In the following (naive) implementation of a pow(x, n) method, ignoring completely any optimized approach, I find the following problem:
public double pow(double x, int n) {
boolean negative = n < 0;
long power = Math.abs(n);
double ans = 1.0;
for(long i = 0; i < power; i++) {
ans = ans * x;
}
return negative ? 1.0/ans: ans;
}
Here I have made the assumption that for the case of negative exponent I simply calculate the x^n and then return 1/(x^n) since e.g. 2^(-3) = 1/(2^3)
Problem:
The code fails in the following case:
pow(2.00000, -2147483648)
The output is 1.00000 while the expected correct result is 0.00000
If I change the code as follows:
public double pow(double x, int n) {
long power = n;
if(power < 0) {
x = 1 / x;
power = -power;
}
double ans = 1.0;
for(long i = 0; i < power; i++) {
ans = ans * x;
}
return ans;
}
The result is correct!
So what is the difference between doing the approaches? I was expecting them to be equivalent but they are not
Math.abs(n) is still an int, and only afterwards it is assigned to a long, Therefore, the absolute value of -2147483648 was -2147483648 again (this is noted in the documentation of Math.abs(int)). With the negative bound, the loop performed no iterations.
Math.abs((long)n) would work around that issue.
I've created some code that generates the Bernoulli Numbers based off of formula 33 on MathWorld. This is given at https://mathworld.wolfram.com/BernoulliNumber.html and should work for all integers n but it diverges from the expected results extremely quickly once it gets to n=14. I think the issue may be in the factorial code, although I have no idea.
It's pretty accurate up until 13, all odd numbers should be 0 besides 1 but the values past 14 give weird values. For instance 14 gives a number like 0.9 when it should give something around 7/6 and say 22 gives a very negative number in the order of 10^-4. The odd numbers give strange values like 15 gives around -11.
Here is all the related code
public static double bernoulliNumber2(int n) {
double bernoulliN = 0;
for (double k = 0D; k <= n; k++) {
bernoulliN += sum2(k,n)/(k+1);
}
return bernoulliN;
}
public static double sum2(double k, int n) {
double result = 0;
for (double v = 0D; v <= k; v++) {
result += Math.pow(-1, v) * MathUtils.nCr((int) k,(int) v) * Math.pow(v, n);
}
return result;
}
public static double nCr(int n, int r) {
return Factorial.factorial(n) / (Factorial.factorial(n - r) * Factorial.factorial(r));
}
public static double factorial(int n) {
if (n == 0) return 1;
else return (n * factorial(n-1));
}
Thank you in advance.
The problem here is that floating point arithmetic doesn't need to overflow to experience catastrophic loss of precision.
A floating point number has a mantissa and an exponent, where the value of the number is mantissa * 10^exponent (real floating point numbers use binary, I'm using decimal). The mantissa has limited precision.
When we add floating point numbers of different signs we can end up with a final result which has lost precision.
e.g. let's say the mantissa is 4 digits.
If we add:
1.001 x 10^3 + 1.000 x 10^4 - 1.000 x 10^4
we expect to get 1.001 x 10^3.
But 1.001 x 10^3 + 1.000 x 10^4 = 11.001 x 10^3, which is represented as 1.100 x 10^4, given that our mantissa has only 4 digits.
So when we subtract 1.000 x 10^4 we get 0.100 x 10^4, which is represented as 1.000 x 10^3 rather than 1.001 x 10^3.
Here's an implementation using BigDecimal which gives better results (and is far slower).
import java.math.BigDecimal;
import java.math.RoundingMode;
public class App {
public static double bernoulliNumber2(int n) {
BigDecimal bernoulliN = new BigDecimal(0);
for (long k = 0; k <= n; k++) {
bernoulliN = bernoulliN.add(sum2(k,n));
//System.out.println("B:" + bernoulliN);
}
return bernoulliN.doubleValue();
}
public static BigDecimal sum2(long k, int n) {
BigDecimal result = BigDecimal.ZERO;
for (long v = 0; v <= k; v++) {
BigDecimal vTon = BigDecimal.valueOf(v).pow(n);
result = result.add(BigDecimal.valueOf(Math.pow(-1, v)).multiply(nCr(k,v)).multiply(vTon).divide(BigDecimal.valueOf(k + 1), 1000, RoundingMode.HALF_EVEN));
}
return result;
}
public static BigDecimal nCr(long n, long r) {
return factorial(n).divide(factorial(n - r)).divide(factorial(r));
}
public static BigDecimal factorial(long n) {
if (n == 0) return BigDecimal.ONE;
else return factorial(n-1).multiply(BigDecimal.valueOf(n));
}
public static void main(String[] args) {
for (int i = 0; i < 20; i++) {
System.out.println(i + ": " + bernoulliNumber2(i));
}
}
}
Try changing the scale passed to the division in sum2 and watch the effect on the output.
Note: Updated on 06/17/2015. Of course this is possible. See the solution below.
Even if anyone copies and pastes this code, you still have a lot of cleanup to do. Also note that you will have problems inside the critical strip from Re(s) = 0 to Re(s) = 1 :). But this is a good start.
import java.util.Scanner;
public class NewTest{
public static void main(String[] args) {
RiemannZetaMain func = new RiemannZetaMain();
double s = 0;
double start, stop, totalTime;
Scanner scan = new Scanner(System.in);
System.out.print("Enter the value of s inside the Riemann Zeta Function: ");
try {
s = scan.nextDouble();
}
catch (Exception e) {
System.out.println("You must enter a positive integer greater than 1.");
}
start = System.currentTimeMillis();
if (s <= 0)
System.out.println("Value for the Zeta Function = " + riemannFuncForm(s));
else if (s == 1)
System.out.println("The zeta funxtion is undefined for Re(s) = 1.");
else if(s >= 2)
System.out.println("Value for the Zeta Function = " + getStandardSum(s));
else
System.out.println("Value for the Zeta Function = " + getNewSum(s));
stop = System.currentTimeMillis();
totalTime = (double) (stop-start) / 1000.0;
System.out.println("Total time taken is " + totalTime + " seconds.");
}
// Standard form the the Zeta function.
public static double standardZeta(double s) {
int n = 1;
double currentSum = 0;
double relativeError = 1;
double error = 0.000001;
double remainder;
while (relativeError > error) {
currentSum = Math.pow(n, -s) + currentSum;
remainder = 1 / ((s-1)* Math.pow(n, (s-1)));
relativeError = remainder / currentSum;
n++;
}
System.out.println("The number of terms summed was " + n + ".");
return currentSum;
}
public static double getStandardSum(double s){
return standardZeta(s);
}
//New Form
// zeta(s) = 2^(-1+2 s)/((-2+2^s) Gamma(1+s)) integral_0^infinity t^s sech^2(t) dt for Re(s)>-1
public static double Integrate(double start, double end) {
double currentIntegralValue = 0;
double dx = 0.0001d; // The size of delta x in the approximation
double x = start; // A = starting point of integration, B = ending point of integration.
// Ending conditions for the while loop
// Condition #1: The value of b - x(i) is less than delta(x).
// This would throw an out of bounds exception.
// Condition #2: The value of b - x(i) is greater than 0 (Since you start at A and split the integral
// up into "infinitesimally small" chunks up until you reach delta(x)*n.
while (Math.abs(end - x) >= dx && (end - x) > 0) {
currentIntegralValue += function(x) * dx; // Use the (Riemann) rectangle sums at xi to compute width * height
x += dx; // Add these sums together
}
return currentIntegralValue;
}
private static double function(double s) {
double sech = 1 / Math.cosh(s); // Hyperbolic cosecant
double squared = Math.pow(sech, 2);
return ((Math.pow(s, 0.5)) * squared);
}
public static double getNewSum(double s){
double constant = Math.pow(2, (2*s)-1) / (((Math.pow(2, s)) -2)*(gamma(1+s)));
return constant*Integrate(0, 1000);
}
// Gamma Function - Lanczos approximation
public static double gamma(double s){
double[] p = {0.99999999999980993, 676.5203681218851, -1259.1392167224028,
771.32342877765313, -176.61502916214059, 12.507343278686905,
-0.13857109526572012, 9.9843695780195716e-6, 1.5056327351493116e-7};
int g = 7;
if(s < 0.5) return Math.PI / (Math.sin(Math.PI * s)*gamma(1-s));
s -= 1;
double a = p[0];
double t = s+g+0.5;
for(int i = 1; i < p.length; i++){
a += p[i]/(s+i);
}
return Math.sqrt(2*Math.PI)*Math.pow(t, s+0.5)*Math.exp(-t)*a;
}
//Binomial Co-efficient - NOT CURRENTLY USING
/*
public static double binomial(int n, int k)
{
if (k>n-k)
k=n-k;
long b=1;
for (int i=1, m=n; i<=k; i++, m--)
b=b*m/i;
return b;
} */
// Riemann's Functional Equation
// Tried this initially and utterly failed.
public static double riemannFuncForm(double s) {
double term = Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s);
double nextTerm = Math.pow(2, (1-s))*Math.pow(Math.PI, (1-s)-1)*(Math.sin((Math.PI*(1-s))/2))*gamma(1-(1-s));
double error = Math.abs(term - nextTerm);
if(s == 1.0)
return 0;
else
return Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s)*standardZeta(1-s);
}
}
Ok well we've figured out that for this particular function, since this form of it isn't actually a infinite series, we cannot approximate using recursion. However the infinite sum of the Riemann Zeta series (1\(n^s) where n = 1 to infinity) could be solved through this method.
Additionally this method could be used to find any infinite series' sum, product, or limit.
If you execute the code your currently have, you'll get infinite recursion as 1-(1-s) = s (e.g. 1-s = t, 1-t = s so you'll just switch back and forth between two values of s infinitely).
Below I talk about the sum of series. It appears you are calculating the product of the series instead. The concepts below should work for either.
Besides this, the Riemann Zeta Function is an infinite series. This means that it only has a limit, and will never reach a true sum (in finite time) and so you cannot get an exact answer through recursion.
However, if you introduce a "threshold" factor, you can get an approximation that is as good as you like. The sum will increase/decrease as each term is added. Once the sum stabilizes, you can quit out of recursion and return your approximate sum. "Stabilized" is defined using your threshold factor. Once the sum varies by an amount less than this threshold factor (which you have defined), your sum has stabilized.
A smaller threshold leads to a better approximation, but also longer computation time.
(Note: this method only works if your series converges, if it has a chance of not converging, you might also want to build in a maxSteps variable to cease execution if the series hasn't converged to your satisfaction after maxSteps steps of recursion.)
Here's an example implementation, note that you'll have to play with threshold and maxSteps to determine appropriate values:
/* Riemann's Functional Equation
* threshold - if two terms differ by less than this absolute amount, return
* currSteps/maxSteps - if currSteps becomes maxSteps, give up on convergence and return
* currVal - the current product, used to determine threshold case (start at 1)
*/
public static double riemannFuncForm(double s, double threshold, int currSteps, int maxSteps, double currVal) {
double nextVal = currVal*(Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s)); //currVal*term
if( s == 1.0)
return 0;
else if ( s == 0.0)
return -0.5;
else if (Math.abs(currVal-nextVal) < threshold) //When a term will change the current answer by less than threshold
return nextVal; //Could also do currVal here (shouldn't matter much as they differ by < threshold)
else if (currSteps == maxSteps)//When you've taken the max allowed steps
return nextVal; //You might want to print something here so you know you didn't converge
else //Otherwise just keep recursing
return riemannFuncForm(1-s, threshold, ++currSteps, maxSteps, nextVal);
}
}
This is not possible.
The functional form of the Riemann Zeta Function is --
zeta(s) = 2^s pi^(-1+s) Gamma(1-s) sin((pi s)/2) zeta(1-s)
This is different from the standard equation in which an infinite sum is measured from 1/k^s for all k = 1 to k = infinity. It is possible to write this as something similar to --
// Standard form the the Zeta function.
public static double standardZeta(double s) {
int n = 1;
double currentSum = 0;
double relativeError = 1;
double error = 0.000001;
double remainder;
while (relativeError > error) {
currentSum = Math.pow(n, -s) + currentSum;
remainder = 1 / ((s-1)* Math.pow(n, (s-1)));
relativeError = remainder / currentSum;
n++;
}
System.out.println("The number of terms summed was " + n + ".");
return currentSum;
}
The same logic doesn't apply to the functional equation (it isn't a direct sum, it is a mathematical relationship). This would require a rather clever way of designing a program to calculate negative values of Zeta(s)!
The literal interpretation of this Java code is ---
// Riemann's Functional Equation
public static double riemannFuncForm(double s) {
double currentVal = (Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s));
if( s == 1.0)
return 0;
else if ( s == 0.0)
return -0.5;
else
System.out.println("Value of next value is " + nextVal(1-s));
return currentVal;//*nextVal(1-s);
}
public static double nextVal(double s)
{
return (Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s));
}
public static double getRiemannSum(double s) {
return riemannFuncForm(s);
}
Testing on three or four values shows that this doesn't work. If you write something similar to --
// Riemann's Functional Equation
public static double riemannFuncForm(double s) {
double currentVal = Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s); //currVal*term
if( s == 1.0)
return 0;
else if ( s == 0.0)
return -0.5;
else //Otherwise just keep recursing
return currentVal * nextVal(1-s);
}
public static double nextVal(double s)
{
return (Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s));
}
I was misinterpretation how to do this through mathematics. I will have to use a different approximation of the zeta function for values less than 2.
I think I need to use a different form of the zeta function. When I run the entire program ---
import java.util.Scanner;
public class Test4{
public static void main(String[] args) {
RiemannZetaMain func = new RiemannZetaMain();
double s = 0;
double start, stop, totalTime;
Scanner scan = new Scanner(System.in);
System.out.print("Enter the value of s inside the Riemann Zeta Function: ");
try {
s = scan.nextDouble();
}
catch (Exception e) {
System.out.println("You must enter a positive integer greater than 1.");
}
start = System.currentTimeMillis();
if(s >= 2)
System.out.println("Value for the Zeta Function = " + getStandardSum(s));
else
System.out.println("Value for the Zeta Function = " + getRiemannSum(s));
stop = System.currentTimeMillis();
totalTime = (double) (stop-start) / 1000.0;
System.out.println("Total time taken is " + totalTime + " seconds.");
}
// Standard form the the Zeta function.
public static double standardZeta(double s) {
int n = 1;
double currentSum = 0;
double relativeError = 1;
double error = 0.000001;
double remainder;
while (relativeError > error) {
currentSum = Math.pow(n, -s) + currentSum;
remainder = 1 / ((s-1)* Math.pow(n, (s-1)));
relativeError = remainder / currentSum;
n++;
}
System.out.println("The number of terms summed was " + n + ".");
return currentSum;
}
public static double getStandardSum(double s){
return standardZeta(s);
}
// Riemann's Functional Equation
public static double riemannFuncForm(double s, double threshold, double currSteps, int maxSteps) {
double term = Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s);
//double nextTerm = Math.pow(2, (1-s))*Math.pow(Math.PI, (1-s)-1)*(Math.sin((Math.PI*(1-s))/2))*gamma(1-(1-s));
//double error = Math.abs(term - nextTerm);
if(s == 1.0)
return 0;
else if (s == 0.0)
return -0.5;
else if (term < threshold) {//The recursion will stop once the term is less than the threshold
System.out.println("The number of steps is " + currSteps);
return term;
}
else if (currSteps == maxSteps) {//The recursion will stop if you meet the max steps
System.out.println("The series did not converge.");
return term;
}
else //Otherwise just keep recursing
return term*riemannFuncForm(1-s, threshold, ++currSteps, maxSteps);
}
public static double getRiemannSum(double s) {
double threshold = 0.00001;
double currSteps = 1;
int maxSteps = 1000;
return riemannFuncForm(s, threshold, currSteps, maxSteps);
}
// Gamma Function - Lanczos approximation
public static double gamma(double s){
double[] p = {0.99999999999980993, 676.5203681218851, -1259.1392167224028,
771.32342877765313, -176.61502916214059, 12.507343278686905,
-0.13857109526572012, 9.9843695780195716e-6, 1.5056327351493116e-7};
int g = 7;
if(s < 0.5) return Math.PI / (Math.sin(Math.PI * s)*gamma(1-s));
s -= 1;
double a = p[0];
double t = s+g+0.5;
for(int i = 1; i < p.length; i++){
a += p[i]/(s+i);
}
return Math.sqrt(2*Math.PI)*Math.pow(t, s+0.5)*Math.exp(-t)*a;
}
//Binomial Co-efficient
public static double binomial(int n, int k)
{
if (k>n-k)
k=n-k;
long b=1;
for (int i=1, m=n; i<=k; i++, m--)
b=b*m/i;
return b;
}
}
I notice that plugging in zeta(-1) returns -
Enter the value of s inside the Riemann Zeta Function: -1
The number of steps is 1.0
Value for the Zeta Function = -0.0506605918211689
Total time taken is 0.0 seconds.
I knew that this value was -1/12. I checked some other values with wolfram alpha and observed that --
double term = Math.pow(2, s)*Math.pow(Math.PI, s-1)*(Math.sin((Math.PI*s)/2))*gamma(1-s);
Returns the correct value. It is just that I am multiplying this value every time by zeta(1-s). In the case of Zeta(1/2), this will always multiply the result by 0.99999999.
Enter the value of s inside the Riemann Zeta Function: 0.5
The series did not converge.
Value for the Zeta Function = 0.999999999999889
Total time taken is 0.006 seconds.
I am going to see if I can replace the part for --
else if (term < threshold) {//The recursion will stop once the term is less than the threshold
System.out.println("The number of steps is " + currSteps);
return term;
}
This difference is the error between two terms in the summation. I may not be thinking about this correctly, it is 1:16am right now. Let me see if I can think better tomorrow ....
For an assignment I must create a method using a binary search to find the square root of an integer, and if it is not a square number, it should return an integer s such that s*s <= the number (so for 15 it would return 3). The code I have for it so far is
public class BinarySearch {
/**
* Integer square root Calculates the integer part of the square root of n,
* i.e. integer s such that s*s <= n and (s+1)*(s+1) > n
* requires n >= 0
*
* #param n number to find the square root of
* #return integer part of its square root
*/
private static int iSqrt(int n) {
int l = 0;
int r = n;
int m = ((l + r + 1) / 2);
// loop invariant
while (Math.abs(m * m - n) > 0) {
if ((m) * (m) > n) {
r = m;
m = ((l + r + 1) / 2);
} else {
l = m;
m = ((l + r + 1) / 2);
}
}
return m;
}
public static void main(String[] args) {
//gets stuck
System.out.println(iSqrt(15));
//calculates correctly
System.out.println(iSqrt(16));
}
}
And this returns the right number for square numbers, but gets stick in an endless loop for other integers. I know that the problem lies in the while condition, but I can't work out what to put due to the gap between square numbers getting much bigger as the numbers get bigger (so i can't just put that the gap must be below a threshold). The exercise is about invariants if that helps at all (hence why it is set up in this way). Thank you.
Think about it: Math.abs(m*m-n) > 0 is always true non-square numbers, because it is never zero, and .abs cannot be negative. It is your loop condition, that's why the loop never ends.
Does this give you enough info to get you going?
You need to change the while (Math.abs(m * m - n) > 0) to allow for a margin of error, instead of requiring it be exactly equal to zero as you do right now.
Try while((m+1)*(m+1) <= n || n < m * m)
#define EPSILON 0.0000001
double msqrt(double n){
assert(n >= 0);
if(n == 0 || n == 1){
return n;
}
double low = 1, high = n;
double mid = (low+high)/2.0;
while(abs(mid*mid - n) > EPSILON){
mid = (low+high)/2.0;
if(mid*mid < n){
low = mid+1;
}else{
high = mid-1;
}
}
return mid;}
As you can see above , you should simply apply binary search (bisection method)
and you can minimize Epsilon to get more accurate results but it will take more time to run.
Edit: I have written code in c++ (sorry)
As Ken Bloom said you have to have an error marge, 1. I've tested this code and it runs as expected for 15. Also you'll need to use float's, I think this algorithm is not possible for int's (although I have no mathematical proof)
private static int iSqrt(int n){
float l = 0;
float r = n;
float m = ((l + r)/2);
while (Math.abs(m*m-n) > 0.1) {
if ((m)*(m) > n) {
r=m;
System.out.println("r becomes: "+r);
} else {
l = m;
System.out.println("l becomes: "+l);
}
m = ((l + r)/2);
System.out.println("m becomes: "+m);
}
return (int)m;
}
How would I write a recursive static method that uses an (n+1) term MacLaurin series to compute e^x, called e(x,n), by using the following recursive formulation:
e(x,0)= 1
e(x,n)= e(x,n-1) + x^n/n!, if n>0
Also, my method signature needs to use the following:
public static double eTwo(double x, long n)
Been stuck for a while, any thoughts guys?
This is simplest solution that get on my mind, did you try it?
public static double eTwo(double x, long n){
if(n==0)
return 1;
else
return eTwo(x,n-1) + Math.pow(x, n)/factorial(n);
}
public double factorial (n){
if(n==0)
return 1;
else
return n*factorial(n-1);
}
A version which is likely a little more efficient would be to use a loop instead of recursion, because only a single stack frame needs to be allocated:
static double eTwo(double x, long n) {
double result = 1;
for (long i = 1; i < n; i++)
result += Math.pow(x, (double) i) / (double) factorial(i);
return result;
}
static long factorial(long n) {
long result = 1;
for (long i = 1; i <= n; i++)
result *= i;
return result;
}