Geometric Approach to Zero-Memory Quantum Dot Reservoir Computing

Bongsu Kim, Oscar Lee, Sangjun Jeon, Kun Woo Kim · 2026

Read Geometric Approach to Zero-Memory Quantum Dot Reservoir Computing by Bongsu Kim, Oscar Lee, Sangjun Jeon, Kun Woo Kim (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-29320.txt.

Introduction

Kim and three colleagues show that a memoryless physical system can still compute with memory, by trading time for space. Reservoir computing uses a physical system's dynamics as a computer, and memory capacity is usually thought to come from the system's own history-dependence. Their approach instead uses multidimensional input nodes as a spatial memory axis, so the reservoir need not remember anything itself.

They test the idea on a generalised quantum dot, whose discrete energy levels supply strong nonlinearity. In simulations, the memoryless quantum reservoir performed well on chaotic Mackey-Glass forecasting and on nonlinear transformation tasks. Looking at the geometry of the quantum state paths, they find the mechanism: the added spatial memory builds a hysteresis loop in the quantum state space. The loop becomes topologically stable when the system's evolution locks to the input signal's rhythm.

Catalog record

CREATORBongsu Kim, Oscar Lee, Sangjun Jeon, Kun Woo Kim
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE64 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.29320

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Geometric Approach to Zero-Memory Quantum Dot Reservoir Computing Bongsu Kim, Oscar Lee, Sangjun Jeon, Kun Woo Kim Abstract Physical reservoir computing offers an energy-efficient alternative to conventional neural networks, where the intrinsic memory capacity in the physical system plays a central role. We demonstrate that memory capacity can be engineered extrinsically in memoryless systems by exploiting the computational space-time tradeoff, substituting temporal memory with spatial degrees of freedom. Our approach utilizes multidimensional input nodes to function as a spatial memory axis, thereby removing the dependency on intrinsic history-dependent dynamics in the reservoir. We validate this framework through numerical simulations of a generalized quantum dot, whose discrete energy levels provide strong nonlinearity crucial for reservoir computing as well. By coupling this inherent nonlinearity with our extrinsic memory, we show that memoryless quantum reservoir can achieve high performance on both chaotic Mackey-Glass future prediction and nonlinear transformation tasks. Furthermore, by analyzing the geometry of the quantum state trajectories, we identify the physical mechanism underlying this memory emergence: extrinsic memory constructs a hysteresis loop within the quantum Hilbert space, and this loop becomes topologically stable when the evolution of the system state synchronizes with the input signal’s frequency. Our work decouples reservoir computing from material-specific memory properties, significantly expanding the range of candidate systems for quantum neuromorphic computing. I Introduction The rapid expansion of artificial intelligence, particularly large-scale deep learning models, has increased computational demand and energy consumption in data centers, raising concerns about the sustainability and cost of training and deploying such models. In response to this challenge, Physical Reservoir Computing (PRC) has emerged as a compelling alternative framework 24 ; 12 . Unlike standard neural networks that require energy-intensive optimization across all layers, PRC exploits the rich dynamics of a physical substrate—such as optical 1 ; 42 ; 56 , spintronic 50 ; 61 ; 7 , or memristor systems 13 ; 8 ; 60 ; 47 —to project inputs into a high-dimensional state space, requiring training only for the final output layer 25 ; 5 ; 55 ; 3 ; 38 . More recently, this has been extended into quantum regime, where quantum reservoir computing utilizes the exponentially large Hilbert space 40 ; 48 ; 37 ; 39 ; 30 . However, the efficacy of a physical reservoir in processing time-series data, such as chaotic forecasting or speech recognition, depends critically on its capacity for memory 23 ; 33 ; 60 ; 13 . Conventionally, this memory is assumed to be an intrinsic property of a physical system, often arising from relaxation processes or hysteresis that allow the system’s current state to retain echoes of past inputs. Consequently, the search for suitable reservoir materials has largely been confined to physical systems that naturally exhibit this memory property, limiting the scope of viable candidates for reservoir computing 49 ; 41 ; 34 . To overcome this material-given property, several approaches have been proposed to manipulate the memory effect extrinsically, such as using time-delayed feedback 18 ; 51 ; 31 ; 20 ; 2 , optical cavities 1 ; 4 ; 57 ; 42 ; 56 , and rich phase diagrams of magnetic systems 33 ; 32 ; 16 ; 55 ; 28 . However, even in the time-delayed framework, while these computers have proven effective at many tasks, their performance hinges on the system operating near the edge of chaos and its internal relaxation timescale matching the input signal rate. These constraints arise because memory is realized through the history-dependent dynamics of the reservoir itself. In this work, we propose a paradigm shift that circumvents the reliance on the intrinsic dynamics by introducing the concept of extrinsic memory within the input layer. We demonstrate that memory capacity need not be an inherent feature of the reservoir but can instead be engineered through a structured input injection scheme. Because this approach decouples memory from the system’s temporal evolution, the entire computation process can operate in discrete time steps. Our scheme employs multidimensional input nodes to function as a spatial memory axis. Specifically, given L input nodes, we simultaneously inject the current input signal u(n) at time step n along with its past history up to u(n-L+1) across these nodes. This strategy is grounded in the computational principle of the space-time tradeoff 46 ; 45 ; 21 : by utilizing more spatial resources, we reduce the dependency on temporal retention within the reservoir 6 ; 58 ; 15 ; 10 . In conventional reservoir computing, to access past input u(n-L) at time n , the system must sustain the signal state through its internal fading dynamics over L time steps. However, in our setup, the parallel injection across L nodes removes the need for the system to remember the past L steps. By setting L to match the desired relaxation timescale, we directly substitute intrinsic temporal memory with extrinsic spatial memory. This allows even inherently memoryless systems to function as effective reservoirs, thereby significantly expanding the range of physical systems applicable to PRC. To validate this framework, we employ a single quantum dot as a physical reservoir. Quantum dots are promising nanodevices, and experimental techniques for their control have reached a mature stage 9 ; 59 ; 27 ; 11 ; 43 . In particular, due to its quantized energy states, a quantum dot inherently possesses rich nonlinearity. However, despite these advantages, its application to PRC is restricted by a lack of intrinsic memory due to its fast relaxation time. By applying our extrinsic memory scheme, we overcome this critical limitation. Our results show that the quantum dot can be tuned as a versatile reservoir, achieving high performance on both nonlinear transformation and the chaotic Mackey-Glass time series prediction task 36 , a benchmark known to require high memory capacity 35 ; 33 ; 16 ; 51 . Moreover, by analyzing the geometry of the quantum state trajectories evolving in time, we uncover the origin of physical mechanisms driving the performance of reservoir computation characterized by memory capacity as well as nonlinearity and complexity. Within the quantum Hilbert space, a hysteresis-like loop structure is developed as the extrinsic memory grows, and its shape and size directly manifest the magnitude of the three performance measures. The topological robustness is demonstrated using 1-dimensional input signals, however, with a different number of incommensurate frequencies which explores a higher dimensional synthetic space. II Extrinsically introduced Memory Capacity Memory refers to the capacity to store and retrieve information about past events. It implies that its current state and its responses are not solely determined by present inputs, but also by its past history. This history dependence is the fundamental characteristic of a system possessing the memory capacity. Physical systems can contain intrinsic memories 29 ; 53 and they are reflected in the static and dynamical properties. For instance, when such a ferromagnetic materials is driven by an external magnetic field, H(t)\sim\sin(\omega t) , its history-dependency causes the magnetic response to exhibit a temporal lag,yielding M(t)\sim\sin(\omega t-\phi) . In addition, the inherent nonlinearity of the magnetization process introduces higher-order harmonic distortions into this response. The combination of the temporal delay and these nonlinear response make the well known magnetic hysteresis loop in the input-response space 26 (see Fig. 1

Read the complete text