Opening Statistical Physics of Generative Grammars
Statistical Physics of Generative Grammars
- Length
- 35 min
- Level
- Research
- Recorded
- International Centre for Theoretical Physics2023
Chapter 1 of 6: Before the talk
DeGiuli assigns grammar rules Gaussian random energies and finds a symmetry-breaking transition to syntax.
Generative grammars can be studied with the tools of disordered systems, in the way spin glasses gave physics a working notion of complexity. DeGiuli defines grammars as string rewriting rules over hidden symbols and shows how context-free grammars build derivation trees with long-range correlations. His random language model assigns Gaussian IID energies to grammar rules, giving an equilibrium ensemble with a temperature parameter. Below a critical temperature, permutation symmetry among hidden symbols breaks and the Shannon entropies of phrase structures drop sharply, which signals grammatical structure. Numerical order parameters show self-averaging across the ensemble. Because exact solution is hard, DeGiuli turns to simpler regular grammars linked to fMRI data, then takes questions on probabilistic grammars, scaling relations and partition functions.
Generative grammars can be studied with the tools of disordered systems, in the way spin glasses gave physics a working notion of complexity. DeGiuli defines grammars as string rewriting rules over hidden symbols and shows how context-free grammars build derivation trees with long-range correlations. His random language model assigns Gaussian IID energies to grammar rules, giving an equilibrium ensemble with a temperature parameter. Below a critical temperature, permutation symmetry among hidden symbols breaks and the Shannon entropies of phrase structures drop sharply, which signals grammatical structure. Numerical order parameters show self-averaging across the ensemble. Because exact solution is hard, DeGiuli turns to simpler regular grammars linked to fMRI data, then takes questions on probabilistic grammars, scaling relations and partition functions.
EDG PRL 2019
Corominas-Murtra, Valverde, Sole 2009
EDG J Phys A 2019; 2022
Mosam et al PRE 2021
Phys. Rev. Lett. 2019
J. Phys. A 2019
PRE 2021

This lecture
Talk 21 of 36 of Workshop on Stochastic Thermodynamics - WOST IV.
Choose a node to see how it connects to this lecture. What comes before is on the leftat the top. What comes after is on the rightat the bottom. The ring is what it teaches. People and places are above. Other lectures and books are below.
Press a chosen node again to open it.
Your comment is kept while you sign up.Sign in