# How do symmetries “define” physical laws?

First of all, I do not have any problems concerning what symmetries are or how to describe them. However, I do not have any knowledge concerning how the reasoning for quantum field theory and thus the standard model works. I hope it is still appropriate to ask such a question this early.

What concerns me is a statement I now have heard numerous times and which goes along these lines:

Electromagnetism is built upon a $U(1)$ Symmetry. If we consider other symmetries, we end up with other forces, for instance, if we consider $SU(2)\times U(1)$, we get the electroweak interaction.

Assuming this statement were true, I imagine something like the following to be done:

• Consider some mathematical framework along the lines of “Configuration space + Function of the latter + Axioms”
• Postulate that said function has a $U(1)$ symmetry
• End up with Maxwell's equations (or the corresponding Lagrangian or something equivalent to that)

I cannot imagine any process along these lines though. How can one postulate a symmetry and find physical laws? Hasn't it always been the other way around? That seems like complete magic to me!

• It doesn't work like that. In this case, you're specifically thinking about Yang-Mills gauge theory which is essentially defined by giving a symmetry group, but you're basically asking us here to reproduce an entire introduction to gauge theory, making this too broad. See, however, e.g physics.stackexchange.com/q/126978/50583 for a question whose answers perhaps approach the overview you're looking for here. – ACuriousMind Sep 21 '16 at 14:53
• Electromagnetism was built upon experimental evidence from the likes of Oersted and Ampere and Faraday and others, and on the mathematical skills of Maxwell and Heaviside and others. Not upon U(1) symmetry. So beware of sweeping statements. Sometimes they're horsesh*t. – John Duffield Sep 21 '16 at 15:55
• I thought, normally we should only have a correspondence between symmetries/invariants and conservation principles (e.g., laws of physics do not change (are time invariant) <=> conservation of energy; laws of physics are the same everywhere (translation invariant) <=> conservation of momentum; laws of physics are anisotropic(rotation invariant) <=> conservation of angular momentum) – Hagen von Eitzen Sep 21 '16 at 17:55
• The phrase "built upon" here does not mean discovered from. It means exist because of. It's like me saying rocks are made of atoms and someone quipping "surely that can't be right, humans must have discovered rocks before atoms" – slebetman Sep 22 '16 at 5:32

In principle there are an infinite number of Lagrangians we could choose as a guess. In practice common sense narrows the range of choices but obviously any way of narrowing this further is a great help, and that's what a gauge symmetry does. For example by requiring that our guesses for the quantum electrodynamics Lagrangian have a $U(1)$ symmetry we are led to a theory that has to have both electrons and photons - without both the symmetry would be violated. It also tells us that the photons have to be massless, which is just as well really. In fact simply by requiring the $U(1)$ symmetry the correct Lagrangian for quantum electrodynamics pretty much falls into our hands.
The other gauge symmetries work in a similar way. For QCD we guess that the gauge symmetry is $SU(3)$, and requiring that the QCD Lagrangian respect this symmetry points very strongly to the correct choice of Lagrangian for the theory. As with QED we find we have to have both quarks and gluons and it even tells us how many gluons there must be, and tells us that the gluons have to be massless, as we observe.