Blog Entry (c) Friday, June 21, 2024, by James Pate Williams, Jr. Comparison of Two Prime Number Sieves

First the C++ results:

Limit = 1000000
Number of primes <= 1000000 78498
Milliseconds taken by Sieve of Atkin: 12
Number of primes <= 1000000 78498
Milliseconds taken by Sieve of Eratosthenes: 14
Limit = 10000000
Number of primes <= 10000000 664579
Milliseconds taken by Sieve of Atkin: 159
Number of primes <= 10000000 664579
Milliseconds taken by Sieve of Eratosthenes: 204
Limit = 100000000
Number of primes <= 100000000 5761455
Milliseconds taken by Sieve of Atkin: 1949
Number of primes <= 100000000 5761455
Milliseconds taken by Sieve of Eratosthenes: 2343
Limit = 0

Next, we have the Java results:

C:\WINDOWS\system32>java -jar k:\SieveOfAtkin\build\Debug\SieveOfAtkin.jar 1000000 0
number of primes less than equal 1000000 = 78498
total computation time in seconds = 0.008

C:\WINDOWS\system32>java -jar k:\SieveOfAtkin\build\Debug\SieveOfAtkin.jar 10000000 0
number of primes less than equal 10000000 = 664579
total computation time in seconds = 0.098

C:\WINDOWS\system32>java -jar k:\SieveOfEratosthenes\build\Debug\SieveOfEratosthenes.jar 1000000 0
number of primes less than equal 1000000 = 78498
total computation time in seconds = 0.011

C:\WINDOWS\system32>java -jar k:\SieveOfEratosthenes\build\Debug\SieveOfEratosthenes.jar 10000000 0
number of primes less than equal 10000000 = 664579
total computation time in seconds = 0.151

C:\WINDOWS\system32>java -jar k:\SieveOfAtkin\build\Debug\SieveOfAtkin.jar 100000000 0
number of primes less than equal 100000000 = 5761455
total computation time in seconds = 1.511

C:\WINDOWS\system32>java -jar k:\SieveOfEratosthenes\build\Debug\SieveOfEratosthenes.jar 100000000 0
number of primes less than equal 100000000 = 5761455
total computation time in seconds = 1.995

Notice that the Java application outperforms the C++ application.

// PrimeSieveComparison.cpp (c) Friday, June 21, 2024
// by James Pate Williams, Jr.
//
//  SieveOfAtkin.java
//  SieveOfAtkin
//
//  Created by James Pate Williams, Jr. on 9/29/07.
//  Copyright (c) 2007 James Pate Williams, Jr. All rights reserved.
//
//  SieveOfEratosthenes.java
//  SieveOfEratosthenes
//
//  Created by James Pate Williams, Jr. on 9/29/07.
//  Copyright (c) 2007 James Pate Williams, Jr. All rights reserved.
//

#include <math.h>
#include <iostream>
#include <chrono>
using namespace std::chrono;
using namespace std;

const int Maximum = 100000000;
bool sieve[Maximum + 1];

void SieveOfAtkin(int limit)
{
	auto start = high_resolution_clock::now();
	int e, k, n, p, x, xx3, xx4, y, yy;
	int primeCount = 2, sqrtLimit = (int)sqrt(limit);

	for (n = 5; n <= limit; n++)
		sieve[n] = false;

	for (x = 1; x <= sqrtLimit; x++) {
		xx3 = 3 * x * x;
		xx4 = 4 * x * x;
		for (y = 1; y <= sqrtLimit; y++) {
			yy = y * y;
			n = xx4 + yy;
			if (n <= limit && (n % 12 == 1 || n % 12 == 5))
				sieve[n] = !sieve[n];
			n = xx3 + yy;
			if (n <= limit && n % 12 == 7)
				sieve[n] = !sieve[n];
			n = xx3 - yy;
			if (x > y && n <= limit && n % 12 == 11)
				sieve[n] = !sieve[n];
		}
	}

	for (n = 5; n <= sqrtLimit; n++) {
		if (sieve[n]) {
			e = 1;
			p = n * n;
			while (true) {
				k = e * p;
				if (k > limit)
					break;
				sieve[k] = false;
				e++;
			}
		}
	}
	
	for (n = 5; n <= limit; n++)
		if (sieve[n])
			primeCount++;

	auto stop = high_resolution_clock::now();
	auto duration = duration_cast<milliseconds>(stop - start);

	std::cout << "Number of primes <= " << limit << ' ';
	std::cout << primeCount << endl;
	std::cout << "Milliseconds taken by Sieve of Atkin: "
		<< duration.count() << endl;
}

void SieveOfEratosthenes(int limit)
{
	auto start = high_resolution_clock::now();
	int i = 0, k = 0, n = 0, nn = 0;
	int primeCount = 0, sqrtLimit = (int)sqrt(limit);

	// initialize the prime number sieve

	for (n = 2; n <= limit; n++)
		sieve[n] = true;

	// eliminate the multiples of n

	for (n = 2; n <= sqrtLimit; n++)
		for (i = 2; i <= n - 1; i++)
			sieve[i * n] = false;

	// eliminate squares

	for (n = 2; n <= sqrtLimit; n++) {
		if (sieve[n]) {
			k = 0;
			nn = n * n;
			i = nn + k * n;
			while (i <= limit) {
				sieve[i] = false;
				i = nn + k * n;
				k++;
			}
		}
	}

	primeCount = 0;

	for (n = 2; n <= limit; n++)
		if (sieve[n])
			primeCount++;

	auto stop = high_resolution_clock::now();
	auto duration = duration_cast<milliseconds>(stop - start);

	std::cout << "Number of primes <= " << limit << ' ';
	std::cout << primeCount << endl;
	std::cout << "Milliseconds taken by Sieve of Eratosthenes: "
		<< duration.count() << endl;
}

int main()
{
	while (true)
	{
		int limit = 0;
		std::cout << "Limit = ";
		cin >> limit;

		if (limit == 0)
			break;

		SieveOfAtkin(limit);
		SieveOfEratosthenes(limit);
	}

	return 0;
}

Blog Entry Tuesday, June 18, 2024 (c) James Pate Williams, Jr. FreeLIP Computation of Euler Numbers and Tangent Numbers

FreeLIP is a free large integer package solely created by Professor Emeritus Arjen K. Lenstra of the Number Field Sieve fame. He developed FreeLIP while he was an employee of AT&T – Lucent in the late 1980s. His copyright notice in the header file, lip.h, states copyright from 1989 to 1997. I have been using this excellent number theoretical package since the late 1990s. See the paper by Donald E. Knuth and Thomas J. Buckholtz for the formula for Tangent Numbers. I can’t remember where I got the Euler Numbers recurrence relation. I wrote a C# application in 2015 for computing Euler Numbers. The code below is in the vanilla C computer language. Excellent resources for the Euler and tangent numbers also known as zag numbers are:

https://oeis.org/A122045

https://oeis.org/A000182

Blog Entry Sunday, June 16, 2024 (c) James Pate Williams, Jr. Chapter 4 Matrices and Systems of Linear Equations from a Textbook by S. D. Conte and Carl de Boor

Blog Entry Friday, June 14, 2024 (c) James Pate Williams, Jr.

For the last week or so I have been working my way through Chapter 3 The Solution of Nonlinear Equations found in the textbook “Numerical Analysis: An Algorithmic Approach” by S. D. Conte and Carl de Boor. I also used some C source code from “A Numerical Library in C for Scientists and Engineers” by H. T. Lau, PhD. I implemented twenty examples and exercises from the previously mentioned chapter.

Blog Entry June 5-7, 2024, (c) James Pate Williams, Jr. All Applicable Rights Reserved Chapter 7 Example and Some Exercises from “Numerical Analysis: An Algorithmic Approach (c) 1980 by S. D. Conte and Carl de Boor (Numerical Differentiation and Numerical Integration)

Blog Entry June 3-4, 2024, (c) James Pate Wiliams, Jr., Solution of Tridiagonal Matrix Problems

The first solution is from the textbook, Elementary Numerical Analysis: An Algorithmic Approach (c) 1980 by S. D. Conte and Carl de Boor. I translated the FORTRAN code to vanilla C using Visual Studio 2019 Community Version. The second solution is from Boundary Value Problems Second Edition (c) 1979 by David L. Powers. It solves a simple second order linear ordinary differential equation using the finite element difference equation method.

Blog Entry for Early Morning Friday, May 31, 2024, Solutions of a Second Order Self-Adjoint Ordinary Differential Equation by Three Methods