# Can physics be represented/approximated by causal models?

Discussion in 'Physics' started by Programmer2134, Oct 8, 2018.

1. ### Programmer2134Guest

There is a literature on “causal models”. See e.g. Judea Pearl’s “Causality” .

A causal model (for current purposes), roughly, is a set of (finite or infinite) time-dependent variables $V$, with for each variable $i$, causal relations: $$v_{i,t}=f_i(v_t)+\epsilon_{i,t}$$

Where $v_t$ is the vector of all variables, and $f_i$ denotes the transition function of variable $i$ depending on the previous variables. $\epsilon_{i,t}$ is a random component for each variable, but may be omitted in case of a deterministic model.

I know that current physics is time-direction invariant, so that physics don’t really think of causality as a “fundamental” thing, but rather more as an informal concept.

However, I am wondering to what extent physics can be represented in the formalism of causal graphs. Perhaps fundamentally quantum physics cannot, but can it work as an approximation, e.g. to classical mechanics? If not, what is the specific reason why causal models are too restrictive?

Disclaimer: I don’t know that much about physics, beyond the basics.