Researchers have developed a new method, called Error Per Circuit Layer (EPCL), to evaluate the performance of large-scale quantum systems. This approach is a significant milestone because it allows measuring the accumulated effect of noise in quantum operations without the need for costly classical simulations or structured gate sets, making it compatible with a wide variety of quantum architectures, including those with non-Clifford gates.

EPCL works by applying identical random circuits to two disjoint quantum registers and measuring the overlap between their output states as a function of circuit depth. The decay of this overlap provides an estimate of the effective layer polarization. Unlike other benchmarking methods, EPCL avoids the need for classically simulating ideal output distributions or recovering a known reference state, which significantly simplifies the characterization process.

Numerical simulations have shown that EPCL recovers the predicted polarization under weak local stochastic noise and remains well-described by a single-exponential decay even at stronger stochastic noise levels. Coherent errors associated with fixed entangling layers may require techniques such as Pauli twirling or randomized compiling to produce the expected decay. Experiments conducted on IBM quantum hardware with 8- and 16-qubit implementations have demonstrated clear EPCL decay, validating its utility for measuring aggregate register performance without the limitations of previous methods.