Any space of modular forms that can be created in Magma is of the form MZ tensor Z R for some ring R and some space MZ of modular forms defined over Z. The basis of M is the image in M of Basis(M_Z), and Basis(M_Z) is in Hermite normal form.
Note that in order to determine this basis over Z, it is necessary to compute q-expansions up to a "Sturm bound" for M (see PrecisionBound). The precision used internally will therefore be at least as large as this bound. If desired, however, one can compute q-expansions to lower precision by working directly with spaces of modular symbols.
The canonical basis of the space of modular forms or half-integral weight forms M.
A sequence containing q-expansions (to the specified precision) of the elements of Basis(M).
Exact: BoolElt Default: false
An integer b such that f + O(qb) determines any modular form f in the given space M of modular forms or half-integral weight forms. If the optional parameter Exact is set to true, or if a q-expansion basis has already been computed for M, then the result is best-possible, ie the smallest integer b such that f + O(qb) determines any modular form f in M. Otherwise it is a "Sturm bound" similar to (although in some cases sharper than) the bounds given in section 9.4 of [Ste07].Note: In some much older versions of Magma the default was Exact := true.
Ring: Rng Default:
An abstract free module isomorphic to the given space of modular forms M, over the same base ring (unless Ring is specified). The second returned object is a map from the abstract module to/from M.This function is needed when one wants to use linear algebra functions on M (since in Magma, a space of modular forms is not a subtype of vector space).
> M := ModularForms(Gamma1(16),3); M; Space of modular forms on Gamma_1(16) of weight 3 and dimension 23 over Integer Ring. > Dimension(CuspidalSubspace(M)); 9 > SetPrecision(M,19); > Basis(NewSubspace(CuspidalSubspace(M)))[1]; q - 76*q^8 + 39*q^9 + 132*q^10 - 44*q^11 + 84*q^12 - 144*q^13 - 232*q^14 + 120*q^15 + 160*q^16 + 158*q^17 - 76*q^18 + O(q^19)We can print the whole basis to less precision:
> SetPrecision(M,10); > Basis(NewSubspace(CuspidalSubspace(M))); [ q - 76*q^8 + 39*q^9 + O(q^10), q^2 + q^7 - 58*q^8 + 30*q^9 + O(q^10), q^3 + 2*q^7 - 42*q^8 + 18*q^9 + O(q^10), q^4 + q^7 - 26*q^8 + 13*q^9 + O(q^10), q^5 + 2*q^7 - 18*q^8 + 5*q^9 + O(q^10), q^6 + 2*q^7 - 12*q^8 + 3*q^9 + O(q^10), 3*q^7 - 8*q^8 + O(q^10) ]Note the coefficient 3 of q7, which emphasizes that this is not the reduced echelon form over a field, but the image of a reduced form over the integers:
> MQ := BaseExtend(M,RationalField()); > Basis(NewSubspace(CuspidalSubspace(MQ)))[7]; 3*q^7 - 8*q^8 + O(q^10)