Quantum amplitude estimation with error mitigation for time-evolving probabilistic networks

TL;DR

This paper introduces a quantum method for modeling time-evolving probabilistic networks, enabling efficient estimation of complex network failure probabilities with error mitigation on noisy quantum hardware.

Contribution

It develops a quantum amplitude estimation approach for dynamic probabilistic networks, including error mitigation techniques, applicable to various real-world network systems.

Findings

Successful low-depth quantum amplitude estimation on a simulator with noise.

Implementation of the method on the AQT PINE quantum computer system.

Introduction of an error model that enhances result accuracy on noisy quantum hardware.

Abstract

We present a method to model a discretized time evolution of probabilistic networks on gate-based quantum computers. We consider networks of nodes, where each node can be in one of two states: good or failed. In each time step, probabilities are assigned for each node to fail (switch from good to failed) or to recover (switch from failed to good). Furthermore, probabilities are assigned for failing nodes to trigger the failure of other, good nodes. Our method can evaluate arbitrary network topologies for any number of time steps. We can therefore model events such as cascaded failure and avalanche effects which are inherent to financial networks, payment and supply chain networks, power grids, telecommunication networks and others. Using quantum amplitude estimation techniques, we are able to estimate the probability of any configuration for any set of nodes over time. This allows us, for example, to determine the probability of the first node to be in the good state after the last time step, without the necessity to track intermediate states. We present the results of a low-depth quantum amplitude estimation on a simulator with a realistic noise model. We also present the results for running this example on the AQT quantum computer system PINE. Finally, we introduce an error model that allows us to improve the results from the simulator and from the experiments on the PINE system.