Using multiplicativity of Hecke operators, $\otherI$ contains $T_n - e_n$, where the $e_{\ell}$ for $\ell$ prime are $1+\ell$ or $\pm 1$ as above, and we extend using the recursive formula for Hecke operators: $T_{p^r} = T_p T_{p^{r-1}} - \eps(p) p T_{p^{r-2}}$, where $\eps(p)=1$ unless $p=13, 43$ in which case $\eps(p)=0$. The first few $e$ are: %link \begin{lstlisting} sage: e = [0,1,3,4,7,6,12,8,15,13,18,12,28,-1,24,24,31,18,39,20,42,32,36,24,60,31,-3,40, ... 56,30,72,32,63,48,54,48,91,38,60,-4,90,42,96,1,84,78,72,48,124,57,93,72,-7,54,120,72, ... 120,80,90,60,168,62,96,104,127,-6,144,68,126,96,144,72,195,74,114,124,140,96,-12,80] \end{lstlisting} %The above was computed using the following too-complicated code! % def evec(B=hecke_bound): % # compute the vector of e, so e[n] = e_n above for all n<=hecke_bound % e = [0,1] + [None for i in range(B-1)] % # first primes: % for p in primes(B+1): % if p == 13: % e[p] = -1 % elif p == 43: % e[p] = 1 % else: % e[p] = p+1 % # now powers of p % k = 2 % while p^k <= B: % if p == 13 or p == 43: % e[p^k] = e[p]^k % else: % e[p^k] = e[p^(k-1)]*e[p] - p*e[p^(k-2)] % k += 1 % # next composite non-prime powers: extend multiplicatively % for n in [1..B]: % if not n.is_prime_power(): % e[n] = prod(e[F[0]^F[1]] for F in factor(n)) % return e Taking the $\Z$-module $M$ generated by $T_n-e_n$, for $n\leq 52$, gives a $\Z$-submodule $M \subset \otherI$, but it turns out that $M\neq \otherI$. However, the ideal generated by $M$ equals $\otherI$, since ...TODO Using $T_n-e_n$ for many more $n$ does not help. %link \begin{lstlisting} sage: def eisenstein_ideal(v, B): return [v * (S.hecke_matrix(i) - e[i]) for i in [1..B]] sage: M = eisenstein_ideal(v, 52) sage: factor(span(ZZ,M).index_in(L)) 2 * 3 * 7^2 sage: M = eisenstein_ideal(v, 103) sage: factor(span(ZZ,M).index_in(L)) 2 * 7^2 \end{lstlisting}%link To determine whether or not a candidate $\Z$-module $M$ is invariant under the action of $\T$, it suffices to check that $M$ is preserved by $T_p$ for all primes $p<52$, since ... The trick we use to do this is to note that for $t \in M$ we have $v (T_p t) = (v T_p) t$, so to compute $T_p M$ we just compute the image $M_p$ of $M$ under the embedding defined by $v T_p$ instead of the embedding defined by $v$. If the $\Z$-modules $M_p$ is contained in $M$, for $p<52$, then we conclude that $M$ is a $\T$-module. More precisely, since the $M$ above is not a $\T$-module, we add to it all module $M_P$ to obtain a $\T$-module. %link \begin{lstlisting} \end{lstlisting}%link %\end{lstlisting}%link %link %\begin{lstlisting} %\begin{lstlisting}\end{lstlisting}