Philosophy of Physics Seminar (Thursday - Week 2, MT24)
Thursday 22nd October, 3:00pm - 5:00pm
Lecture Room, Radcliffe Humanities
Peter Morgan (Yale University): 'A Dataset & Signal Analysis Interpretation of Quantum Field Theory’
Abstract: Elementary signal analysis, with a dependence only on time, can be thought of as a 1+0-dimensional classical field theory, for which Fourier and other integral transforms and the use of Hilbert spaces are familiar. Signal analysis is less constrained than traditional classical mechanics insofar as its relationship with the available datasets does not assume a priori that the data is about object properties, making it a helpful intermediary for thinking about quantum field theory. The presence of several distinct kinds of noise requires statistical methods and, to accommodate the intervention and causal modelling aspect of signal analysis in a Hilbert space formalism, generalized probability theory.
I will show how we can use the Poisson bracket to extend classical mechanics to be as unconstrained as signal analysis, giving what I have called 'CM+'. The greater generality of CM+ includes measurement incompatibility, so that it has a measurement problem, which allows us to rethink the measurement problem as we have it for quantum mechanics. If we further require a CM+ model to differentiate between thermal noise and quantum noise (spoiler: it's about Lorentz invariance), then we must work in at least 1+1-dimensions and we can make even closer contact with quantum field theory.
I will also show that the nonlinear response to applied modulations that we expect in a signal analysis perspective suggests a way to rethink renormalization as a surreptitious way to introduce nonlinearity into an axiomatic quantum field theory. Signal analysis thus gives us ways to rethink both the measurement problem and the renormalization 'problem'.