Add support for simulation over classical distributions #1682
Conversation
- These classes will be used within the classical simulator to represent distributions of classical measurements. - For instance, if measuring in the X-basis, you will get a random result (0 or 1) possibly with an additional phase. - This is part of the measurement-based uncomputation feature (MBUC).
| ) | ||
|
|
||
|
|
||
| def test_get_soquet(): |
qualtran/_infra/composite_bloq.py
Outdated
| We can uniquely address a Soquet by the arguments to this function. | ||
| Args: |
There was a problem hiding this comment.
I think we need some newlines in this docstring for it to be parsed correctly; although this is a private function so it won't actually show up anywhere
| class _FixedClassicalValHandler(_ClassicalValHandler): | ||
| """Returns a random classical value using a fixed value per instance. | ||
|
|
||
| Useful for deterministic testing. |
There was a problem hiding this comment.
Args:
binst_i_to_val: mapping from BloqInstance.i instance indices to the fixed classical value.
qualtran/simulation/classical_sim.py
Outdated
|
|
||
|
|
||
| class _FixedClassicalValHandler(_ClassicalValHandler): | ||
| """Returns a random classical value using a fixed value per instance. |
qualtran/simulation/classical_sim.py
Outdated
|
|
||
| Args: | ||
| reg_name: Name of the register | ||
| idx: Index of the register wite(s). |
qualtran/simulation/classical_sim.py
Outdated
| ) -> 'PhasedClassicalSimState': | ||
| """Initiate a classical simulation from a CompositeBloq. | ||
|
|
||
| Args: | ||
| cbloq: The composite bloq | ||
| vals: A mapping of input register name to classical value to serve as inputs to the | ||
| procedure. | ||
| rng: A random number generator to use for classical random values, such a np.random. | ||
| fixed_random_vals: A dictionary of instance to values to perform fixed calculation |
There was a problem hiding this comment.
instance -> bloq instance index (?)
qualtran/simulation/classical_sim.py
Outdated
| random_handler: '_ClassicalValHandler' = _RandomClassicalValHandler( | ||
| rng=np.random.default_rng() | ||
| ), |
There was a problem hiding this comment.
I think this should be _BannedClassicalValHandler, no? throughout these PRs, I tried to leave the "normal" classical simulator the same, non-phased, and deterministic; and the fancy stuff is "opt-in" by using the PhasedClassicalSimulator
There was a problem hiding this comment.
It's awfully difficult to use though if you set that as the default, since the top-level functions don't have a super great way to pipe in a random_handler (and the random_handler is private).
Are you sure about this?
dstrain115
changed the title
Original work done by mpharrigan in #1678