enumitem: enumerate in a theorem environment

Question

I use the enumitem package:

\usepackage[shortlabels]{enumitem}

And I put some numeration like this in a theorem environment with \theoremstyle{plain}:

\begin{prop}
    We have
        \begin{enumerate}[(i)]
            \item $χ(c_M)=-\Id_{σM}$,

        and if $π_0(M)=0$, 
            \item $χ(f_M)=-α_M^{-1}$.
        \end{enumerate}
\end{prop}

the [(i)] after \begin{enumerate} is supposed to make items indexed by (i)(ii)(iii). But in this theorem environment, as texts are italic, the (i)(ii)(iii) become funny things like (i)(ii)(iii). How do I fix it?

Answer 1

Instead of using the short labels method, you should specify the label with the key-value syntax that enumitem provides:

\begin{enumerate}[label={\upshape(\roman*)}]

Answer 2

You can also use the following:

\begin{enumerate}[label={\upshape(\roman*)}, align=left, widest=iii, leftmargin=*]

to get left-aligned roman (i), (ii), (iii), with no leftmargin, and the widest spacing taken to be (iii).