Deducing irreducible characters from non-irreducible characters

charactersrepresentation-theory

Say I have a complex character $\alpha$ of a finite group $G$ with the inner product $(\alpha, \alpha) = 2$. Since the only decomposition of $2$ as a sum of squares is $2 = 1^2 + 1^2$, the representation belonging two $\alpha$ must be a direct sum of two irreducible representations.

In the particular example I am working with, it is easy to calculate that for $\chi_1$ the trivial representation, the class function $\gamma(g) = \alpha(g) – \chi_1(g)$ has inner product $(\gamma, \gamma) = 1$. Is this enough to show that $\alpha$ is the direct sum of $\chi_1$ and $\gamma$? Since it's not even guaranteed that the representation for $\alpha$ has the subrepresentation for $\chi_1$, this does not seem like something I could deduce.

What about the case that $(\gamma, \chi_1) = 0$? Since this means that $\gamma$ is orthogonal to $\chi_1$ in the vector space of class functions on $G$, and all the irreducible characters are orthogonal to each other, does this imply $\gamma$ is a character (and thus irreducible)?

I would be interested in knowing this in the most generality possible, but I can supply concrete numbers if necessary, I just omitted them as the calculations are somewhat tedious.

Best Answer

Any difference of characters is still an integral linear combination of irreducible characters. So if your $\gamma$ has $\langle \gamma, \gamma \rangle = 1$, then $\gamma$ is $\pm 1$ times an irreducible character. In your case, it cannot be $-1$, so it is in fact an irreducible character.