# Identity as a trivial reducible representation

In particle physics, I was taught that a representation of a group is a function $r: group \rightarrow matrices\,(n\times n)$ such that $r(g_1)r(g_2)=r(g_1g_2)$ and $r(e)=I_{n\times n}$. Then, that a representation is reducible when you can find a matrix $A$ such that $Ar(g)A^{-1}$ is in diagonal-block form for every element of the group.

Then the professor tried to find in complicated ways reducible representations of $SO(N)$, $SU(N)$ and so on. But the trivial function that assigns $I_{n\times n}$ to every value of $g$ is not already a reducible transformation? I know it must be somehow useless, but what did I lose?

-
If $n>1$ then yes, the identity representation is trivially reducible and irreducibly trivial. :) You are saying that if you have a bunch of objects which don't transform at all under the group then they don't transform into each other. This is a true statement but we tend to leave such trivial cases out of the discussion. – Michael Brown Feb 13 '13 at 1:31
Yes that would be a reducible representation decomposed into direct sums of the trivial (singlet) representation $1\oplus\dots\oplus 1$. What do you mean by what you lost? You can construct reducible representations by taking direct sums of irreducible ones in all sort of ways, so this is just the most trivial example of them all (using only the singlet representation). We are typically much more interested in irreducible representations. – Heidar Feb 13 '13 at 1:31

What you have constructed is a representation, but not a faithful one. Since your homomorphism $r$ is not injective, you lose some of the structure of the group. In fact, since $r$ is trivial, you lose all the structure of the group. While most useful statements about $G$ apply to $r(G)$ equally well, you cannot pull back anything useful from $r(G)$ to $G$, so your representation doesn't tell you anything about $G$, defeating the whole purpose of using representations in the first place.
For example, addition of angular momentum in quantum mechanics involves writing reducible representations of $\mathrm{SU}(2)$ as directs sums of irreducible ones.