aub_htp.pdf.skorohod#

Functions

generate_pdf_skorohod_vectorized(x, alpha, beta)

Vectorized dispatcher that combines Skorohod formulas into a stable pdf approximation across parameter regimes.

skorohod_formula_1(x, alpha, beta[, N])

Skorohod's Formula 1: series for 0 < alpha < 1 (tails).

skorohod_formula_2(x, beta[, N])

Skorohod's Formula 2: alpha = 1 case (Cauchy-like).

skorohod_formula_2_bk(beta, k)

Helper for Skorohod's Formula 2 (alpha = 1): compute b_k(beta).

skorohod_formula_3(x, alpha, beta[, N])

Skorohod's Formula 3: series for 1 < alpha < 2 (tails).

skorohod_formula_3_an(alpha, beta, n)

Coefficient a_n(alpha, beta) for Skorohod's Formula 3 (1 < alpha < 2 tails).

skorohod_formula_4(x, alpha)

Skorohod's Formula 4: small-|x| asymptotic for 0 < alpha < 1.

skorohod_formula_4_A(alpha)

Asymptotic prefactor A(alpha) for 0 < alpha < 1 near zero (β = ±1 edges).

skorohod_formula_4_B(alpha)

Asymptotic exponent scale B(alpha) for 0 < alpha < 1 near zero.

skorohod_formula_4_Lambda(alpha)

Lambda(alpha) = alpha / (1 - alpha) for 0 < alpha < 1.

skorohod_formula_5(x)

Skorohod's Formula 5: alpha = 1, near zero correction (heuristic).

skorohod_formula_6(x, alpha)

Skorohod's Formula 6: small-|x| asymptotic for alpha > 1.

skorohod_formula_6_A_prime(alpha)

Asymptotic prefactor A'(alpha) for alpha > 1 near zero (β = ±1 edges).

skorohod_formula_6_B_prime(alpha)

Asymptotic exponent scale B'(alpha) for alpha > 1 near zero.

skorohod_formula_6_lambda_prime(alpha)

Lambda'(alpha) = alpha / (alpha - 1) for alpha > 1.

aub_htp.pdf.skorohod.generate_pdf_skorohod_vectorized(x, alpha, beta, epsilon=1e-06)#

Vectorized dispatcher that combines Skorohod formulas into a stable pdf approximation across parameter regimes.

Inputs - x: array of evaluation points (after normalization if needed). - alpha ∈ (0, 2], beta ∈ [-1, 1]. - epsilon: small threshold around zero to switch to near-zero asymptotics.

Logic - alpha < 1:

  • beta = 1: use Formula 1 on x>ε; Formula 4 near zero on (0, ε].

  • beta = -1: mirror to left tail; use |x| and flip beta where needed.

  • else: use Formula 1 on both sides, flipping beta for x<=0.

  • alpha = 1:
    • beta ∈ [-1,1]:
      • main: Formula 2 on each side with |x| as needed.

      • near zero: Formula 5 to stabilize behavior.

  • alpha > 1:
    • beta = 1: Formula 3 on x>0; Formula 6 near/below zero.

    • beta = -1: swap sides; Formula 6 on x>0; Formula 3 on x<=0 with flip.

    • else: Formula 3 on both sides, flipping beta for x<0.

Returns - result: array of pdf approximations, same shape as x.

aub_htp.pdf.skorohod.skorohod_formula_1(x, alpha, beta, N=170)#

Skorohod’s Formula 1: series for 0 < alpha < 1 (tails).

  • Uses truncated series with N terms.

  • Valid only when 0 < alpha < 1 and |beta| <= 1.

  • Works on x > 0 for right tail; use |x| and flipped beta for left tail.

Returns vector of pdf approximations at x.

aub_htp.pdf.skorohod.skorohod_formula_2(x, beta, N=10)#

Skorohod’s Formula 2: alpha = 1 case (Cauchy-like).

  • Adjust x by log term from skewness.

  • Use truncated inverse-power series with coefficients b_k(beta).

  • Valid for x ≠ 0; typically used on x > 0 and mirrored for x < 0.

aub_htp.pdf.skorohod.skorohod_formula_2_bk(beta, k)#

Helper for Skorohod’s Formula 2 (alpha = 1): compute b_k(beta).

  • Integrates e^{-v} v^k * Im[ 1j + beta*1j - (2*beta/π) ln v ] dv over v∈[0, ∞).

  • Caches results by (beta, k) to speed up repeated calls.

aub_htp.pdf.skorohod.skorohod_formula_3(x, alpha, beta, N=20)#

Skorohod’s Formula 3: series for 1 < alpha < 2 (tails).

  • Uses a_n(alpha, beta) with N terms.

  • Typically applied to |x| large.

aub_htp.pdf.skorohod.skorohod_formula_3_an(alpha, beta, n)#

Coefficient a_n(alpha, beta) for Skorohod’s Formula 3 (1 < alpha < 2 tails).

  • Returns array for vector n.

  • a_n = (-1)^{n-1} (1 + β^2 tan^2(πα/2))^{n/2} sin(n(πα/2 + atan(β tan(πα/2)))) Γ(nα+1)/n!

aub_htp.pdf.skorohod.skorohod_formula_4(x, alpha)#

Skorohod’s Formula 4: small-|x| asymptotic for 0 < alpha < 1.

  • Used when x→0 for extreme skew β=±1 handling.

  • Returns the core exponential form A x^{-1-λ/2} exp(-B x^{-λ}).

aub_htp.pdf.skorohod.skorohod_formula_4_A(alpha)#

Asymptotic prefactor A(alpha) for 0 < alpha < 1 near zero (β = ±1 edges).

aub_htp.pdf.skorohod.skorohod_formula_4_B(alpha)#

Asymptotic exponent scale B(alpha) for 0 < alpha < 1 near zero.

aub_htp.pdf.skorohod.skorohod_formula_4_Lambda(alpha)#

Lambda(alpha) = alpha / (1 - alpha) for 0 < alpha < 1.

aub_htp.pdf.skorohod.skorohod_formula_5(x)#

Skorohod’s Formula 5: alpha = 1, near zero correction (heuristic).

  • Empirical correction term to stabilize behavior around 0.

  • Not a strict series; acts as a patch for numerical issues.

aub_htp.pdf.skorohod.skorohod_formula_6(x, alpha)#

Skorohod’s Formula 6: small-|x| asymptotic for alpha > 1.

  • Used when x→0 for extreme skew β=±1 handling.

  • Returns A’ x^{-1+λ’/2} exp(-B’ x^{λ’}).

aub_htp.pdf.skorohod.skorohod_formula_6_A_prime(alpha)#

Asymptotic prefactor A’(alpha) for alpha > 1 near zero (β = ±1 edges).

aub_htp.pdf.skorohod.skorohod_formula_6_B_prime(alpha)#

Asymptotic exponent scale B’(alpha) for alpha > 1 near zero.

aub_htp.pdf.skorohod.skorohod_formula_6_lambda_prime(alpha)#

Lambda’(alpha) = alpha / (alpha - 1) for alpha > 1.