Qgate – Classical trajectory filtering for noisy quantum circuits

Hi HN, I’m Ran. I built an open-code middleware package for Qiskit that uses classical post-selection to filter thermal noise out of deep quantum circuits without requiring global error correction.The current bottleneck in near-term quantum computing (NISQ) is the decoherence death spiral. When IBM proved "Quantum Utility" on 127 qubits in 2023, they used Zero-Noise Extrapolation (ZNE) to statistically guess the correct expectation values from the noise. I wanted to see if deep circuits retain exploitable statistical structure beyond apparent decoherence collapse.How it works: qgate adds a single ancilla probe (an O(1) quantum overhead) whose measurement outcome provides a trajectory score. During classical post-processing, shots are ranked using a dynamic Galton-style threshold, and the lowest-scoring (most decohered) trajectories are discarded. Unlike symmetry-based post-selection, this does not rely on known conserved operators.The Stress Test: I ran it on a 133-qubit heavy-hex lattice (IBM Torino) at an ISA depth of 16,700+ gates (~37x T1). In this regime, expectation values approach the infinite-temperature limit. After trajectory filtering, an 11.9% subset of shots exhibited a statistically significant negative energy shift (δ = -0.0798).I’ve published the telemetry, the math, and the codebase under a Source-Available Evaluation License.

Repo/Docs: https://github.com/ranbuch/qgate-trajectory-filter

The Story/Data: https://medium.com/@ranbuch/beating-the-quantum-noise-wall-e...

Install: pip install qgate