DLR Develops Scalable Unitary Data Encoding Beating Exponential State-Preparation Depth

Researchers at the German Aerospace Center (DLR) Institute of Software Technology and the University of Cologne have introduced a deterministic quantum state-preparation framework that reduces two-qubit gate depth by up to two orders of magnitude compared to standard multiplexer compilers. The architecture combines tensor-network representations, Boolean logic minimization (BT-BE Algorithm) and Walsh–Hadamard transformations (BT-WHT Algorithm) to achieve sub-linear depth scaling across structured function classes. Memory Compression using Tensor Cross Interpolation lowers classical memory scaling from O(2^l) to O(2^l χ^2), enabling encoding of out-of-core datasets up to 36 qubits. Physical validation on the 20-qubit IQM Emerald processor encoded a 2D Gaussian state at total circuit depth 56 (CZ depth 26) with state infidelity I = 0.011 ± 0.002 over 256,000 shots, staying within coherence limits. Benchmarks on Lorentzian, polynomial and financial jump-diffusion distributions show asymptotic depth plateaus at fixed infidelity (~5 × 10⁻³), breaking the exponential scaling of standard Qiskit tools. ¹