Sufficiency of Longer Windows Under Weak Dissipation

Determine whether input windows longer than six steps suffice to meet the history criterion for quantum reservoir emulation under weak-loss conditions where windows of up to six inputs fail.

Background

The numerical experiments examine whether a reset-window quantum extreme learning machine (QELM) receives enough recent input history to reproduce the outputs of a recurrent quantum reservoir computer (QRC). The history criterion is based on the remaining variation in reservoir outputs across histories that share the same recent input window. Under the strongest tested loss, windows of two inputs satisfy the criterion for the tested reservoir sizes; under weak loss, no tested window of up to six inputs does.

The paper does not determine whether extending the window beyond six inputs would eventually make the omitted history negligible in the weak-loss regime. This leaves unresolved the amount of context required for accurate QELM emulation when reservoir forgetting is slow.

References

At weak loss, no tested window of up to six inputs does; longer windows remain unresolved.

— When Can Quantum Extreme Learning Machines Replace Quantum Reservoirs?  (2609.31027 - Baumann et al., 25 Sep 2026) in Section 5.1, “When Is the History Sufficient?”