Ten open problems.
The frontiers that still gate trust in quantum hardware.
Scroll
01 / 10
Logical qubits still Markovian?
Physical noise forgets; syndrome memory can break logical GST / RB.
02 / 10
Can we untangle crosstalk?
Separate SPAM, gate error, and neighbor coupling—no independence assumed.
03 / 10
What noise stays unlearnable?
Mid-circuit measurements unlock Pauli parameters cycle benchmarking can't see.
04 / 10
Few shots, real certificates?
Classical shadows turn random measurements into scalable checks.
05 / 10
Is one number enough?
A single fidelity doesn't predict useful quantum advantage.
06 / 10
Can one prover certify BQP?
Information-theoretic single-prover QPIP is still open.
07 / 10
Can benchmarks be certificates?
A high score is not a proof the device did the right quantum job.
09 / 10
Does crosstalk go logical?
Low-weight coupling can become logical ZZ in multi-logical blocks.
10 / 10
Can RB survive memory?
DD and randomized compiling help only in part—not a full fix.
