorrectionScheme: | ||
| """QuantumErrorCorrectionScheme represents a quantum error correction scheme. |
There was a problem hiding this comment.
this is an unhelpful line.
A full quantum error correction scheme would include a particular code, decoder, way of doing gates, and many more things. As I understand it: this class captures a small slice of the relevant characteristics of an error correcting code. Consider a more helpful name for this class. And/or consider making this abstract and giving it a concise name (and meaningful names to the implementations).
| error_rate_scaler: Logical error rate coefficient. | ||
| error_rate_threshold: Logical error rate threshold. |
There was a problem hiding this comment.
Please document these constants in terms of the mathematical formula in which they appear. I don't think "error_rate_scaler" is a common term and "error_rate_threshold" would imply that this is a property of the code whereas in reality this is yet another fuzzy parameter that captures the empirical behavior of the error suppression.
Note that we measure/target/think about the parameter
There was a problem hiding this comment.
is
There was a problem hiding this comment.
The definition of lambda definitely includes the physical error rate
you can use https://arxiv.org/abs/2207.06431 as a reference for lambda. It's the error at d divided by the error at d+2; i.e. how much suppression you get by increasing the code distance.
Co-authored-by: Matthew Harrigan <[email protected]>
part 1 of splitting #327