Mathematical Foundations

The Distribution of Prime Numbers

Core Definitions

Die Primzahlen (Prime Numbers)

Prime numbers are natural numbers strictly greater than 1 that possess exactly two distinct positive divisors: 1 and the number itself.

• The Case of 1: The number 1 is formally excluded from the primes because it does not satisfy the structural requirement of having two distinct positive divisors.
• Natural Numbers: These are positive integers ($1,2,3,\dots$) utilized primarily for counting discrete elements.

---

The Prime Number Theorem (PNT)

Originally conjectured by Carl Friedrich Gauss, the Prime Number Theorem describes the asymptotic distribution of prime numbers across large limits. It establishes that the count of primes up to a given bound $x$ can be closely approximated by the continuous logarithmic integral function, $Li(x)$:

$\pi (x)\approx Li(x)={\int }_2^x\frac{dt}{\ln t}$

Deconstructing the Equation

• Bound Variable ($x$): Represents the independent upper limit of the calculation. If estimating primes below a million, then $x=1,000,000$.
• Prime Counting Function ($\pi (x)$): A discrete function counting the exact quantity of prime numbers less than or equal to $x$. Because primes occur as discrete steps, the graph of $\pi (x)$ forms a jagged staircase that runs flat across composite ranges and steps up by exactly 1 upon hitting a prime value.
• Logarithmic Integral ($Li$): Deriving from the German term Logarithmische Integral, this denotes the integral functional operator itself rather than a standalone algebraic variable.
• Density Fraction ($\frac{1}{\ln t}$): Represents the local density probability of a given integer $t$ being prime. Near $t=100$, roughly 1 in every $\ln (100)\approx 4.6$ numbers is prime ($\approx 0.217$). Near $t=1,000,000$, the density drops to 1 in every $\ln (1,000,000)\approx 13.8$ numbers ($\approx 0.072$), showing that primes thin out predictably at larger scales.

Geometric Interpretation

Geometrically, the integral behaves like a continuous accumulation machine. Instead of summing integers sequentially, it determines the total area bounded beneath the probability curve $\frac{1}{\ln t}$ from the baseline prime boundary $t=2$ up to the designated limit $t=x$.

---

Empirical Comparison: $\pi (x)$ vs. $Li(x)$

The table below illustrates the performance of Gauss’s smooth continuous approximation relative to the exact prime count up to $10^{10}$:

Bound ($x$)	$\pi (x)$ (Exact Primes)	$Li(x)$ (Gauss’s Approximation)	Absolute Error ($Li(x)-\pi (x)$)	Percentage Error	Ratio ($\frac{Li(x)}{\pi (x)}$)
$10^1$	4	5.12	+1.12	28.00000%	1.2800000
$10^2$	25	29.08	+4.08	16.32000%	1.1632000
$10^3$	168	176.56	+8.56	5.09524%	1.0509524
$10^4$	1,229	1,245.09	+16.09	1.30919%	1.0130919
$10^5$	9,592	9,628.80	+36.80	0.38365%	1.0038365
$10^6$	78,498	78,626.50	+128.50	0.16370%	1.0016370
$10^7$	664,579	664,917.41	+338.41	0.05092%	1.0005092
$10^8$	5,761,455	5,762,207.33	+752.33	0.01306%	1.0001306
$10^9$	50,847,534	50,849,233.92	+1,699.92	0.00334%	1.0000334
$10^{10}$	455,052,511	455,055,613.54	+3,102.54	0.00068%	1.0000068

Error Analysis & Sign Dynamics

• Asymptotic Convergence: While the absolute error margin ($Li(x)-\pi (x)$) expands in magnitude, the percentage error decays sharply, shrinking well below $0.001\%$ at the $10^{10}$ threshold. Concurrently, the ratio approaches $1$, validating the asymptotic behavior predicted by the PNT.
• The Littlewood Multiplicity: Although initial computations show a persistent overestimation by $Li(x)$, mathematician J.E. Littlewood rigorously proved in 1914 that the error term $Li(x)-\pi (x)$ oscillates and changes sign infinitely many times. This demonstrates that the exact count and the smooth curve bypass each other infinitely often, though the first crossover occurs at an immense scale known as Skewes’ number.