INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
The Probability of Coincidence in Citation and Alphabetical  
Ordering an Exhaustive Analysis of the Rencontres Problem  
Dr Partha Majumdar  
Department of Computer Science, Swiss School of Business and Management (SSBM), Geneva,  
Switzerland  
DOI : https://doi.org/10.51583/IJLTEMAS.2026.150100010  
Received: 31 December 2025; Accepted: 09 January 2026; Published: 23 January 2026  
ABSTRACT  
This analysis explores the likelihood that a reference's original citation number, based on its initial position in  
the text, matches its final number after sorting the bibliography alphabetically. It models this as counting the  
fixed points in a random permutation—a classic problem in combinatorial probability known as the Rencontres  
or Hat-Check problem—assuming a uniform random order. Using the Principle of Inclusion-Exclusion, an exact  
probability mass function for the number of references (n) with exactly k matches is derived, based on  
derangements. The study reveals a key asymptotic result: as n grows, the distribution of matches quickly and  
strongly approaches a Poisson distribution with a mean of λ=1. This means that for typical bibliography sizes (n  
≥ 10), the chance of any given number of matches becomes nearly independent of total references. Thus, the  
expected number of matches remains 1, regardless of the bibliography's length. The study also provides specific  
probabilities for at least k matches, computed via the Poisson approximation and involving the Incomplete  
Gamma Function. Notably, there is a 63.2% chance of at least one match, 26.4% for two or more, and 8.0% for  
three or more, indicating that multiple coincidences are rare. By distinguishing this from phenomena like the  
Birthday Paradox and highlighting its use in assessing shuffling algorithms and cryptographic security, the report  
presents a clear framework for understanding fixed-point matches in permutations. It concludes that despite the  
apparent randomness of reordering references, the number of matches follows a strict, predictable probabilistic  
law governed by the constant e.  
Keywords: Rencontres problem, Fixed points in permutations, Derangements, Poisson distribution, Citation  
ordering  
INTRODUCTION:  
The Intersection of Bibliographic Structure and Combinatorial Probability  
The organisation of academic literature relies heavily on structured referencing systems. When a researcher  
drafts a manuscript, references are typically accumulated in a chronological or "order of appearance" sequence.  
This initial listing represents the temporal flow of the author's engagement with existing literature. However, the  
final presentation of these references often adheres to style guides—such as those of the American Psychological  
Association (APA) or the Modern Language Association (MLA)—that mandate an alphabetical arrangement by  
the primary author's surname. This transformation from a chronological input to an alphabetical output  
constitutes a permutation of the original set of reference indices.  
The query addressed in this report poses a fundamental probabilistic question arising from this editorial process:  
given a manuscript with n references, what is the likelihood that the original citation number (based on order of  
appearance) and the final reference number (based on alphabetical sorting) are identical in at least k instances?  
This problem, while rooted in the mundane mechanics of academic formatting, maps directly onto one of the  
most celebrated and rigorously studied problems in combinatorial probability: the Rencontres problem (or the  
Problem of Encounters), also widely known as Montmort’s matching problem or the Hat-Check problem.  
www.ijltemas.in  
Page 126  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
This report provides a comprehensive, expert-level analysis of this phenomenon. We will explore the  
mathematical formalisation of the problem as a search for fixed points in random permutations. We will derive  
the exact probability distributions for finite bibliographies, investigate the asymptotic behaviour as the number  
of references grows large, and demonstrate the remarkable convergence to the Poisson distribution. Furthermore,  
we will delve into the deep connections between this problem and the Incomplete Gamma function, providing  
both theoretical proofs and computational algorithms for assessing these probabilities. By synthesising historical  
context, rigorous derivation, and modern computational statistical methods, we aim to offer a definitive answer  
regarding the probability of at least k ordinal matches.  
Formalising the Bibliographic Permutation  
To rigorously analyse the probability of reference numbers remaining unchanged, we must first establish a  
mathematical model of the "listing" and "sorting" processes.  
Let R = {r1, r2, …, rn} be the set of n distinct references cited in the paper.  
The listing process assigns a unique integer index i ∈ {1, …, n} to each reference based on its first appearance  
in the text. This defines a bijection L: {1, …, n} → R, where L(i) is the reference cited at position i.  
The sorting process rearranges the elements of R into a sequence defined by the lexicographical order of the  
author names (and titles/years for same-author works). This defines a second bijection S: {1, …, n} → R, where  
S(j) is the reference that appears at the j-th position in the sorted bibliography.  
A "match" occurs for a specific reference if its position in the citation list is identical to its position in the sorted  
bibliography. Mathematically, we are comparing the two orderings. We can define a permutation σ of the set  
{1, …, n} that maps the citation index to the sorted index. Alternatively, and more intuitively for this problem,  
we can consider the "alphabetical rank" of the reference cited at position i. Let π be the alphabetical rank of the  
reference that appeared i-th in the text.  
If the first reference cited (i=1) happens to be the one that comes first alphabetically, then π(1) = 1.  
If the first reference cited (i=1) is the one that comes last alphabetically, then π(1) = n.  
The condition that the "original reference number and the reference number after sorting will be the same" is  
mathematically equivalent to the condition π(i) = i. In the theory of permutations, such an index i is called a  
fixed point of the permutation π.  
The Assumption of Randomness  
A critical assumption in this analysis is the distribution of the permutation π. In the absence of specific  
information regarding the author's citation habits (e.g., a propensity to cite authors with surnames beginning with  
'A' earlier in the text), we model π as a uniform random permutation. This means that any of the n! possible  
orderings of the n alphabetical ranks are equally likely to occur as the citation order.  
While real-world citation behaviours might exhibit slight biases—perhaps foundational texts (often older and  
potentially distributed differently alphabetically) are cited first—the uniform random permutation model is the  
standard "null hypothesis" for such problems and provides the baseline probabilistic truth. Under this  
assumption, the problem transforms into finding the probability that a permutation chosen uniformly at random  
from the symmetric group Sn has at least k fixed points. [15]  
Historical and Theoretical Context: The Legacy of Montmort  
The mathematical lineage of the fixed-point problem dates back to the early 18th century, embedding the  
question in a rich history of gaming and probability. Understanding this history clarifies why certain  
mathematical tools (like the subfactorial) were developed and how they apply to the sorting of references.  
www.ijltemas.in  
Page 127  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
The Problem of Rencontres  
The specific framing of "matches" is classically referred to as the Rencontres problem. Pierre Rémond de  
Montmort first formulated it in his seminal work Essay d'analyse sur les jeux de hazard (1708). Montmort was  
investigating a card game called Treize (Thirteen), where a dealer turns over cards while counting "one, two,  
three..." up to thirteen. A "rencontre" (encounter or meeting) occurred if the card revealed matched the number  
spoken.  
The scenario is an isomorphism of this game:  
The Counter: The citation numbers 1, 2, …, n correspond to the dealer calling out numbers.  
The Card/Object: The reference's actual alphabetical rank corresponds to the value of the card revealed.  
The Win Condition: A reference number matching the sorted number corresponds to a rencontre.  
Montmort, along with correspondents Nicholas Bernoulli and later Leonhard Euler, sought to calculate the  
probability of having no encounters (a derangement) and, by extension, the probability of exactly k encounters.  
The Hat Check Problem Analogy  
In modern combinatorial literature, this problem is frequently illustrated as the Hat Check Problem. [2]  
n men check their hats at a cloakroom. The attendant, having lost the claim tickets, returns the hats entirely at  
random. What is the probability that exactly k men receive their own hats?  
This analogy perfectly mirrors the bibliography sorting problem:  
The "men" are the positions in the original citation list.  
The "hats" are the alphabetical ranks associated with the references.  
A man receiving his own hat is equivalent to a reference where the citation index i equals the sorted index  
π(i).  
The widespread use of this analogy underscores the counterintuitive nature of the result: as we will see, the  
probability of a specific number of matches stabilises very quickly. Whether there are 10 men (references) or  
10,000, the probability that exactly one man gets his hat back remains almost constant. This property, known as  
asymptotic stability, is a central theme of our analysis and provides a robust answer to the query regardless of  
the exact length of their bibliography (assuming n is reasonably large). [15]  
Mathematical Significance  
The study of fixed points in permutations serves as a gateway to broader topics in group theory and statistics.  
Cycle Structure: A fixed point is simply a cycle of length 1. John Baez and other mathematicians have  
highlighted that the expected number of cycles of length k in a random permutation is 1/k. For k=1, this yields  
the elegant result that the expected number of fixed points is 1. [1]  
Derangements: The case where k=0 (no matches) defines the derangement numbers or subfactorials,  
denoted !n or Dn. The calculation of Dn is fundamental to solving the general case for k matches. [12]  
Combinatorial Derivation of Exact Probabilities  
To answer the question regarding the probability of at least k matches, we must first construct the probability  
mass function for exactly k matches. [4] This derivation relies on the Principle of Inclusion-Exclusion.  
www.ijltemas.in  
Page 128  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
The Number of Derangements (Dn)  
We begin by calculating the number of permutations of n elements that have zero fixed points. Let Sn denote the  
set of all permutations of {1, …, n}, so |Sn| = n!. Let Pi be the property that a permutation π fixes the element i  
(i.e., π(i) = i). We seek the number of permutations that satisfy none of the properties P1, P2, …, Pn.  
According to the Principle of Inclusion-Exclusion, the number of permutations with none of these properties is:  
[13]  
Dn = 푇 ⊆ {1,…,푛}(−1)|푇| (푁(ꢀ)  
∑
where N(T) is the number of permutations that fix at least the elements in the set T.  
If we fix a set of elements T with |T| = j, the remaining n-j elements can be permuted in any order. Thus:  
N(T) = (n-j)!  
There are (푛푗 ) distinct subsets T of size j. Therefore, the term corresponding to size j in the inclusion-exclusion  
sum is: (푛푗) (n-j)!  
Expanding the binomial coefficient:  
푛!  
(푛푗) (n-j)! =  
(n-j)! = 푛!  
(
)
푗! 푛−푗 !  
푗!  
Substituting this back into the sum:  
ꢁ
푛!  
푗!  
(−1)  
푛
푛
Dn = 푇 ⊆ {1,…,푛}(−1)|푇| (푁(ꢀ) = 푗=0(−1)푗  
= n!  
)
∑
∑
∑
푗=0  
푗!  
ꢂ
Dn = n! (1 -  
+
-
+ ... + (−1)  
1
1
1
0! 1!  
2! 3!  
푛!  
This formula for Dn (often written as !n) is the exact count of bibliographies where no reference retains its  
original number after sorting. [14]  
Generalising to Exactly k Matches  
Let X be the random variable representing the number of fixed points (matches) in a random permutation of n  
elements. We wish to find the probability P(X=k).  
To form a permutation with exactly k fixed points, we perform a two-step constructive process: [9]  
Selection of Matches: We must choose exactly k indices out of the n available positions to be the fixed points.  
(푛)  
The binomial coefficient gives the number of ways to choose these k indices  
.
푘
Derangement of the Remainder: The remaining n-k elements must be permuted such that none of them map  
to their own locations (otherwise, we would have more than k fixed points). The number of ways to arrange  
these n-k elements as a derangement is Dn-k.  
Combining these steps, the total number of permutations with exactly k fixed points, denoted Nn(k), is:  
www.ijltemas.in  
Page 129  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
(푛)  
Nn(k) =  
Dn-k  
푘
The probability P(X=k) is the ratio of favourable permutations to the total number of permutations (n!):  
ꢂ
ꢃ
P(X = k) = ( ) 퐷  
ꢂ−ꢃ  
푛!  
We can simplify this expression by expanding the binomial coefficient and the derangement term:  
푛!  
퐷
퐷
ꢂ−ꢃ  
ꢂ−ꢃ  
P(X = k) =  
=
(
)
(
)
푘! 푛−푘 ! 푛!  
푘! 푛−푘 !  
Now, substitute the inclusion-exclusion formula for Dn-k:  
ꢁ
ꢁ
1
(−1)  
1
(−1)  
푛−푘  
푛−푘  
푗=0  
∑
∑
푗=0  
P(X = k) =  
(n – k)!  
=
(
)
푘! 푛−푘 !  
푗!  
푘!  
푗!  
This equation provides the exact probability that the reference number matches the sorted number in exactly k  
cases for a bibliography of size n. [7]  
The "At Least k" Formulation  
We need to find the probability of the numbers being the same in at least k cases. This corresponds to the  
cumulative probability of the upper tail of the distribution:  
푛
∑
P(X ≥ k) = ꢄ=푘 푃(푋 = 푚)  
Substituting the derived formula for P(X=m):  
ꢁ
1
(−1)  
푛−ꢄ  
푛
ꢄ=푘  
∑
∑
푗=0  
P(X ≥ k) =  
(
)
ꢄ!  
푗!  
While this double summation provides the precise answer for any finite n, calculating it manually is tedious.  
However, the structure of the inner sum (the partial series for 1/e) hints strongly at a convergent limit, which we  
will explore in the next section. This limit allows for a much simpler "closed form" approximation that is highly  
accurate for typical bibliography sizes (n > 10).  
Asymptotic Analysis: Convergence to the Poisson Distribution  
One of the most powerful results in combinatorial probability is the convergence of the fixed-point distribution  
to the Poisson distribution. This fact implies that the length of the bibliography (n) effectively "drops out" of the  
equation once it becomes sufficiently large, simplifying the problem to a universal probability calculation.  
The Limit of the Derangement Sum  
Consider the inner summation derived earlier:  
ꢁ
(−1)  
푛−푘  
푗=0  
∑
Sn-k  
=
푗!  
We recognise this immediately as the partial sum of the Maclaurin series expansion for the exponential function  
ex, evaluated at x = -1:  
ꢁ
ꢁ
푥
(−1)  
x
∞
푗=0  
-1  
∞
∑
∑
푗=0  
e =  
⇒ e =  
푗!  
푗!  
As n → ∞ (assuming k is fixed), the upper limit of the sum n-k also approaches infinity. Therefore:  
www.ijltemas.in  
Page 130  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
ꢁ
(−1)  
푛−푘  
푗=0  
∑
lim푛 → ∞  
= e-1 = 1  
푗!  
푒
The Poisson Probability Mass Function  
Applying this limit to the probability formula for exactly k matches:  
ꢁ
1
(−1)  
1 1  
푛−푘  
푗=0  
∑
lim 푃(푋 = ꢅ) =  
lim  
=
푘!  
푗!  
푘! 푒  
푛 → ∞  
푛 → ∞  
−1  
ꢃ
1
P(X = k) ≈ 푒  
푘!  
−휆  
ꢃ
푒
ꢆ
This expression,  
with λ = 1, is the probability mass function (PMF) of a Poisson distribution with mean  
푘!  
parameter λ = 1.  
Thus, we formally state:  
The number of fixed points in a uniform random permutation of n elements converges in distribution to a  
Poisson(1) random variable as n → ∞.  
This result is robust and well-documented in the research literature. [11] It implies that for a reasonably long  
bibliography (e.g., n=50), the probability of having exactly k matching reference numbers is independent of n.  
Error Analysis and Rate of Convergence  
You might wonder: "How large does n need to be for this approximation to be valid?"  
ꢁ. The absolute  
(−1)  
∑
The speed of convergence of the alternating series determines the error in the approximation  
value of the first neglected term bounds the error of an alternating series truncation.  
푗!  
ꢁ
ꢁ
(−1)  
(−1)  
푛−푘  
푗=0  
∞
∑
∑
푗=0  
The approximation replaces  
with  
.
푗!  
푗!  
The difference is the tail of the series starting from j = n-k+1:  
|Error| = |(−1)(ꢂ−ꢃ+1| =  
1
(
)
(
)
푛−푘+1 !  
푛−푘+1 !  
For a bibliography of n=20 references and checking for k=2 matches:  
1
1
|Error| =  
=
(
)
20−2+1 !  
19!  
This error is infinitesimally small (< 10-17).  
Even for very small n (like n=5), the approximation is remarkably good.  
Exact P(X=0) for n=5: 0.36666...  
Poisson Approximation (1/e): 0.36787...  
Difference: ≈0.0012 (0.12%)  
Unless the paper has fewer than five references, the Poisson model is functionally exact. We can ignore this, as  
most journals require a minimum of 15 references.  
www.ijltemas.in  
Page 131  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
Implications of λ = 1  
The fact that the parameter λ = 1 is profoundly significant.  
Expected Value: The mean of a Poisson(1) distribution is λ = 1. This confirms the result derived via linearity  
of expectation: on average, exactly one reference number will remain unchanged after sorting. [5]  
Variance: The variance of Poisson(1) is also 1. The spread of the distribution is moderate; extremely high  
numbers of matches are exponentially rare. [3]  
Calculating the "At Least k" Probability  
Having established the Poisson(1) model, we can now provide the specific computational solution for our  
requirement: determining the probability of at least k matches.  
The Complementary Approach  
Directly summing the Poisson probabilities for k, k+1, k+2, … to infinity (or n) is valid, but it is often  
computationally easier to use the complement. The probability of at least k matches is 1 minus the probability  
of fewer than k matches.  
푘−1  
푗=0  
∑
P(X ≥ k) = 1 - P(X < k) = 1 -  
푃(푋 = ꢇ)  
Using the Poisson approximation:  
−1  
푒
1
= 1 - 1  
푘−1  
푗=0  
푘−1  
∑
∑
푗=0  
P(X ≥ k) ≈ 1 -  
푗!  
푒
푗!  
This formula is the most efficient way to calculate the probability manually or with a simple calculator. One  
sums the inverse factorials up to 1/(k-1)!, divides by e, and subtracts from 1.  
Connection to the Incomplete Gamma Function  
For more advanced analysis or software implementation, this probability is often expressed using the Incomplete  
Gamma Function. This connection is explicitly highlighted in the research snippets. [11]  
The cumulative distribution function (CDF) of a Poisson random variable with mean λ is related to the  
Regularised Upper Incomplete Gamma Function, denoted Q(s, x) or Γreg(s, x). [8]  
The identity states:  
P(Y < k) = 훤(푘,ꢆ) = Q(k, λ)  
훤(푘)  
where Y ~ Poisson(λ).  
Note the slight nuance in definitions. The probability of at most k-1 events (which corresponds to X < k) is often  
given as:  
−휆  
ꢁ
푒
ꢆ
훤(푘,ꢆ)  
푘−1  
푗=0  
∑
=
= Q(k, λ)  
(
)
푗!  
푘−1 !  
Therefore, the probability we seek, P(X ≥ k), is the complement of this CDF:  
P(X ≥ k) = 1 - P(X < k) = 1 - Q(k, λ).  
www.ijltemas.in  
Page 132  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
Since P(s, x) + Q(s, x) = 1 (where P is the Regularised Lower Gamma function), we have:  
P(X ≥ k) = P(k, λ) = 훾(푘,ꢆ)  
훤(푘)  
For our specific case where λ = 1:  
훾(푘,1)  
P(X ≥ k) ≈ (  
)
푘−1 !  
where γ(k, 1) = 1 푡푘−1ꢈ−ꢉ 푑푡.  
∫
0
This "closed form" is extremely useful because many statistical software packages (like Python's scipy.stats or  
MATLAB) implement optimised routines for Incomplete Gamma functions, allowing for high-precision  
calculation of these tail probabilities. [11]  
Tabulated Probabilities  
To provide immediate value, we calculate the probabilities for typical values of k using the Poisson(1) model.  
ퟏ
Formula (1− ퟏ  
)
풌−ퟏ  
풋=ퟎ  
Minimum  
Matches  
(k)  
Probability  
Percentage  
Interpretation  
∑
풆
풋!  
1 - 1 (1)  
1 - 0.3679 =  
0.6321  
More likely than not to  
have ≥ 1 match.  
At least 1  
At least 2  
At least 3  
At least 4  
63.21%  
26.42%  
8.03%  
1.90%  
푒
1 - 1 (1 + 1)  
1 - 0.7358 =  
0.2642  
About a 1 in 4 chance.  
푒
1 - 1 (1 + 1 + 0.5)  
1 - 0.9197 =  
0.0803  
Fairly rare (less than 1 in  
10).  
푒
1 - 1 (1 + 1 + 0.5 +  
1 - 0.9810 =  
0.0190  
Very rare (~2%).  
푒
0.1667)  
1 - 1 (1 + 1 + 0.5 +  
0.00365  
0.00058  
Extremely unlikely.  
Less than 1 in 1000.  
At least 5  
At least 6  
0.37%  
0.06%  
푒
0.1667 + 0.04166667)  
1 - 1 (1 + 1 + 0.5 +  
푒
0.1667 + 0.04166667 +  
0.00833333)  
Table 1: Probability of Reference Coincidences.  
Insight: The table reveals a rapid decay. While a single match is expected and common (63%), finding three or  
more references that keep their original numbers is a statistically significant anomaly (only 8%).  
Computational Algorithms and Implementation  
For researchers wishing to verify these results or compute exact probabilities for small N where the Poisson  
approximation might have a theoretically distinct (though numerically negligible) error, specific algorithms are  
preferred over the summation formula.  
www.ijltemas.in  
Page 133  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
Recursive Calculation of Derangements  
Computing factorials for large N leads to overflow errors in standard integer arithmetic. A more stable approach  
for finding the probability relies on the recurrence relation for derangements.  
Recall Dn = (n-1)(Dn-1 + Dn-2).  
Dividing by n!, we can derive a recurrence for the probability pn = Dn/n! (the probability of 0 matches in a list  
of size n):  
)
pn = (  
=
p
+ 1 pn-2  
푛−1 (퐷  
+ 퐷  
)
푛−1  
ꢂ−1  
ꢂ−2  
n-1  
푛!  
푛
푛
Wait, the simpler recurrence is Dn = n Dn-1 + (-1)n.  
Dividing this by n!:  
ꢂ
퐷
푛 퐷  
ꢂ−1  
(−1)  
ꢂ
=
+
푛!  
푛!  
푛!  
ꢂ
(−1)  
Or, pn = pn-1  
+
푛!  
ꢁ
(−1)  
푛
푗=0  
∑
This confirms the partial sum series directly: pn =  
.
푗!  
Dynamic Programming for Exact "At Least k"  
To find the exact probability of at least k matches for a fixed n without approximation:  
Construct D array: Generate D0, D1, …, Dn using Di = i Di-1 + (-1)i.  
(푛 )  
푛
ꢄ=푘  
∑
Sum Favourable Outcomes: Count =  
Dn-m  
ꢄ
Calculate Probability: P = 퐶표푢푛ꢉ  
푛!  
This approach avoids floating-point precision issues inherent in summing alternating series by keeping all  
calculations in integers until the final division. It is the preferred method for computer algebra systems. [6]  
Code Logic  
A Python code for the exact calculation:  
import math  
# Precompute Derangements using D_n = n * D_{n-1} + (-1)^n  
from functools import lru_cache  
@lru_cache(maxsize=None)  
def derangements(n):  
if n < 0:  
raise ValueError("Derangements are not defined for negative numbers.")  
if n == 0:  
return 1  
result = n * derangements(n - 1) + (-1) ** n  
return result  
def probability_at_least_k_matches(n, k) -> float:  
www.ijltemas.in  
Page 134  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
if n < 0:  
raise ValueError("'Number of references' (n) should not be negative.")  
if k < 0:  
raise ValueError("'Number of matches' (k) should not be negative.")  
if k > n:  
return 0.0  
# Sum favourable permutations  
favourable_permutations = 0  
fact_n = math.factorial(n)  
for m in range(k, n + 1):  
# n choose m  
n_choose_m = fact_n // (math.factorial(m) * math.factorial(n - m))  
# Add to total: (ways to choose fixed) * (ways to derange rest)  
favourable_permutations += n_choose_m * derangements(n - m)  
return favourable_permutations / fact_n  
probability_at_least_k_matches(25, 2)  
0.26424111765711533  
probability_at_least_k_matches(50, 2)  
0.26424111765711533  
probability_at_least_k_matches(30, 4)  
0.01898815687615381  
probability_at_least_k_matches(35, 1)  
0.6321205588285577  
This algorithm provides the exact probability P(X ≥ k), valid for any n and k.  
Broader Context and Applications  
While the query arises from a specific academic task, the underlying probabilistic mechanics have far-reaching  
implications in fields ranging from cryptography to statistical testing.  
The "Birthday Paradox" Distinction  
A common misconception is equating "matches" or "collisions" in this problem with the famous Birthday  
Paradox. [10] It is crucial to distinguish them:  
Birthday Problem (Sampling With Replacement): We ask if any two items share a value. The mappings are  
independent. The sample space is nn. Probabilities scale with √ꢊ.  
Rencontres Problem (Sampling Without Replacement): We ask if an item maps to its own specific location.  
The mappings are dependent (a permutation). The sample space is n!. The probability is constant (1/e).  
In the birthday problem, the probability of a match approaches 1 rapidly as n grows (for fixed days). In the  
www.ijltemas.in  
Page 135  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
bibliography problem, the probability of a match stabilises at ≈63.2% and never reaches 1. This highlights a  
fundamental difference between "collisions between arbitrary elements" and "collisions with a fixed index."  
Applications in Random Number Testing  
The distribution of fixed points is a standard metric in evaluating the quality of Random Number Generators  
(RNGs) and shuffling algorithms (like the Fisher-Yates shuffle).  
If a software library claims to shuffle a list "randomly," one can test this by shuffling a list of n items millions  
of times and counting the fixed points.  
● The mean number of fixed points should be close to 1.  
● The variance should be close to 1.  
● The distribution should match Poisson(1). Significant deviation from these expected values (e.g., if the  
average number of matches is 1.5 or 0.5) indicates a flaw in the shuffling algorithm, known as bias. The  
sorting process can effectively be seen as a "physical" test of randomness. If they consistently find 5 or  
6 matches in every paper, it suggests their citation process is not random but correlated with alphabetical  
order (e.g., they systematically cite 'A' authors earlier).  
Cryptography and Hashing  
In cryptography, the "birthday attack" relies on finding collisions. However, the Rencontres model applies to  
fixed-point attacks on permutation-based cyphers. If a block cypher (which is a permutation of the message  
space) has significantly more fixed points than a random permutation, it is considered a weakness. The  
probability analysis derived here (1/e for no fixed points) is the baseline against which cryptographic security is  
measured.  
CONCLUSION  
The problem of determining the likelihood that a reference's original citation number matches its sorted number  
is a classical application of the Rencontres problem. By modelling the citation and sorting process as a uniform  
random permutation, we have derived the following exhaustive conclusions:  
Exact Probability: For a paper with n references, the probability of exactly k matches is given by $\frac{1}{k!}  
\sum_{j=0}^{n-k} \frac{(-1)^j}{j!}$.  
Universal Approximation: For any typical bibliography size (n ≥ 10), the distribution of matches is effectively  
independent of n. It follows a Poisson distribution with parameter λ = 1.  
The Answer to "At Least k": The probability that the reference numbers match in at least k cases is  
approximately:  
1
P(Matches ≥ k) ≈ 1 - 1  
푘−1  
푖=0  
∑
푒
푖!  
훾(푘 ,1)  
Alternatively, this can be calculated using the Regularised Lower Incomplete Gamma Function: P ≈  
.
(
)
푘−1 !  
Key Probabilities:  
○ There is a 63.2% chance of at least one match.  
○ There is a 26.4% chance of at least two matches.  
○ There is an 8.0% chance of at least three matches.  
www.ijltemas.in  
Page 136  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
Expectation: Regardless of the length of the bibliography, the expected number of references that retain their  
original number is exactly 1.  
These findings illustrate a fascinating interplay between order and chaos: while the shuffling of references  
appears random, the number of coincidences adheres to a strict and predictable probabilistic law governed by  
the constant e. Whether analysing bibliographic data, shuffling cards, or securing data, the principles of the  
Rencontres problem provide the definitive mathematical framework for understanding fixed points in  
permutations.  
REFERENCES  
1. Arratia, R., & Desalvo, S. (2016, June). Poisson and independent process approximation for random  
combinatorial structures with a given number of components, and near-universal behavior for low rank  
assemblies. ResearchGate. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/303993076_Poisson_and_independent_process_approximatio  
n_for_random_combinatorial_structures_with_a_given_number_of_components_and_near-  
universal_behavior_for_low_rank_assemblies  
2. Arratia, R., Barbour, A. D., & Tavaré, S. (2003, December). Logarithmic Combinatorial Structures: A  
Probabilistic Approach. ResearchGate. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/240097339_Logarithmic_Combinatorial_Structures_a_Proba  
bilistic_Approach  
3. Brizzi, M., Nani, D., & Betti, L. (2021, December). Poisson distribution and process as a well-fitting  
pattern for counting variables in biologic models. International Journal of High Dilution Research.  
Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/357252061_Poisson_distribution_and_process_as_a_well-  
fitting_pattern_for_counting_variables_in_biologic_models  
4. Diaconis, P., Fulman, J., & Guralnick, R. (2007, September). On fixed points of permutations. Journal of  
Algebraic Combinatorics. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/1760933_On_fixed_points_of_permutations  
5. Feller, W. (1968). An Introduction to Probability Theory and Its Applications (3rd ed.). Wiley.  
6. Hashempour, M., Mohammadi, M., & Doostparast, M. (2025, December). Statistical Inference for Load-  
Sharing Systems with Heterogeneous Populations Using Sequential Order Statistics. ResearchGate.  
Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/398840264_Statistical_Inference_for_Load-  
Sharing_Systems_with_Heterogeneous_Populations_Using_Sequential_Order_Statistics  
7. HASHIMOTO, Y., NUIDA, K., Shinagawa, K., INAMURA, M., & Hanaoka, G. (2018, September).  
Toward Finite-Runtime Card-Based Protocol for Generating a Hidden Random Permutation without  
Fixed Points. IEICE Transactions on Fundamentals of Electronics. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/327379955_Toward_Finite-Runtime_Card-  
Based_Protocol_for_Generating_a_Hidden_Random_Permutation_without_Fixed_Points  
8. Kuba, M. (2025, February). A note on mixed Poisson distributions. ResearchGate. Retrieved December  
31, 2025, from  
https://www.researchgate.net/publication/388954982_A_note_on_mixed_Poisson_distributions  
9. Lawrence, W. (2025, December). Theory of Derived Probability and Entanglement Compression.  
ResearchGate. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/397000891_Theory_of_Derived_Probability_and_Entanglem  
ent_Compression  
10. Lindo, A., & Sagitov, S. (2015, July). Asymptotic results for the number of Wagner's solutions to a  
generalised birthday problem. Statistics & Probability Letters. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/280243331_Asymptotic_results_for_the_number_of_Wagner  
's_solutions_to_a_generalised_birthday_problem  
11. Mantalos, P., & Zografos, K. (2008, December). Interval estimation for a binomial proportion: A  
bootstrap approach. Journal of Statistical Computation and Simulation. Retrieved December 31, 2025,  
from  
www.ijltemas.in  
Page 137  
\INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue I, January 2026  
https://www.researchgate.net/publication/243043021_Interval_estimation_for_a_binomial_proportion_  
A_bootstrap_approach  
12. Pan, Q., Wang, Y., & Zeng, J. (2025, December 1). Counting permutations by alternating runs via  
Hetyei-Reiner trees. ResearchGate. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/398721020_Counting_permutations_by_alternating_runs_via  
Hetyei-Reiner_trees  
13. Pinsky, R. G. (2025, May). The inversion statistic in derangements and in other permutations with a  
prescribed number of fixed points. ResearchGate. Retrieved December 31, 2025, from  
https://www.researchgate.net/publication/391461367_The_inversion_statistic_in_derangements_and_i  
n_other_permutations_with_a_prescribed_number_of_fixed_points  
14. Riordan, J. (1958). An Introduction to Combinatorial Analysis. Wiley.  
15. Ross, S. M. (2014). Introduction to Probability Models (11th ed.). Academic Press.  
www.ijltemas.in  
Page 138