Node derivatives and the Euler–Poisson system for L #
The native L-integral satisfies Carlson (1987), (2.7) and (3.5). Node analyticity and the Euler–Poisson system are instances of the general Dirichlet average theorems, not independent integration-by-parts proofs.
The power-logarithm kernel is holomorphic on the principal slit plane.
theorem
DirichletTransform.hasDerivAt_carlsonLKernel
(t : ℂ)
{w : ℂ}
(hw : w ∈ Complex.slitPlane)
:
HasDerivAt (carlsonLKernel t) (t * carlsonLKernel (t - 1) w + w ^ (t - 1)) w
Differentiating the kernel lowers the exponent and adds a power kernel.
theorem
DirichletTransform.continuousOn_carlsonLKernel_affine
{ι : Type u_1}
[Fintype ι]
(t : ℂ)
{z : ι → ℂ}
(hz : z ∈ carlsonRVariableDomain)
:
ContinuousOn (fun (u : ι → ℝ) => carlsonLKernel t (carlsonAffineForm z u)) (Convexity.StdSimplex.coordinateSet ℝ ι)
The log-power integrand is continuous on the compact simplex.
theorem
DirichletTransform.analyticOnNhd_regCarlsonLIntegral_nodes
{ι : Type u_1}
[Fintype ι]
(t : ℂ)
{b : ι → ℂ}
(hb : b ∈ Complex.mvBetaConvergent)
:
On the native convergence region, L is holomorphic jointly in all nodes.
theorem
DirichletTransform.carlsonEulerPoissonOperator_regCarlsonLIntegral
{ι : Type u_1}
[Fintype ι]
(t : ℂ)
{b z : ι → ℂ}
(hb : b ∈ Complex.mvBetaConvergent)
(hz : z ∈ carlsonRVariableDomain)
(i j : ι)
:
Equation (2.7): the complete Euler–Poisson system, including equal indices.
theorem
DirichletTransform.regCarlsonDirichletAverage_deriv_LKernel
{ι : Type u_1}
[Fintype ι]
(t : ℂ)
{b z : ι → ℂ}
(hb : b ∈ Complex.mvBetaConvergent)
(hz : z ∈ carlsonRVariableDomain)
:
regCarlsonDirichletAverage b z (deriv (carlsonLKernel t)) = t * regCarlsonLIntegral (t - 1) b z + regCarlsonRIntegral (t - 1) b z
Averaging the derivative of the kernel gives the inhomogeneous lowering formula.
theorem
DirichletTransform.carlsonPartialDeriv_regCarlsonLIntegral
{ι : Type u_1}
[Fintype ι]
(t : ℂ)
{b z : ι → ℂ}
(hb : b ∈ Complex.mvBetaConvergent)
(hz : z ∈ carlsonRVariableDomain)
(i : ι)
:
carlsonPartialDeriv i (regCarlsonLIntegral t b) z = b i * (t * regCarlsonLIntegral (t - 1) (addDirichletUnit b i) z + regCarlsonRIntegral (t - 1) (addDirichletUnit b i) z)
Equation (3.5), in regularized form.