Digital Twins for Complex Systems: Theory and Algorithms

A working group at the Santa Fe Institute (SFI) met for three days at the end of September 2026 to work on the theory and algorithms of digital twins and on applying them to complex systems. It was closed to the public. Two of the three organisers are professors at the University of Texas at Austin: Karen Willcox in aerospace engineering, who is also an SFI external professor, and Omar Ghattas in mechanical engineering and geological sciences. The third, Bart van Bloemen Waanders, is a computational scientist at Sandia National Laboratories. SFI's news story names the other participants' fields as aerospace and structural engineering, disaster response and applied mathematics, and names one speaker, Caroline Chung of the MD Anderson Cancer Center, who works on twins of cancer patients.

The event page uses the strict definition: a digital twin is "a computational model or set of coupled models that evolves over time", and data flows both ways between the model and the thing it models. It credits the National Academies' 2024 report with pulling the field together around three kinds of mathematics. Data assimilation folds new observations into a running model. Optimal control picks the actions that steer a system. Uncertainty quantification says how far to trust the answer.

The questions SFI reports from the meeting are how much computing a twin needs, how much detail it needs to be useful, and how it earns public trust. SFI says the organisers are preparing an edited book with tutorials on methods and chapters on applications. The SFI story is filed as its own entry: Building better digital twins.