Information Physics
Published on 11 Oct 2007 at 12:01 am.
1 Comment.
Filed under Information, Physics, Probability, Research.
Three of us here at the University at Albany are becoming collectively known as the Albany Group: Ariel Caticha, Carlos Rodriguez and myself. We three are exploring some relatively radical ideas about information, probability, geometry and physics, and often we find our ideas converging in unexpected ways.
One convergence that I am particularly interested in is this notion that the Laws of Physics are in reality a special application of the Laws of Inference. The idea that is developing is that the laws of physics are not so much laws that describe the universe, but rather are laws that describe how an ideal observer, or a rational agent, could make predictions about the universe. This sounds rather like some of the ideas about quantum mechanics with cruel observers and abused cats, but there are some critical differences. Here I have said nothing about the necessity of an observer. Instead, I posit that the laws of physics describe how an ideal observer would make predictions. This distinction is crucial.
But this is a strange idea that we are taking right to the limit. Even laws as seemingly physical as classical mechanics, such as F = ma, are not laws that describe the object being accelerated as much as they are laws about our making inferences about the object and its position or state of motion when it is interacting with another object. (See Caticha’s MaxEnt 2007 paper)
This is not a particularly unusual idea when one considers thermodynamics, or more generally, statistical mechanics. Here it is almost obvious that the theory is about our ability to make inferences about macrostates given that we can only measure macroscopic variables. This is where entropy appears as a way to assign probability distributions subject to constraints, such as the total energy. And it is here that it is most clear that entropy refers to the optimal state of knowledge of an ideal observer and not the observed system. This becomes even more clear when one considers that a macrostate is not really a “state” at all, but rather is an equivalence class of microstates.
(see my MaxEnt 2007 tutorial)
But this begs the question: What is a state?
My belief is that a state is a description.
State = Statement
That’s rather nice!
And in that sense, a macrostate is a statement. Or more precisely, a macrostate is a disjunction of statements. It is the statement that the system can be described by microstate a or microstate b or microstate c, etc; where stating that the system is in “microstate a” is a statement itself.
Statements, or descriptions, are clearly not properties of a system. They are instead properties of observers caught in the act of describing a system. This space of statements is the space in which physics takes place.
In this sense, physics is all about information…
Information Physics.
Kevin Knuth
Albany NY
Jonathan on 29 Feb 2008 at 3:57 pm: 1
I’m so glad to read this. I feel like this approach clears up a lot of ambiguity when trying to discuss the concept of “Nature”. Additionally, looking at the process of science from this perspective makes it easier to potentially automate.