A Convergent Greedy Residue Expansion for the MRB Constant

Marvin Burns

August 2026

Abstract / Overview

This notebook organizes the main structure of the paper around the residue-magnitude family
A(m)
2
π
-1/m
m
sin
π
m
, m≥2,
and the greedy procedure built from it.
The principal goals are to document and computationally support the following claims:
◼
  • the family
    A(m)
    arises naturally from the reciprocal-pole structure in the analytic theory of the MRB constant;
  • ◼
  • A(m)
    is strictly decreasing for
    m≥4
    ;
  • ◼
  • any sufficiently small residual can be greedily reduced by choosing the nearest value among neighboring residue magnitudes;
  • ◼
  • this yields a uniform contraction once the residual lies below
    A(4)1/π
    ;
  • ◼
  • every real number admits a convergent greedy expansion in signed members of the family
    A(m)
    ;
  • ◼
  • for nonterminating orbits, the residual satisfies the asymptotically quadratic estimate
  • 
    E
    n+1
    ≤
    1
    4
    +o(1)
    E
    n
    2
    |
    .
    This notebook is meant as a research notebook and proof-structure scaffold, with symbolic checks, numerical experiments, and theorem-oriented sections.

    1. The MRB constant and the residue family

    The MRB constant is defined by the alternating series
    C
    MRB
    
    ∞
    ∑
    n1
    n
    (-1)
    (
    1/n
    n
    -1).
    We store a high-precision reference value for experiments:
    In[]:=
    ClearAll[mrbN,mrbTerm,mrbPartialSum];​​​​mrbN=SetPrecision[​​0.1878596424620671202485179340542732300559030949,​​80​​];​​​​mrbTerm[n_Integer?Positive]:=(-1)^n(n^(1/n)-1)​​mrbPartialSum[n_Integer?Positive]:=Sum[mrbTerm[k],{k,1,n}]
    The residue-magnitude family is
    In[]:=
    ClearAll[a];​​a[m_Integer?Positive]:=(2/Pi)m^(-1/m)Sin[Pi/m]
    A continuous extension is useful for analysis:
    In[]:=
    ClearAll[aContinuous];​​aContinuous[x_]:=(2/Pi)x^(-1/x)Sin[Pi/x]

    2. The distinguished term
    A(8)

    A special role is played by
    A(8)
    2
    π
    -1/8
    8
    sin
    π
    8
    
    5/8
    2
    sin(π/8)
    π
    
    2-
    2
    3/8
    2
    π
    .
    Symbolic and numerical evaluation:
    In[]:=
    ClearAll[a8Exact,a8Numeric,a8Error];​​​​a8Exact=FullSimplify[a[8]]​​a8Numeric=N[a8Exact,40]​​a8Error=N[mrbN-a8Exact,40]
    Out[]=
    5/8
    2
    Sin
    π
    8
    
    π
    Out[]=
    0.1878596613801383990492532045996126144757
    Out[]=
    -1.891807127880073527054533938441983110588×
    -8
    10
    You can also compare absolute error directly:
    In[]:=
    Abs[a8Error]
    Out[]=
    1.891807127880073527054533938441983110588×
    -8
    10
    This is the starting point of the MRB greedy residue expansion.

    3. Large-
    x
    asymptotics of
    A(x)

    The continuous function
    A(x)
    2
    π
    -1/x
    x
    sin
    π
    x
    , x>1,
    has asymptotic expansion
    A(x)
    2
    x
    -
    2logx
    2
    x
    +
    2
    (logx)
    -
    2
    π
    3
    3
    x
    +O
    3
    (logx)
    4
    x
    .
    This can be explored symbolically.

    3.1 Series scaffolding

    In[]:=
    ClearAll[x];​​Normal@Series[Exp[-Log[x]/x],{x,Infinity,3}]
    Out[]=
    1-
    Log[x]
    x
    +
    2
    Log[x]
    2
    2
    x
    -
    3
    Log[x]
    6
    3
    x
    In[]:=
    Normal@Series[Sin[Pi/x],{x,Infinity,3}]
    Out[]=
    -
    3
    π
    6
    3
    x
    +
    π
    x
    You can combine these pieces manually or through asymptotic expansion workflows.

    3.2 Numerical confirmation

    In[]:=
    ClearAll[aAsym2];​​aAsym2[x_]:=2/x-2Log[x]/x^2+((Log[x])^2-Pi^2/3)/x^3​​​​Table[​​{​​n,​​N[aContinuous[n],20],​​N[aAsym2[n],20],​​N[aContinuous[n]-aAsym2[n],20]​​},​​{n,{10,20,50,100,200,500}}​​]
    Out[]=
    {{10,0.15626527733388680807,0.15596032811690103146,0.00030494921698577661760},{20,0.085735574936024959297,0.085731906597369608914,3.6683386553503826777×
    -6
    10
    },{50,0.036965348018315666278,0.036966494042547904046,-1.1460242322377679763×
    -6
    10
    },{100,0.019096710075790007003,0.019096883687110598866,-1.7361132059186248114×
    -7
    10
    },{200,0.0097381618927803906639,0.0097381819190204691748,-2.0026240078510964378×
    -8
    10
    },{500,0.0039505648361280462754,0.0039505657870980886920,-9.5097004241653252375×
    -10
    10
    }}

    4. Derivative and monotonicity

    A key formula is
    ′
    A
    (x)
    A(x)
    
    logx-1-πcot(π/x)
    2
    x
    .
    We verify this symbolically.
    In[]:=
    ClearAll[derivativeRatio];​​derivativeRatio=FullSimplify[​​D[aContinuous[x],x]/aContinuous[x],​​Assumptions->x>1​​]
    Out[]=
    -1-πCot
    π
    x
    +Log[x]
    2
    x
    To check the exact claimed form:
    In[]:=
    FullSimplify[​​derivativeRatio-(Log[x]-1-PiCot[Pi/x])/x^2,​​Assumptions->x>1​​]
    Out[]=
    0

    4.1 Numerical evidence for strict decrease on
    [4,∞)

    In[]:=
    Table[​​{n,N[a[n],30],N[a[n+1],30],a[n]>a[n+1]},​​{n,4,20}​​]

    4.2 Plot of the continuous extension

    For derivative sign:

    5. Discrete spacing of successive residue magnitudes

    A scaled error check:
    The values should trend toward 2.
    Symbolic verification:
    And numerically:

    6.1 Bracketing a residual

    Example checks:

    7. Greedy step definition

    7.1 Nearest-residue index

    7.2 One greedy update

    8. Global greedy expansion algorithm

    Display as a table:
    Example use:

    10. Predictor-based search for very large greedy indices

    10.1 Predictor formula

    10.2 Local search near the predictor

    This is the practical tool for reproducing the huge indices reported in the paper.

    11. High-precision predictive greedy computation for MRB

    Presentation grid:
    Example:

    12. Verifying entry into the contraction range

    Examples:

    13. Testing the contraction numerically

    We can test this experimentally.
    Suggested experiments:

    14. Testing the quadratic regime

    Example for MRB:
    For predictive large-index computations:

    15. Universal Greedy Residue Expansion Theorem (UGRET)

    This section records the theorem in notebook form.

    Theorem (UGRET)

    ◼
  • some residual $E_N$ vanishes, giving a finite exact representation, or
  • 16. Nontermination for nonzero rational targets

    This is an important conceptual point:
    ◼
  • nontermination alone does not prove irrationality of the target;
  • Notebook note

    This theorem is best kept as a conceptual proof cell rather than a computational one.

    17. Suggested experiments

    17.1 Compare direct MRB partial sums with the greedy approximation

    17.2 Compare first greedy terms

    17.3 Inspect monotonicity numerically at large scale

    17.4 Inspect asymptotic slope of the spacing law

    17.5 Inspect the quadratic constant experimentally

    18. Paper-structure summary

    This notebook separates the argument into four layers:
    1. Origin of the family
    2. Monotonicity and spacing
    3. Global convergence
    4. Asymptotic acceleration
    Once convergence is known, the spacing estimate yields asymptotically quadratic residual decay.
    This makes the greedy residue expansion both analytically motivated and mathematically robust.

    19. Documentation links

    Useful Wolfram Language references:

    20. Closing summary

    For the MRB constant specifically:
    ◼
  • subsequent greedy choices from the same family reduce the residual very rapidly;
  • ◼
  • the convergence is globally guaranteed by a contraction argument;
  • ◼
  • the near-doubling of correct digits is explained by the asymptotic spacing of neighboring residue magnitudes.
  • If you want, I can next turn this into one of these:
    1. a paper-style notebook with theorem/lemma/proof formatting,
    2. a fully executable experimental notebook with evaluated cells and plots,
    3. a seminar notebook with shorter slide-like sections,
    4. or a proof-outline notebook separating rigorous arguments from numerical evidence.

    Acknowledgements

    ​
    This investigation developed through a combination of analytic calculation, high - precision numerical experimentation, and symbolic verification . ChatGPT, developed by OpenAI, was used interactively during the development of this work
    to assist in exploring contour representations, branch - cut transformations, residue calculations, asymptotic expansions, and related analytic questions, as well as in the preparation and revision
    of portions of the manuscript . Wolfram Mathematica was used for high - precision numerical computation and symbolic verification where appropriate . All mathematical claims, derivations, numerical results, and conclusions presented in the final
    manuscript were reviewed and accepted by the author, who assumes full responsibility for the content of the work .
    The Wolfram Notebook Assisted designed this notebook.