Search papers, labs, and topics across Lattice.
This paper operationalizes quantum complexity by extending computational mechanics to quantum space series, defining structure via causal generation memory. It demonstrates that for stationary, memoryful series with finite-dimensional causal models, a memory-minimal isometric generator exists without a discarded environment, whose canonical matrix product state representation yields minimum memory dimension and entropy. This minimum memory entropy is shown to be the entanglement entropy across the past-future bipartition, providing an operational interpretation of entanglement as irreducible causal memory.
Entanglement entropy directly quantifies irreducible memory for causal generation in quantum processes.
What makes one quantum process more complex than another? We adopt an operational approach inspired by classical complexity science, which characterises structure through the memory required for causal generation. We develop a fully-quantum extension of computational mechanics for quantum space series: Sequences of quantum outputs that do not return to interact with their generator. For stationary, memoryful quantum space series admitting finite-dimensional, primitive causal models, we show that a memory-minimal model can be chosen to generate the outputs isometrically, without a discarded environment. The resulting canonical matrix product state representation determines both the minimum memory dimension and the minimum memory entropy. The latter is precisely the entanglement entropy across the past-future bipartition, giving entanglement an operational interpretation as irreducible memory for causal generation. Within this class, we further establish a vanishing fundamental entropy rate, the absence of causal asymmetry, and infinite quantum Markov order. These results establish an operational framework for quantifying intrinsic quantum structure, revealing deep connections between causal generation, complexity, memory, entropy, and entanglement.