Scientific Research

Mineral/Geothermal Exploration: what is the optimal sequence of exploration or appraisal drilling/geophysics that reduces maximally uncertainty on key economic parameters?

Mineral Processing: how to optimize mineral processing accounting for the variability and impurities in feed-stock material from mines?

Mineral Supply Chains: how to create a strategy to build a domestic supply chain of critical minerals from exploration to processing?

Geothermal Energy Production: what is the optimal location and operation of injectors & producers that maximize heat extraction while de-risking earthquake occurrence?

Carbon Storage & Sequestration: how to maximize storage of CO2 in subsurface formations while optimally monitoring for leakage hazards?

All these questions lead to sequential planning under uncertainty problems, and the most optimal solution is given by Artificial Intelligence.

Mineral-X pioneered the use of Intelligent Agents for sequential decision making towards net-zero 2050 goals, with a focus on developing Earth Resources. Mineral-X developed the intelligent prospector, an Intelligent Agent (IA) which helps exploration and mining companies in decision making for optimal data acquisition and engineering operations. An intelligent agent is an AI system that can optimize goals over long-term horizons, accounting for how future information will inform system variables to improve the present decision. This general formulation has also been applied on Carbon Capture and storage and Geothermal Energy in collaboration with OMV, and now extended into mineral processing and building mineral supply chains. Together with our industrial partners, we develop AI methodologies directly into actual practice. 

Our work has been widely featured in domestic and international media

• Stanford Earth: how better mineral exploration makes better batteries 
• Stanford Human-Centered AI institute: Building Intelligent Agents to Reach Net-zero 2050
• MIT: the big tech quest to find the metals needed for the energy overhaul 
• Wired: these mining algorithms are hunting for an EV battery mother lode
• Bloomberg: future of energy requires cleaner electric batteries to solve cobalt problem
• Vanity Fair: the billionaire clubs run on cobalt says everything about our battery powered future 

Since 2019, we have been working with Kobold Metals. We jointly developed an AI algorithm for drillhole planning that has led to an ultra-high grade discovery in Zambia. You can download the AI algorithm used in Zambia:

Mern, J. and Caers, J., 2023. The Intelligent prospector v1. 0: geoscientific model development and prediction by sequential data acquisition planning with application to mineral exploration. Geoscientific Model Development, Geoscientific Model Development, 16(1), pp.289-313.

Mern, J., Corso, A., Burch, D., House, K.Z, & Caers, J., 2024. Intelligent prospector v2.0: exploration drill planning under epistemic model uncertainty. Geoscientific Model Development, arXiv preprint arXiv:2410.10610.

Publications on AI for decision making

• Caers, J., Scheidt, C., Yin, Z., Wang, L., Mukerji, T. and House, K., 2022. Efficacy of Information in Mineral Exploration Drilling. Natural Resources Research, pp.1-17.
• Hall, T., Scheidt, C., Wang, L., Yin, Z., Mukerji, T. and Caers, J., 2022. Sequential Value of Information for Subsurface Exploration Drilling. Natural Resources Research, 31(5), pp.2413-2434.
• Wang, Y., Zechner, M., Mern, J.M., Kochenderfer, M.J. and Caers, J.K., 2022. A sequential decision-making framework with uncertainty quantification for groundwater management. Advances in Water Resources, 166, p.104266.
• Corso, A., Wang, Y., Zechner, M., Caers, J. and Kochenderfer, M.J., 2022. A POMDP Model for Safe Geological Carbon Sequestration. arXiv preprint arXiv:2212.00669. Slides + Video.
• Wang, Y, Zechner, M., Wen, G., Corso, A.L., Mern, J. Kochenderfer, M.J. and Caers, J., 2023. Optimizing carbon storage for long-term safety, arXiv:2304.09352. 
• Moss, R.J., Corso, A., Caers, J. and Kochenderfer, M.J., 2024. BetaZero: Belief-state planning for long-horizon POMDPs using learned approximations. arXiv preprint arXiv:2306.00249.

Data Science for the Geosciences

Wang, L, Zhen Y., and Caers, J., 2023. Data Science for the Geosciences, Cambridge University Press.

Mineral-X has 25 years of experience in developing data science methods with a focus on natural resources. We build on this strength by developing methodologies in the areas of geostatistics, 3D geological modeling, geochemistry, geophysical inversion, spatial data analysis & spatially-aware machine learning.

Topics of ongoing research but not limited to are

• Optimal data acquisition through Bayesian optimal design
• Turning qualitative geological hypothesis into quantitative Bayesian prior models
• Monte Carlo approaches to uncertainty quantification in geophysical inversion
• Accelerating multi-physics geophysical inversion through machine-learning based models
• Developing state-of-the art multivariate analysis to understand the processes revealed by geochemical soil surveys.
• Stochastic level set methodologies for modeling intrusion, stratigraphy in complex structural settings

The following publications cover ongoing work in this area

• Wang, J., Zuo, R. and Caers, J., 2017. Discovering geochemical patterns by factor-based cluster analysis. Journal of Geochemical Exploration, 181, pp.106-115.
• Scheidt, C., Li, L. and Caers, J. eds., 2018. Quantifying uncertainty in subsurface systems (Vol. 236). John Wiley & Sons.
• Wang, Z., Yin, Z., Caers, J. and Zuo, R., 2020. A Monte Carlo-based framework for risk-return analysis in mineral prospectivity mapping. Geoscience Frontiers, 11(6), pp.2297-2308.
• Fouedjio, F., Scheidt, C., Yang, L., Achtziger-Zupančič, P. and Caers, J., 2021. A geostatistical implicit modeling framework for uncertainty quantification of 3D geo-domain boundaries: Application to lithological domains from a porphyry copper deposit. Computers & Geosciences, 157, p.104931.
• Yang, L., Achtziger-Zupančič, P. and Caers, J., 2021. 3D modeling of large-scale geological structures by linear combinations of implicit functions: Application to a large banded iron formation. Natural Resources Research, 30(5), pp.3139-3163.
• Athens, N. and Caers, J., 2022. Stochastic inversion of gravity data accounting for structural uncertainty. Mathematical Geosciences, 54(2), pp.413-436.
• Wei, X., Yin, Z., Scheidt, C., Darnell, K., Wang L. & Caers, 2024. Constructing Priors for Geophysical Inversions Constrained by Surface and Borehole Geochemistry. Surv Geophys 45, 1047–1079 (2024). https://doi.org/10.1007/s10712-024-09843-x

• Scheidt, C., Mathieu,L., Yin,Z., Wang, L. and Caers,J., 2024. Masked Autoregressive Flow for Geochemical Anomaly Detection with Application to Li–Cs–Ta Pegmatites Exploration of the Superior Craton, Canada, Natural Resources Research ( 2024) https://doi.org/10.1007/s11053-024-10409-2