document.json

back to table · edit · history · where entries came from · files · download

9244 bytes, as of the version from 2026-10-31 12:06. Recorded here, not run.

{
  "Title": "Chandrasekhar H-function moments for conservative isotropic scattering",
  "Definition": "The raw angular moments of the conservative isotropic Chandrasekhar $H$-function are $\\alpha_n(d)=\\int_0^1\\mu^nH_d(\\mu)G_d(\\mu)\\,d\\mu$, where $H_d$ is the physical positive solution and $G_d$ is the normalized angular weight CITE{dEM}.",
  "Parameters": {
    "d": {
      "type": "Z",
      "title": "Spatial dimension",
      "display": "$d$",
      "constraints": "$d\\geq2$"
    },
    "n": {
      "type": "Z",
      "title": "Moment order",
      "display": "$n$",
      "constraints": "$n\\geq0$"
    }
  },
  "Comments": {
    "model": "The medium occupies the half-space $x>0$. Particles have one speed, independent exponential free paths of mean one, and uniformly distributed scattering directions. The boundary has no incoming radiation or internal reflection. The single-scattering albedo is exactly $c=1$. Plane-parallel symmetry reduces the depth coordinate without changing the spatial dimension.",
    "normalization": "The defining equation is CITE{h-equation}, with angular weight CITE{angular-weight}. The direction cosine is $\\mu\\in[0,1]$. The physical solution has $H_d(0)=1$ and is the limit of the isotropic $H$-function as $c\\to1$ from $c<1$. These are raw moments with $\\alpha_0(d)=2$. They are neither divided by $\\alpha_0(d)$ nor weighted by an additional outgoing-current factor $\\mu$. The characteristic function is $\\Psi_d=G_d/2$; that factor of one half is not part of the moment measure.",
    "geometry": "The case $d=2$ is Flatland CITE{Flatland}; $d=3$ is the usual three-dimensional transport problem, with independent numerical values in CITE{Kawabata}. Dimensions greater than three are members of the same mathematical model. The separate one-dimensional rod uses only directions $+1$ and $-1$; its moments are all 2 and are not represented by the Gamma-density formula.",
    "ingredients": "The angular-weight normalization uses HREF{Values_of_the_Gamma_function_at_rational_numbers}[$\\Gamma$]. The dispersion function $\\lambda_d$ in CITE{phase} uses the HREF{Values_of_the_Gauss_hypergeometric_function}[Gauss hypergeometric function]. The phase $\\theta_d$ is the continuous argument in $[0,\\pi]$ with $\\theta_d(0)=0$, so it stays on the physical branch when $\\lambda_d$ becomes negative.",
    "distance": "The second moment determines the HREF{Milne_extrapolation_distances_(Hopf_constant_in_three_dimensions)}[Milne extrapolation distance] $z_0(d)$ in mean-free-path units through CITE{distance-identity}. Higher angular moments carry additional information about the emerging angular distribution."
  },
  "Formulas": {
    "angular-weight": "$G_d(\\mu)=\\dfrac{2\\Gamma(d/2)}{\\sqrt\\pi\\,\\Gamma((d-1)/2)}(1-\\mu^2)^{(d-3)/2}$ for $0<\\mu<1$, with $\\int_0^1G_d(\\mu)\\,d\\mu=1$ CITE{dEM}.",
    "h-equation": "$\\dfrac1{H_d(\\mu)}=1-\\dfrac\\mu2\\int_0^1\\dfrac{H_d(v)G_d(v)}{\\mu+v}\\,dv$ for $0<\\mu\\leq1$ CITE{dEM}.",
    "lower-moments": "$\\alpha_0(d)=2$ and $\\alpha_1(d)=2/\\sqrt d$ CITE{dEM}.",
    "distance-identity": "$z_0(d)=\\alpha_2(d)/\\alpha_1(d)=\\sqrt d\\,\\alpha_2(d)/2$, where $z_0(d)$ is the HREF{Milne_extrapolation_distances_(Hopf_constant_in_three_dimensions)}[Milne extrapolation distance] CITE{dEM}.",
    "low-dimensions": "$\\alpha_2(2)=\\sqrt2(1/2+1/\\pi)$ and $\\alpha_2(4)=2/\\pi$, where HREF{Pi}[$\\pi$] is the circle constant CITE{dEM} CITE{Flatland}.",
    "reciprocal": "$\\dfrac1{H_d(z)}=1-\\dfrac12\\sum_{n=0}^{\\infty}(-1)^n\\alpha_n(d)z^{-n}$ for $|z|>1$ CITE{dEM}.",
    "phase": "$\\lambda_d(t)={}_2F_1(1,1-d/2;1/2;t^2)$, $\\theta_d(t)=\\operatorname{Arg}(\\lambda_d(t)+i\\pi tG_d(t)/2)$, and $H_d(z)=\\sqrt d(1+z)\\exp[-\\pi^{-1}\\int_0^1\\theta_d(t)/(t+z)\\,dt]$ for real $z>0$. This is the conservative limit of CITE{dEM}, equation (68).",
    "coefficient-extraction": "Let $J_k(d)=\\pi^{-1}\\int_0^1t^k\\theta_d(t)\\,dt$, $\\ell_j=(-1)^{j-1}(J_{j-1}(d)-1/j)$ for $j\\geq1$, $e_0=1$, and $e_j=j^{-1}\\sum_{k=1}^j k\\ell_k e_{j-k}$. Then $\\alpha_n(d)=2(-1)^{n+1}e_{n-1}/\\sqrt d$ for $n\\geq1$, by expanding CITE{phase} and CITE{reciprocal} at infinity.",
    "busbridge": "$4g_{2m}(d)+\\sum_{k=1}^{2m-1}(-1)^k\\alpha_{2m-k}(d)\\alpha_k(d)=0$ for $m\\geq1$, where $g_{2m}(d)=\\Gamma(d/2)\\Gamma(m+1/2)/(\\sqrt\\pi\\,\\Gamma(d/2+m))$ is the even angular-weight moment CITE{dEM}."
  },
  "References": {
    "dEM": {
      "bib": "E. d'Eon and N. J. McCormick, Radiative transfer in half spaces of arbitrary dimension (2019), equations (7), (40), (50)-(52), (56)-(68), (179)-(181), and Appendix B.",
      "arxiv": "1905.07825"
    },
    "Flatland": {
      "bib": "E. d'Eon and M. M. R. Williams, Isotropic Scattering in a Flatland Half-Space (2018), Sections 3 and 4.",
      "arxiv": "1802.02120"
    },
    "Kawabata": {
      "bib": "K. Kawabata, 15-Digit Accuracy Calculations of Chandrasekhar's H-function for Isotropic Scattering by Means of the Double Exponential Formula (2016), equations (11)-(13) and Table 6.",
      "arxiv": "1611.10197"
    }
  },
  "Similar tables": [
    {
      "table": "HREF{Milne_extrapolation_distances_(Hopf_constant_in_three_dimensions)}",
      "relation": "holds $z_0(d)=\\sqrt d\\,\\alpha_2(d)/2$ for the same conservative isotropic transport model"
    },
    {
      "table": "HREF{Values_of_the_Gamma_function_at_rational_numbers}",
      "relation": "holds Gamma values used to normalize the angular weight $G_d$"
    },
    {
      "table": "HREF{Values_of_the_Gauss_hypergeometric_function}",
      "relation": "holds values of the function defining the dispersion function $\\lambda_d$"
    },
    {
      "table": "HREF{Pi}",
      "relation": "holds the circle constant in the two- and four-dimensional second-moment formulas"
    }
  ],
  "Keywords": [
    "Ambartsumian",
    "Ambarzumian",
    "H-function",
    "angular moments",
    "radiative transfer"
  ],
  "Tags": [
    "physics",
    "integral",
    "special functions"
  ],
  "Links": {},
  "Display properties": {
    "number-header": "$\\alpha_n(d)$"
  },
  "Data properties": {
    "type": "R",
    "rigour": "heuristic (agreement-checked)",
    "rigour details": "Computed by expanding the reciprocal conservative phase factorization at infinity, with mpmath tanh-sinh integration at 150 and 200 decimal working digits and different interval subdivisions. Across all 63 moments, the largest absolute change was less than $1.3\\times10^{-151}$. Values are stored to 100 significant digits only after at least 120 digits of relative agreement. This is an agreement check, not a rigorous quadrature error bound.\n\nIndependent direct angular quadrature evaluates $H_d$ from the Fock integral in CITE{dEM}, equation (52), without using the phase-series generator. Two quadrature refinements agree with every stored moment to an absolute error less than $8\\times10^{-15}$. The known zeroth and first moments provide controls, with errors less than $1.1\\times10^{-14}$. The sine-integral control and the independent Flatland value of $H_2(1)$ precede these comparisons. These double-precision checks validate normalization and coefficient extraction; they do not establish 100-digit accuracy.\n\nThe Busbridge identities, the relation to the companion extrapolation distances, the exact two- and four-dimensional second-moment formulas, and the published three-dimensional moments in CITE{Kawabata}, Table 6, are also checked. The attached generator, independent checks, argument-selection note and numerical evidence preserve the computation and its limitations.",
    "complete": "no",
    "complete-note": "it holds every integer spatial dimension $2\\leq d\\leq10$ at every moment order $2\\leq n\\leq8$, covering Flatland, physical three-space and initial even- and odd-dimensional extensions before extending the moment hierarchy more deeply. These noncanonical bounds keep independent checks practical across both discrete indices. Orders zero and one are given by exact identities instead of repeated rows; the one-dimensional rod has all moments equal to 2 and is omitted. Absorption, anisotropic scattering, other free-path laws and direct angular values of $H_d$ are outside this family"
  },
  "Programs": {
    "mpmath-example": {
      "language": "Python",
      "code": "import mpmath as m\nm.mp.dps = 150\nd, n = 3, 2                  # dimension >= 2, moment order >= 2\nhalf = m.mpf(1)/2\nc = m.sqrt(m.pi)*m.gamma(m.mpf(d)/2)/m.gamma(m.mpf(d-1)/2)\ndef phase(t):\n    if t == 0: return m.mpf(0)\n    if t == 1: return m.pi/2 if d == 2 else m.pi\n    real = m.hyp2f1(1, 1-m.mpf(d)/2, half, t*t)\n    imag = c*t*((1-t)*(1+t))**(m.mpf(d-3)/2)\n    return m.atan2(imag, real)\nJ = [m.quad(lambda t: phase(t)*t**k, [0, half, 1])/m.pi\n     for k in range(n-1)]\nL = [m.mpf(0)] + [(-1)**(j-1)*(J[j-1]-m.mpf(1)/j)\n                  for j in range(1,n)]\ne = [m.mpf(1)]\nfor j in range(1,n):\n    e.append(m.fsum(k*L[k]*e[j-k] for k in range(1,j+1))/j)\nalpha = 2*(-1)**(n+1)*e[n-1]/m.sqrt(d)\nprint(m.nstr(alpha,100))",
      "comment": "This example computes one moment. The attached generator checks agreement at two working precisions before accepting its digits."
    }
  },
  "Numbers": {}
}