Mathematical Programming-based Multi-Agent Systems (MPMAS)

Spatial data representation and interdependencies

MPMAS is a software package for simulating land use change in agriculture and forestry. It combines economic models of farm household decision-making with a range of biophysical models simulating production response to changes in water supply and soil nutrients (Schreinemachers & Berger 2011).

MPMAS is part of a family of models called multi-agent systems models of land-use/cover change (MAS/LUCC). These models couple a cellular component representing a physical landscape with an agent-based component representing land-use decision-making (Parker et al. 2002). MAS/LUCC models have been applied in a wide range of settings (Kremmydas et al. 2018, Utomo et al. 2018) yet have in common that agents are autonomous decision-makers who interact and communicate and make decisions that can alter their environment.

Other MAS/LUCC applications have been implemented with software packages such as Cormas, NetLogo, RePast, and Swarm (Parker et al. 2003). The main difference between MPMAS and these alternative packages is the use of whole farm mathematical programming to simulate land-use decision-making. With this decision-making component MPMAS is firmly grounded in agricultural economics (Nolan et al. 2009). MPMAS has been applied in a dozen countries around the world. It is flexible in terms of the spatial extent that it can cover and has been used in small-scale as well as large-scale applications. Selected applications are shown in the table below.

Hybrid intelligence – arising from the sensible, targeted fusion of human minds and cutting-edge computational systems – holds great potential for enhancing the sustainability of agriculture. Hybrid intelligence systems can be understood as ‘systems of systems’, with the lower-order subsystems being humans and artificially intelligent machines (Dellermann et al. 2919). These subsystems are constantly adapting to dynamic environments, and their behavior is strongly and iteratively intertwined so that intelligent behavior emerges from the interaction of subsystems. Hybrid intelligence is particularly apt to addressing complex problems, capitalizing on the rapid development of AI and harnessing the complementary strengths of human and machine intelligence (Peeters et al. 2021).

Empirically grounded agent-based models such as MPMAS have now reached maturity and provide a valuable starting point for achieving hybrid intelligence in agriculture (Berger et al. 2024). They are able to implement geospatially-explicit computational systems replicating the state and processes of agricultural landscapes while representing the multitude of plants, animals, humans, and their technological artifacts. Because computational entities directly represent real-world entities (e.g., farms, plots), MPMAS allows for each single agent and each single plot to be checked and validated one-to-one. This differs from ‘black-box’ AI models, which are often considered nontransparent and unexplainable. In this sense, the one-to-one representation in MPMAS helps create trust in interactive modeling, especially when computational agents resemble the holdings of real-world farmers (Moessinger et al. 2022).

Table. Applications of MPMAS 
ApplicationNo. of farm agentsSpatial
dimension
Temporal
dimension
Type of
agriculture  
extent [km2]resolution [m]duration [years]time step [days]
Brazil,
Mato Grosso
84416,65710051-365 *Market-oriented and commercial; Soybean, maize, cotton and cattle

Chile,
Maule Basin

3,5945,3001002030Market-oriented and commercial

Ethiopia,
Multiple locations across the country

130133515365Semi-subsistence; teff, maize, wheat, beans, eucalyptus and livestock

Germany,
Southwest

1,7001,297100221-365 *

Market-oriented and commercial

Ghana,
White Volta Basin
34,6913,7791001530-365 *Semi-subsistence; rice, millet, maize, onion and tomato 
Thailand,
Northern uplands
1,309140401530-365 *Commercial fruit, vegetable and flower production
Uganda 1,
Southeast
52012711630-365 *Semi-subsistence; maize, cassava, bean and plantain 

Uganda 2,
Central

1,71630321630-365 *Semi-subsistence; coffee, plantain, maize, beans and livestock
Vietnam,
Northern uplands
1,4716102530-365 *

Semi-subsistence; maize, cassava and rice

Note: * Components of the model have different time steps. The decision-making follows an annual sequence while land, labor, crop water requirements, irrigation water supply, and rainfall are specified on a daily or monthly basis.

References

  • Dellermann, D., Ebel, P., Söllner, M., Leimeister, J.M., 2019. Hybrid Intelligence. Business & Information Systems Engineering 61, 637–643.
  • Kremmydas D., Athanasiadis I.N., Rozakis S., 2018. A review of Agent Based Modeling for agricultural policy evaluation. Agricultural Systems 164, 95–106.
  • Moessinger, J., Troost, C., Berger, T., 2022. Bridging the gap between models and users: A lightweight mobile interface for optimized farming decisions in interactive modeling sessions. Agricultural Systems 195, 103315.
  • Nolan, J., Parker, D., van Kooten, G.C., Berger, T., 2009. An Overview of Computational Modeling in Agricultural and Resource Economics. Canadian Journal of Agricultural Economics 57, 417-429.
  • Parker, D.C., Berger, T., Manson, S.M., 2002. Agent-Based Models of Land-Use/Land-Cover Change: Report and Review of an International Workshop, in, LUCC Focus 1, Publication 6, LUCC Focus 1 Office, Indiana University, Bloomington.
  • Parker, D.C., Manson, S.M., Janssen, M.A., Hoffmann, M.J., Deadman, P., 2003. Multi-agent systems for the simulation of land-use and land-cover change: a review. Annals of the Association of American Geographers. 93(2), 314-337.
  • Peeters, M.M.M., van Diggelen, J., van den Bosch, K., Bronkhorst, A., Neerincx, M.A., Schraagen, J.M., Raaijmakers, S., 2021. Hybrid collective intelligence in a human–AI society. AI & Society 36, 217–238.
  • Schreinemachers, P., Berger, T., 2011. An agent-based simulation model of human environment interactions in agricultural systems. Environmental Modelling & Software 26, 845-859
  • Utomo D.S., Onggo B.S., Eldridge, S., 2018. Applications of agent-based modelling and simulation in the agri-food supply chains. European Journal of Operational Research 269, 794–805.