Counting the Eisenstein series with character #
The normalized raised Eisenstein series E_k^(ψ, φ, t) of weight k and level N are indexed
by TauCeti.EisensteinSeries.CharIndex N k: primitive characters ψ modulo u and φ modulo
v with ψ(-1) φ(-1) = (-1)^k, and a raising parameter t with t u v ∣ N. This file counts
these indices. Away from the levels 1, 2 and 4 there are exactly as many as there are
cusps of Γ₁(N):
2 · #CharIndex N k = ∑_{g ∣ N} totient(g) totient(N / g) = 2 ε∞(Γ₁(N)),
where totient is Euler's function Nat.totient.
The series are linearly independent, so in weight k ≥ 3 the Eisenstein subspaces
TauCeti.eisensteinSubspace of the spaces M_k(N, χ) have dimensions summing to this number
(TauCeti.finrank_eisensteinSubspace). Comparing it with the number of cusps, which bounds the
dimension of the Eisenstein part of M_k(Γ₁(N)) through the constant terms, is how
Diamond–Shurman show that the series span that part (Theorem 4.5.2); this file supplies the
count of the series.
The count goes through a bijection. Raising ψ and φ to the levels g = t u and N / g
sends an index to a divisor g ∣ N and a pair of characters modulo g and N / g of
product parity (-1)^k; conversely, u and v are the conductors of such a pair, ψ and φ
their primitive characters, and t = g / u. This gives the count card_eq_sum at every
level. When g or N / g exceeds 2, an odd character at that level exchanges the two
parities, so exactly half of the totient(g) totient(N / g) pairs have parity (-1)^k
(DirichletCharacter.two_mul_card_apply_neg_one_mul_apply_neg_one_eq). Every divisor g of
N ∉ {1, 2, 4} has this property.
At the levels 1, 2 and 4 the count is direct. Modulo 1 and 2 the only character is
trivial and even, so a pair of such characters has parity (-1)^k exactly when k is even. There
are 1, 2 and 3 indices in even weight, and 0, 0 and 2 in odd weight. In odd weight the
level-4 count is one less than the three cusps of Γ₁(4), matching its irregular cusp.
Main results #
TauCeti.EisensteinSeries.CharIndex.card_eq_sum: the number of indices as a sum over the divisorsgofNof the number of character pairs modulogandN / gwith product parity(-1)^k.TauCeti.EisensteinSeries.CharIndex.two_mul_card:2 · #CharIndex N k = ∑_{g ∣ N} totient(g) totient(N / g)forN ∉ {1, 2, 4}.TauCeti.EisensteinSeries.CharIndex.card_eq_card_cuspOrbit_Gamma1: forN ∉ {1, 2, 4}, the indices are as many as the cusps ofΓ₁(N).TauCeti.EisensteinSeries.CharIndex.card_one,TauCeti.EisensteinSeries.CharIndex.card_two,TauCeti.EisensteinSeries.CharIndex.card_four: the counts at the levels1,2and4.
References #
- F. Diamond and J. Shurman, A first course in modular forms,
Section 4.5, Theorem 4.5.2, and Section 3.8 for the cusps of
Γ₁(N).
The number of Eisenstein indices at level N: indices (ψ, φ, t) correspond to the
divisors g = t u of N together with a pair of characters modulo g and N / g whose product
parity is (-1)^k.
The Eisenstein indices are as many as the cusps of Γ₁(N) away from the levels 1,
2 and 4.