Interactive measurements represent the most powerful way in which agents can learn about their environments. When toddlers want to understand cause and effect for example, they don't merely observe. Instead they try different actions, and observe the reactions. This interactivity is key to learning the causal relations between objects. As such, in Artificial intelligence and learning, interaction is key.
We know the quantum uncertainty principles constrains how much we can simultaneously know about two standard quantum measurements - is there a similar constraint for interactive measurements? Our work here builds such a theory, and as a side-product, we demonstrate that it shows us measurements of different causal structures can be incompatible - a phenomena unique to quantum correlations that can aid us in inferring causal structure.
- Quantum Uncertainty Principles for Measurements with Interventions
Yunlong Xiao, Yuxiang Yang, Ximing Wang, Qing Liu, and Mile Gu
Phys. Rev. Lett. 130, 240201
