Hayk Tepanyan

I tried this little fix, it works fine sometimes but more often than not it gives a segmentation fault: ```c++ void init_state(const py::array_t &input_vector) { StateSpace state_space = factory.CreateStateSpace(); py::buffer_info...

I have all the variables in global scope, and the reference is kept - but still getting the segmentation fault. ```python import cirq import qsimcirq def get_all_H(N): circuit = cirq.Circuit()...

hmmm - the free at [pybind_main.cpp:788](https://github.com/quantumlib/qsim/blob/235ae2fc039fb4a98beb4a6114d10c7f8d2070f7/pybind_interface/pybind_main.cpp#L788) releases the state back to python. Where is the second (double) free?

In that case there should be an issue on the second deallocation, e.g. when python gc kicks in after global scope variables are cleaned up. However I see seg fault...

that issue should be deterministic though right? However I see the same code working ~20% of the time and seg faulting 80% of the time.

+1

@rht - Thanks for creating this request. Dear @leofang and @mtjrider - Do you have any estimates on when we could expect qsim v0.17 in cuQuantum?

Hi @trbromley, I just added final touches to the tutorial - please let me know if there is anything thats needs fixing. 1) One thing that's remaining is the Pennylane...

I will send a separate commit soon to address the remaining comments. @ikurecic, btw I already have a Pennylane.ai account - username is [hayk_tepanyan](https://discuss.pennylane.ai/u/hayk_tepanyan/summary).

hi @ikurecic - just created a pennylane.ai profile with [hayk_tepanyan](https://pennylane.ai/profile/hayk_tepanyan) Also pushed a commit to address all the remaining comments. Let me know what else is needed! Re Poetry hash...