quantumlib/Cirq

Is your feature request related to a use case or problem? Please explain np.RandomState has been effectively deprecated by numpy but it is still used throughout much of Cirq. Additionally, when multiple threads are used, RandomState also becomes a shared object, causing blocking between threads. Describe the solution you would prefer We should instead use numpy.random.Generator, the modern approach. We should provide a separate path for generating random numbers with Generators and ensure multithreading support. The old RandomState based paths may then be deprecated. Per previous work on this issue, the general approach should be: A new type such as PRNG_OR_SEED_LIKE should be added, allowing for either a generator or a seed. parse_random_generator should be added, which returns a np.random.Generator. Existing calls to parse_random_state within cirq should be replaced with parse_random_generator. parse_random_state should be deprecated. How urgent is this for you? Is it blocking

Describe the issue cirq.defer_measurements does not preserve the record index of a BitMaskKeyCondition. When a measurement key is reused, a condition referring to an earlier record can become an unconditional operation after deferral, changing the measurement result. The index should be preserved, or unsupported conditions should be rejected explicitly. Explain how to reproduce the bug or problem from collections import Counter import cirq q0, q1, q2 = cirq.LineQubit.range(3) condition = cirq.BitMaskKeyCondition( cirq.MeasurementKey("a"), index=0, target_value=0, equal_target=True ) circuit = cirq.Circuit( cirq.X(q0), cirq.measure(q0, key="a"), # First record: 1. cirq.measure(q1, key="a"), # Second record: 0. cirq.X(q2).with_classical_controls(condition), cirq.measure(q2, key="out"), ) simulator = cirq.Simulator() for label, candidate in [ ("original", circuit), ("deferred", cirq.defer_measurements(circuit)), ]: result = simulator.run(candidate,

Is your feature request related to a use case or problem? Please explain We try to tag issues and PRs with tags such as area/gates or area/performance to improve findability, but we currently do tagging manually on an ad hoc basis. There exist numerous GitHub Actions workflows for automatically labeling things based on such things as the paths to files they involve. We should put in place a workflow with a configurable list of labels and rules (kept separate from the workflow) so that we don’t have to spend time doing it and so that labeling is done more consistently. Describe the solution you would prefer How urgent is this for you? Is it blocking important work? P3 – I'm not really blocked by it; it's an idea I'm proposing based on principle

Is your feature request related to a use case or problem? Please explain Mypyc compiles Python modules to C extensions. It uses standard Python type hints to generate fast code. Mypyc uses mypy to perform type checking and type inference. Most type system features in the stdlib typing module are supported. Existing code with type annotations is often 1.5x to 5x faster when compiled. Code tuned for mypyc can be 5x to 10x faster. Describe the solution you would prefer Figure out how to make use of mypyc Implement a limited use of mypyc Test before-and-after results on Cirq benchmarks Report the results How urgent is this for you? Is it blocking important work? P3 – I'm not really blocked by it; it's an idea I'm proposing based on principle

Is your feature request related to a use case or problem? Please explain Numba is an open source, NumPy-aware JIT optimizing compiler for Python. It uses the LLVM compiler project to generate machine code from Python syntax. Numba can compile a large subset of numerically-focused Python, including many NumPy functions. Additionally, Numba has support for automatic parallelization of loops, generation of GPU-accelerated code, and creation of ufuncs and C callbacks. Describe the solution you would prefer Numba claims "You don't need to replace the Python interpreter, run a separate compilation step, or even have a C/C++ compiler installed. Just apply one of the Numba decorators to your Python function, and Numba does the rest." (from https://numba.pydata.org/). So one approach to testing its use is: Find a case where Numba could potentially be used to speed up some aspect of Cirq Test its use and run before-and-after benchmarks Report the results How urgent is this for you? Is it block