Work

How noisy quantum time loops select forecasts that match the future

Year: 2026
Type: preprint
Abstract: Used as a forecast test, a quantum time loop simulated by postselected teleportation selects the runs in which an early random forecast matches a later choice, and it selects well only if a later operation can stop the loop from closing when the two differ. We prove an exact formula for the least closing probability of n loops under energy loss. At any non-zero loss, one loop can never be stopped completely, so some wrong forecasts always survive. Two loops operated jointly can be stopped completely by a single entangling gate for loss up to 82.8%, and n loops for loss up to 1 − (2^(1/n) − 1)^2. No wrong forecast then survives, for any number M of possible choices, but only a fraction p_max/M of runs survives. For two loops, complete stopping needs a nonseparable operation once loss exceeds 50%. Under a simulated noise model for a six-qubit superconducting circuit, the joint gate reaches 2.9 bits of contrast per loop against 2.4 for the best separate operations at 83% loss, and wrong three-bit forecasts fall from 21% to 11% of surviving runs. In a pilot on an IBM Heron processor at a programmed loss of 86%, the joint gate left both loops closed in 0.90% of runs against 1.19% for a separate veto on each loop, a little under half the simulated gain.
Language: English
Open Access status: green
OpenAlex ID: W7235999328
Cites: 16
Cited by:
Related to: 10