Środowiskowe Seminarium z Informacji i Technologii Kwantowych

In a bipartite scenario, nonlocality can be thought of as an indicator of "true bipartiteness" of a correlation, i.e., that the full description of the system cannot be reproduced by studying each party individually. Extending this idea to the multipartite case has led to the introduction of Genuine Multipartite Nonlocality (GMNL); a GMNL correlation is nonlocal with respect to any bipartition of the parties. However, it turns out that in some cases, this definition can be cheated, e.g., two bipartite correlations can pass a test for GMNL in a tripartite scenario. This led to the introduction of a modified GMNL definition, called LOSR-GMNL, based on a network with Local Operations and Shared Randomness. In this talk, I will introduce the concept of LOSR-GMNL, show why it is resilient to the types of "attacks" to which the old definition is vulnerable, and discuss the inflation technique - a useful proof method for a LOSR network scenario. Lastly, I will present the results of our work concerning LOSR-GMNL, in which we show that all caterpillar graph states, as well as some other classes of states, can produce LOSR-GMNL correlations.