Quantum Agent-Based Modelling

Agent-based model (ABM is a model simulating autonomous agents, environment, and interactions (agent-agent, agent-environment). The idea is brilliant by design: by performing the actions at a "microscopic" scale, it becomes possible to predict the general behavior of the phenomena. 

Let's consider the simulation of the stock market. We can easily imagine the agents called "traders" acting in the environment of the stock market. Each "trader" has properties, for example, portfolio, balance, and risk appetite. "Market" environment contains information about all available assets, total market metrics, etc. To predict market behavior, one may try to enrich the agents with interactions allowing buying and selling the assets.

There are a lot of systems represented as interacting agents. That's why ABMs used in different areas: epidemiology, finance, physics, complex systems, sociology, and many more. 

But the advantage of this model appears not only if the behavior of a system naturally maps to interacting agents. Agents, environment, and interactions can be purely abstract. Let's consider the next complex satisfiability problem: we want to design the material with the required properties, and we want this material to participate in different experiments validating its properties. The number of variables grows fast with the complexity of the input constraints. We can represent each candidate material as an agent, define the properties space as the environment, and properties optimization between nearest neighbors as interaction.

One of the limitations of the agents is complexity. The problem can require a significant volume of agents (and interactions) to perform the simulation with the required accuracy. Another side is the ability of agents to execute complex strategies, process complex input data, and keep the previous states of the system.

Quantum computing can potentially bring value in this area. The recent spike in QNNs opens the door for "smart" agents or new kinds of heuristics. Moreover, there is a complete framework handling the input-output processes using the power of computational mechanics and quantum theory. According to recent articles, quantum agents can significantly reduce memory consumption. All these results provide a strong feeling that qABMs can become the next breakthrough in simulations of complex systems.

References:

1. Barnett, N. & Crutchfield, J. P. Computational mechanics of input-output processes: Structured transformations and the epsilon-transducer. J. Stat. Phys. 161, 404–451 (2015). ArXiv: https://arxiv.org/pdf/1412.2690.pdf
2. Thomas J. Elliott, Mile Gu, Andrew J. P. Garner, Jayne Thompson. Quantum adaptive agents with efficient long-term memories. 2021. ArXiv: https://arxiv.org/abs/2108.10876
3. J. Thompson, A. J. P. Garner, V. Vedral, and M. Gu, Using quantum theory to simplify input-output processes, npj Quantum Information 3, 6 (2017). npj: https://www.nature.com/articles/s41534-016-0001-3

Image credit

Ira Winder. Agent-based modeling technique that communicates relative flow and directionality of a random OD matrix. https://ira.mit.edu/blog/agent-based-visualization