Math 2.0

Number theory Proof

Proof of Kummer's theorem from Legendre's formula

Builds on P-adic valuationLegendre's formulaFactorial formula for binomial coefficientsP-adic valuation of a productDivision algorithmFloor functionBinomial coefficient

Proof

  1. Let pp be a prime and let m,nm, n be natural numbers. For a natural number xx with base-pp digits x=∑j≥0xjpjx = \sum_{j \ge 0} x_j p^j, 0≤xj<p0 \le x_j < p, let s(x)=∑jxjs(x) = \sum_j x_j be its digit sum.
  2. Base-pp digits are unique. If ∑jajpj=∑jbjpj\sum_j a_j p^j = \sum_j b_j p^j with all digits in {0,…,p−1}\set{0, \dots, p-1}, the division algorithm by pp gives a0=b0a_0 = b_0. Subtracting, dividing by pp and repeating gives aj=bja_j = b_j for all jj.
  3. Fix xx as in step 1 and i≥1i \ge 1. Write x=piA+Bx = p^i A + B with A=∑j≥ixjpj−iA = \sum_{j \ge i} x_j p^{j-i} and B=∑j<ixjpjB = \sum_{j < i} x_j p^j.
  4. Then 0≤B≤∑j<i(p−1)pj=pi−10 \le B \le \sum_{j<i} (p-1) p^j = p^i - 1. So A≤x/pi<A+1A \le x / p^i < A + 1, and A=⌊x/pi⌋A = \lfloor x/p^i \rfloor by the definition of the floor function.
  5. By Legendre's formula and step 4, νp(x!)=∑i≥1∑j≥ixjpj−i=∑j≥1xj(pj−1+⋯+p+1)\nu_p(x!) = \sum_{i \ge 1} \sum_{j \ge i} x_j p^{j-i} = \sum_{j \ge 1} x_j (p^{j-1} + \dots + p + 1).
  6. Since pj−1+⋯+1=(pj−1)/(p−1)p^{j-1} + \dots + 1 = (p^j - 1)/(p - 1), this gives νp(x!)=∑jxj(pj−1)/(p−1)=(x−s(x))/(p−1)\nu_p(x!) = \sum_j x_j (p^j - 1)/(p-1) = (x - s(x))/(p - 1).
  7. By the factorial formula, (m+nm) m! n!=(m+n)!\binom{m+n}{m} \, m! \, n! = (m+n)!. By the valuation of a product and step 6, νp(m+nm)=s(m)+s(n)−s(m+n)p−1\nu_p \binom{m+n}{m} = \dfrac{s(m) + s(n) - s(m+n)}{p - 1}.
  8. Now we describe addition in base pp. Choose KK with m+n<pKm + n < p^K, and write m=∑j=0Kmjpjm = \sum_{j=0}^{K} m_j p^j and n=∑j=0Knjpjn = \sum_{j=0}^{K} n_j p^j with digits in {0,…,p−1}\set{0, \dots, p-1}.
  9. Set c−1=0c_{-1} = 0. For j=0,1,…,Kj = 0, 1, \dots, K in turn, let cj=1c_j = 1 if mj+nj+cj−1≥pm_j + n_j + c_{j-1} \ge p and cj=0c_j = 0 otherwise. Put rj=mj+nj+cj−1−pcjr_j = m_j + n_j + c_{j-1} - p c_j.
  10. By induction on jj, cj∈{0,1}c_j \in \set{0, 1} and 0≤rj≤p−10 \le r_j \le p - 1. Indeed 0≤mj+nj+cj−1≤2p−10 \le m_j + n_j + c_{j-1} \le 2p - 1. The carry cj=1c_j = 1 is exactly a carry at place jj, so the number of carries is C=∑j=0KcjC = \sum_{j=0}^{K} c_j.
  11. Summing rjpjr_j p^j over jj, the carry terms telescope: ∑j=0Krjpj=m+n+c−1−cKpK+1=m+n−cKpK+1\sum_{j=0}^{K} r_j p^j = m + n + c_{-1} - c_K p^{K+1} = m + n - c_K p^{K+1}.
  12. The left side is at least 00. If cK=1c_K = 1, this gives m+n≥pK+1m + n \ge p^{K+1}, against m+n<pKm + n < p^K. So cK=0c_K = 0, and ∑j=0Krjpj=m+n\sum_{j=0}^{K} r_j p^j = m + n.
  13. By step 2, the rjr_j are the base-pp digits of m+nm + n. So s(m+n)=∑jrjs(m+n) = \sum_j r_j.
  14. Summing the definition of rjr_j over jj gives s(m+n)=s(m)+s(n)+∑j=0Kcj−1−p∑j=0Kcjs(m+n) = s(m) + s(n) + \sum_{j=0}^{K} c_{j-1} - p \sum_{j=0}^{K} c_j.
  15. Since c−1=0c_{-1} = 0 and cK=0c_K = 0, we have ∑j=0Kcj−1=C\sum_{j=0}^{K} c_{j-1} = C. So s(m)+s(n)−s(m+n)=pC−C=(p−1)Cs(m) + s(n) - s(m+n) = pC - C = (p - 1) C.
  16. With step 7, νp(m+nm)=C\nu_p \binom{m+n}{m} = C, the number of carries when one adds mm and nn in base pp.

Idea

Legendre's formula can be rewritten as νp(x!)=(x−s(x))/(p−1)\nu_p(x!) = (x - s(x))/(p - 1), with s(x)s(x) the base-pp digit sum. So the valuation of (m+nm)\binom{m+n}{m} measures how much the digit sum drops when one adds mm and nn. Each carry replaces pp in one place by 11 in the next place, so it lowers the digit sum by exactly p−1p - 1.

Reading path

To understand this page from the ground up, read these 76 pages in order. The foundations come first. The path takes the proofs Proof of Bézout's identity, Proof of the approximation property of the supremum, Proof of the generalized Euclid's lemma, Proof of the commutativity of addition by induction, Proof of Euclid's lemma, Proof that the reals form an ordered field, with the Cauchy sequence construction, Proof of strong induction from ordinary induction, Proof that the p-adic valuation of a product is the sum of the valuations, Proof of Pascal's rule from the recursive definition, Proof of the factorial formula for binomial coefficients by induction, Proof of the well-ordering principle from strong induction, Proof of the division algorithm, Proof that the reals are Archimedean from the completeness axiom, Proof of Legendre's formula by counting multiples of prime powers.

  1. Binomial coefficient definition
  2. Construction of the reals from Cauchy sequences concept
  3. Definition by recursion concept
  4. Divisibility definition
  5. Element definition
  6. Factorial definition
  7. Integers object
  8. Mathematical induction concept
  9. Natural numbers object
  10. Ordered field definition
  11. Peano axioms concept
  12. Proposition concept
  13. Rational numbers object
  14. Real numbers object
  15. Set bounded above definition
  16. Proof of Pascal's rule from the recursive definition proof
  17. Greatest common divisor definition
  18. Prime number definition
  19. Set definition
  20. Ring of integers statement
  21. Archimedean ordered field definition
  22. Successor function definition
  23. Implication definition
  24. Predicate definition
  25. The reals form a field statement
  26. Supremum definition
  27. Pascal's rule statement
  28. Coprime integers definition
  29. P-adic valuation definition
  30. Ordered pair definition
  31. Addition of natural numbers definition
  32. The rationals form a field statement
  33. The successor function is injective statement
  34. Universal quantifier definition
  35. Zero is not a successor statement
  36. Completeness axiom statement
  37. Proof of the factorial formula for binomial coefficients by induction proof
  38. Cartesian product definition
  39. Subset definition
  40. Proof of the commutativity of addition by induction proof
  41. Order on the natural numbers definition
  42. The rationals form an ordered field statement
  43. Binary relation definition
  44. Factorial formula for binomial coefficients statement
  45. Order on the reals definition
  46. Commutativity of addition statement
  47. Proof of strong induction from ordinary induction proof
  48. Antisymmetric relation definition
  49. Reflexive relation definition
  50. Transitive relation definition
  51. Strong induction statement
  52. Preorder definition
  53. Proof of the well-ordering principle from strong induction proof
  54. Partial order definition
  55. Well-ordering principle statement
  56. Total order definition
  57. Proof of the division algorithm proof
  58. Division algorithm statement
  59. Proof that the reals form an ordered field, with the Cauchy sequence construction proof
  60. Proof of Bézout's identity proof
  61. The reals form an ordered field statement
  62. Bézout's identity statement
  63. Proof of the approximation property of the supremum proof
  64. Proof of the generalized Euclid's lemma proof
  65. Approximation property of the supremum statement
  66. Generalized Euclid's lemma statement
  67. Proof that the reals are Archimedean from the completeness axiom proof
  68. Proof of Euclid's lemma proof
  69. The reals are Archimedean statement
  70. Euclid's lemma statement
  71. Floor function definition
  72. Proof that the p-adic valuation of a product is the sum of the valuations proof
  73. P-adic valuation of a product statement
  74. Proof of Legendre's formula by counting multiples of prime powers proof
  75. Legendre's formula statement
  76. Proof of Kummer's theorem from Legendre's formula proof

Where this leads

1 page builds on this one.

Further reading

  1. Web https://en.wikipedia.org/wiki/Kummer%27s_theorem, Statement
  2. Web https://en.wikipedia.org/wiki/Legendre%27s_formula, Alternate form
About this page
Tag9933XVSBPV
URIm2://9933XVSBPV
Kindproof
WritingW0 stub
MSC11B65
ProvesKummer's theorem
UsesP-adic valuation, Legendre's formula, Factorial formula for binomial coefficients, P-adic valuation of a product, Division algorithm, Floor function, Binomial coefficient
Changed4 November 2026

Markdown · JSON · Edit on GitHub

What this page needs for W1

  • an inventory row: the row of the inventory that the node accounts for (0 of 1)

The writing levels say how complete the text of a page is. They do not measure its truth.

What links here

No page links to this page yet.