Dirichlet L-function with the principal pole removed
DefinitionLL_Decomp530a_regularPartFor and a Dirichlet character modulo , the function is defined as follows. If is the principal character, let be the entire extension of , whose value at is the residue , and set for and . Otherwise , the entire Dirichlet -function.
import Mathlib
open Classical Complex
namespace LL_Decomp530a
noncomputable def regularPart {q : ℕ} [NeZero q] (χ : DirichletCharacter ℂ q) : ℂ → ℂ := by
classical
exact if χ = 1 then dslope (DirichletCharacter.LFunctionTrivChar₁ q) 1 else χ.LFunction
end LL_Decomp530aRead-back
What the Lean code literally says, in plain math · claude-haiku
Declaration LL_Decomp530a.regularPart (a noncomputable definition, not a theorem, so it asserts no proposition and proves nothing). It takes a natural number with the instance NeZero q, which means , and a Dirichlet character modulo with values in , that is, a multiplicative map from to extended by on non-units. It returns a function , written . The body first enables classical decidability, then splits on whether equals the trivial character modulo (the character that is on integers coprime to and otherwise). If , the function is , where LFunctionTrivChar₁ q is a function whose definition is not shown in this item; dslope is the difference quotient for and for , so here
If , the function is simply , the Dirichlet -function as Mathlib defines it (by analytic continuation), with no subtraction. Every case is total on , so no value is undefined, and there are no hypotheses beyond and the character itself. Concerns for the auditor: the item is a definition only, so nothing is claimed about it. For nontrivial the output is the full -function, which is entire, so nothing is removed, as expected. For the trivial character, a mathematician would normally expect the regular part of at its simple pole , namely , which is holomorphic at and equals the constant term of the Laurent expansion at . The Lean code instead returns the difference quotient of at , and this equals the expected object only if LFunctionTrivChar₁ q is defined in a way that makes this difference quotient coincide with the regular part; whether it is, depends on the Mathlib definition of LFunctionTrivChar₁, which I did not inspect because the auditing instructions limit me to this item. If is itself the regular part, then the trivial branch returns a difference quotient rather than , and the two differ away from . The value at is a derivative, not the constant term of a Laurent expansion, unless the definition of is arranged to make these agree. The item contains no theorem, so no inequality direction, quantifier, or iff needs checking.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.