Universality

Matsuo Sato

We formulate a quantum theory of the Universe based on Bayesian probability. In this theory, the probability of the Universe is not a frequency probability, which can be obtained by observing experimental results several times, but is a Bayesian probability, which can define a probability of an event that occurs just once. As an example, by applying the quantum theory of the Universe to an action of a scalar field theory in the four dimensions as a toy model for the theory of the Universe, we explicitly obtain the probability of the Universe and the action of matters in the Universe.

Tomáš Gonda, Coves, Gemma De les

A universal Turing machine is a powerful concept - a single device can compute any function that is computable. A universal spin model, similarly, is a class of physical systems whose low energy behavior simulates that of any spin system. Our categorical framework for universality (arXiv:2307.06851) captures these and other examples of universality as instances. In this article, we present an accessible account thereof with a focus on its basic ingredients and ways to use it. Specifically, we show how to identify necessary conditions for universality, compare types of universality within each instance, and establish that universality and negation give rise to unreachability (such as uncomputability).

Amit Hagar

Tobias Reinhart, Gemma De les Coves

One of the theoretical pillars that sustain certain machine learning models are universal approximation theorems, which prove that they can approximate all functions from a function class to arbitrary precision. Independently, classical spin models are termed universal if they can reproduce the behavior of any other spin model in their low energy sector. Universal spin models have been characterized via sufficient and necessary conditions, showing that simple models such as the 2d Ising with fields are universal. In this work, we prove that universal spin models are universal approximators of probability distributions. This enables us to leverage the characterization of the former to reveal conditions which are sufficient for universal approximation. Deriving universal approximation theorems thus amounts to verifying these conditions, yielding a unified recipe for universal approximation theorems applicable to a wide range of models. We explicitly test this recipe for restricted and deep Boltzmann machines, as well as for deep belief networks. This work illustrates that independently discovered universality statements may be intimately related, enabling the transfer of results.