We Keep Coding

Rabin-Karp not working for large primes (gives wrong output)

So I was solving this problem (Rabin Karp's algorithm) and wrote this solution:
private static void searchPattern(String text, String pattern) {
int txt_len = text.length(), pat_len = pattern.length();
int hash_pat = 0, hash_txt = 0; // hash values for pattern and text's substrings
final int mod = 100005; // prime number to calculate modulo... larger modulo denominator reduces collisions in hash
final int d = 256; // to include all the ascii character codes
int coeff = 1; // stores the multiplier (or coeffecient) for the first index of the sliding window
/*
* HASHING PATTERN:
* say text = "abcd", then
* hashed text = 256^3 *'a' + 256^2 *'b' + 256^1 *'c' + 256^0 *'d'
*/
// The value of coeff would be "(d^(pat_len - 1)) % mod"
for (int i = 0; i < pat_len - 1; i++)
coeff = (coeff * d) % mod;
// calculate hash of the first window and the pattern itself
for (int i = 0; i < pat_len; i++) {
hash_pat = (d * hash_pat + pattern.charAt(i)) % mod;
hash_txt = (d * hash_txt + text.charAt(i)) % mod;
}
for (int i = 0; i < txt_len - pat_len; i++) {
if (hash_txt == hash_pat) {
// our chances of collisions are quite less (1/mod) so we dont need to recheck the substring
System.out.println("Pattern found at index " + i);
}
hash_txt = (d * (hash_txt - text.charAt(i) * coeff) + text.charAt(i + pat_len)) % mod; // calculating next window (i+1 th index)
// We might get negative value of t, converting it to positive
if (hash_txt < 0)
hash_txt = hash_txt + mod;
}
if (hash_txt == hash_pat) // checking for the last window
System.out.println("Pattern found at index " + (txt_len - pat_len));
}
Now this code is simply not working if the mod = 1000000007, whereas as soon as we take some other prime number (large enough, like 1e5+7), the code magically starts working !
The line at which the code's logic failed is:
hash_txt = (d * (hash_txt - text.charAt(i) * coeff) + text.charAt(i + pat_len)) % mod;
Can someone please tell me why is this happening ??? Maybe this is a stupid doubt but I just do not understand.

In Java, an int is a 32-bit integer. If a calculation with such number mathematically yields a result that needs more binary digits, the extra digits are silently discarded. This is called overflow.
To avoid this, the Rabin-Karp algorithm reduces results modulo some prime in each step, thereby keeping the number small enough that the next step will not overflow. For this to work, the prime chosen must be suitably small that
d * (hash + max(char) * coeff) + max(char)) < max(int)
Since
0 ≤ hash < p,
1 ≤ coeff < p,
max(char) = 216
max(int) = 231
any prime smaller than 27=128 will do. For larger primes, it depends on what their coeff ends up being, but even if we select one with the smallest possible coeff = 1, the prime must not exceed 223, which is much smaller than the prime you used.
In practice, one therefore uses Rabin-Karp with an integer datatype that is significantly bigger that the character type, such as a long (64 bits). Then, any prime < 239 will do.
Even then, if it worth noting that your reasoning
our chances of collisions are quite less (1/mod) so we dont need to recheck the substring
is flawed, because the probability is determined not by chance, but by the strings being checked. Unless you know the probability distribution of your inputs, you can't know what the probability of failure is. That's why Rabin-Karp rechecks the string to make sure.

Algorithm to solve an equation

I have this problem for the course "Algorithm and data structures"
You have a equation x^2+s(x)+200·x=N, where x and N are natural numbers and S(x) is the sum of digits of number x.
On the input we have N and A, B such that A≤B and A, B≤1,000,000,000. You need to check if there is a natural number x in the interval [A, B] that solves the equation. If found you need to return that number, otherwise return -1.
Example Input:
1456
10 80
Output
-1
I managed to solve this problem by using some math and a bit modified version of brute force algorithm. But are there any more effective(algorithm based) ways to solve this problem?
This is my code:
import java.io.BufferedReader;
import java.io.InputStreamReader;
import java.util.StringTokenizer;
public class Range {
static int proveri(long N, long A, long B) {
long res = 0;
long start = (long)((-200 + Math.sqrt(4*N + 4))/2);
//System.out.println(start);
for (long i = Math.max(A, start); i <= B; i++) {
res = i * i + S(i) + 200 * i;
if(res == N)
return (int)i;
if(res > N)
return -1;
}
return -1;
}
static int S(long x) {
int sum = 0;
while(x > 0) {
sum += x % 10;
x /= 10;
}
return sum;
}
public static void main(String[] args) throws Exception {
int i,j,k;
BufferedReader br = new BufferedReader(new InputStreamReader(System.in));
long N = Long.parseLong(br.readLine());
StringTokenizer st = new StringTokenizer(br.readLine());
long A = Long.parseLong(st.nextToken());
long B = Long.parseLong(st.nextToken());
int res = proveri(N, A, B);
System.out.println(res);
br.close();
}
}

Here's a way where you can cut down on the amount of numbers you have to search.
Consider the equation anxn +
an-1xn-1 + ... + a1x + a0 = 0.
The rational root theorem states that if x = p/q is a solution,
then p divides a0 and q divides an
In your case, an is 1 and a0 is equal to S(x)-N. Thus, we know that any solution must divide S(x)-N.
This is where ben75's tip comes in. Since S(x) can't be bigger than 81, we can loop through all of the possible values of S(x), and solve separately. Something like this:
for each possible value of S(x)
loop through every factor x of S(x) - N
check if it is between A and B, if its digits sum to S(x)
and if it is a solution to x*x + 200x + S(x) = N.
if it is, return it.
return -1
There's also a pretty slick way for you to loop through all of the factors of a number, but I'll let you work that one out for yourself since this is for a course. My hint there is to look at the prime factorization of a number.

For the equation x^2+s(x)+200·x=N, consider
x^2 + 200·x + (N - s(x)) = 0
For a solution to a*x^2 + b*x + c = 0 equation with integer solutions, we need to have:
b^2 - 4*a*c >= 0 and must be a perfect square
Hence 200^2 - 4 * (N - s(x)) >=0 and a square or
10000 >= (N - s(x)) and (10,000 - (N - s(x)) must be a square. The square value is therefore less than 10,000 and hence there can be at most 100 values you need to check. With proper values of N it can be much lesser.
Also note that since N < 10,000, s(x) can be at most 36. These should cut down the range quite a bit.

Can I make this function more efficient (Project Euler Number 9)?

I just finished Project Euler problem 9 (warning spoilers):
A Pythagorean triplet is a set of three natural numbers, a < b < c, for which,
a^2 + b^2 = c^2
For example, 3^2 + 4^2 = 9 + 16 = 25 = 5^2.
There exists exactly one Pythagorean triplet for which a + b + c = 1000.
Find the product abc.
Here's my solution:
public static int specPyth(int num)
{
for (int a = 1; a < num; a++)
for (int b = 2; b < a; b++)
{
if (a*a +b*b == (num-a-b)*(num-a-b))
return a*b*(num-a-b); //ans = 31875000
}
return -1;
}
I can't help but think that there's a solution that involves only one loop. Anyone have ideas? I'd prefer answers using only one loop, but anything that's more efficient than what I currently have would be nice.

if a + b +c = 1000
then
a + b + sqroot(a² + b²) = 1000
-> (a² + b²) = (1000 - a - b)²
-> a² + b² = 1000000 - 2000*(a+b) + a² + 2*a*b + b²
-> 0 = 1000000 - 2000*(a+b) + 2*a*b
-> ... (easy basic maths)
-> a = (500000 - 1000*b) / (1000 - b)
Then you try every b until you find one that makes a natural number out of a.
public static int specPyth(int num)
{
double a;
for (int b = 1; b < num/2; b++)
{
a=(num*num/2 - num*b)/(num - b);
if (a%1 == 0)
return (int) (a*b*(num-a-b));
}
return -1;
}
EDIT: b can't be higher than 499, because c>b and (b+c) would then be higher than 1000.

I highly recommend reading http://en.wikipedia.org/wiki/Pythagorean_triple#Generating_a_triple and writing a function that will generate the Pythagorean triples one by one.
Not to give too much of a spoiler, but there are a number of other PE problems that this function will come in handy for.
(I don't consider this giving away too much, because part of the purpose of PE is to encourage people to learn about things like this.)

First, since a is the smallest, you need not to count it up to num, num/3 is sufficient, and even num/(2+sqrt(2)).
Second, having a and constraints
a+b+c=num
a^2+b^2=c^2
we can solve this equations and find b and c for given a, which already satisfy this equations and there is no need to check if a^2+b^2=c^2 as you do now. All you need is to check if b and c are integer. And this is done in one loop
for (int a = 1; a < num/3; a++)

Runs in 62 milli seconds
import time
s = time.time()
tag,n=True,1000
for a in xrange (1,n/2):
if tag==False:
break
for b in xrange (1,n/2):
if a*a + b*b - (n-a-b)*(n-a-b) ==0:
print a,b,n-a-b
print a*b*(n-a-b)
tag=False
print time.time() - s

C solution
Warning : solution assumes that GCD(a, b) = 1. It works here but may not always work. I'll fix the solution in some time.
#include <stdio.h>
#include <math.h>
int main(void)
{
int n = 1000; // a + b + c = n
n /= 2;
for(int r = (int) sqrt(n / 2); r <= (int) sqrt(n); r++)
{
if(n % r == 0)
{
int s = (n / r) - r;
printf("%d %d %d\n", r*r - s*s, 2*r*s, r*r + s*s);
printf("Product is %d\n", (2*r*s) * (r*r - s*s) * (r*r + s*s));
}
}
return 0;
}
Solution uses Euclid's formula for triplets which states that any primitive triple is of form a = r^2 - s^2, b = 2rs, c = r^2 + s^2.
Certain restrictions like sqrt(n / 2) <= r <= sqrt(n) can be added based on the fact that s is positive and r > s.
Warning: you may need long long if the product is large

Definitely not the most optimal solution, but my first instinct was to use a modified 3SUM. In Python,
def problem_9(max_value = 1000):
i = 0
range_of_values = [n for n in range(1, max_value + 1)]
while i < max_value - 3:
j = i + 1
k = max_value - 1
while j < k:
a = range_of_values[i]
b = range_of_values[j]
c = range_of_values[k]
if ((a + b + c) == 1000) and (a*a + b*b == c*c):
return a*b*c
elif (a + b + c) < 1000:
j += 1
else:
k -= 1
i += 1
return -1

You say a < b < c, then b must always be bigger than a, so your starting point in the second loop could be b = a + 1; that would lead certainly to fewer iterations.
int specPyth(int num)
{
for (int a = 1; a < num/3; a++)
for (int b = a + 1; b < num/2; b++)
{
int c = num - a - b;
if (a * a + b * b == c * c)
return a * b * c; //ans = 31875000
}
return -1;
}

In the first given equation, you have three variables a, b, c. If you want to find-out matching values for this equation, you have to run 3 dimension loop. Fortunately there is another equation a+b+c=N where N is known number.
Using this, you can reduce down the dimension to two because if you know two among the three, you can calculate the rest. For instance, if you know a and b, c equals N - a - b.
What if you can reduce one more dimension of the loop? It is possible if you fiddle with the two given equations. Get a pen and paper. Once you get the additional equation with two variables and one constant (N), you will be able to acquire the result in O(n). Solve the two equations a+b+c=n; a^2+b^2=c^2 taking n and a to be constant and solve for b and c:
public static void main(String[] args) {
Scanner in = new Scanner(System.in);
int t = in.nextInt();
for(int a0 = 0; a0 < t; a0++){
int n = in.nextInt();
int max=-1;
int multi=0;
int b=1,c=1;
for(int i=1;i<=n;i++)
{
if(2*(i-n)!=0)
b=(2*i*n-(n*n))/(2*(i-n));
c=n-b-i;
if( (i*i+b*b==c*c)&& i+b+c==n && b>0 && c>=0 && i+b>c && c+i>b && b+c>i)
{
multi=i*b*c;
if(max<multi)
max=multi;
}
}
if(max==-1)
System.out.println(-1);
else
System.out.println(max);
}
}

Python:
Before jump into the code, do a small exercise in algebra.
a^2 + b^2 = c ^2
a + b + c = 1000 --> c = 1000 - (a+b)
a^2 + b^2 = (1000 - (a+b))^2
a = 1000*(500-b)/(1000-b)
Now the time to write the code.
for b in range(2, 500):
a = 1000*(500 - b)/(1000 - b)
if a.is_integer():
c = 1000 - (a+b)
print(a, b, c, a*b*c)
break

Canonical coefficients from Newton polynomial

A while ago I implemented a Polynom approximation for a game I programmed.
I am using Newton's pyramide method.
It took me quite a while to figure it out, but my solution requires to calculate the binomial coefficients and I also have to sum up all the coefficients for the final coefficient of each power (since solving this problem is similar to squaring, cubing.. terms and calculating the binomial coefficients)
For example:
pick k out of n of the bionomeal terms and add them
one pick is multiplied
a*(x+b)(x+c)(x+d) ==> a*x^3 + a*x^2*(b+c+d) + a*x(bc+bd+cd) +a*b*c*d
so b*c*d would be one pick b*c and b*d too
My question now is:
Is there a way calculating the Polynominterpolation with the newton scheme without having to calculate all the bionomial coefficients?
My code:
https://github.com/superphil0/Polynominterpolation/blob/master/PolynomInterpolation.java
It works pretty good, although if one gives too many points it will be rather slow
because of the selection of terms which have all be summed up
(I am really bad at explaining this in english, I hope someone can understand what I want to know though)
cheers

Judging from this description, I take it that your “pyramid scheme” generates coefficients ci such that the polynomial p(x) can be written as
p(x) =
c0 + (x ‒ x0)(
c1 + (x ‒ x1)(
c2 + (x ‒ x2)(
c3 + (x ‒ x3)(
… (
cn-1 + (x ‒ xn‒1)
cn ) … ))))
Now you can compute canonical coefficients recursively from the back. Start with
pn = cn
In every step, the current polynomial can be written as
pk =
ck + (x ‒ xk)pk+1 =
ck + (x ‒ xk)(b0 +
b1x + b2x2 + …)
assuming that the next smaller polynomial has already been turned into canonical coefficients.
Now you can compute the coefficients ai of pk using those coefficients bi of pk+1. In a strict formal way, I'd have to use indices instead of a and b, but I believe that it's clearer this way. So what are the canonical coefficients of the next polynomial?
a0 = ck − xkb0
a1 = b0 − xkb1
a2 = b1 − xkb2
…
You can write this in a loop, using and reusing a single array a to hold the coefficients:
double[] a = new double[n + 1]; // initialized to zeros
for (int k = n; k >= 0; --k) {
for (int i = n - k; i > 0; --i)
a[i] = a[i - 1] - x[k]*a[i];
a[0] = c[k] - x[k]*a[0];
}

Java: Implementing simple equation

I am looking to implement the simple equation:
i,j = -Q ± √(Q2-4PR) / 2P
To do so I have the following code (note: P = 10. Q = 7. R = 10):
//Q*Q – 4PR = -351 mod 11 = -10 mod 11 = 1, √1 = 1
double test = Math.sqrt(modulo(((Q*Q) - ((4*P)*R))));
// Works, but why *-10 needed?
i = (int)(((-Q+test)/(P*2))*-10); // i = 3
j = (int)(((-Q-test)/(P*2))*-10); // j = 4
To put it simply, test takes the first part of the equation and mods it to a non-zero integer in-between 0 and 11, then i and j are written. i and j return the right number, but for some reason *-10 is needed to get them right (a number I guessed to get the correct values).
If possible, I'd like to find a better way of performing the above equation because my way of doing it seems wrong and just works. I'd like to do it as the equation suggests, rather than hack it to work.

The quadratic equation is more usually expressed in terms of a, b and c. To satisfy ax2+bx+c = 0, you get (-b +/- sqrt(b^2-4ac)) / 2a as answers.
I think your basic problem is that you're using modulo for some reason instead of taking the square root. The factor of -10 is just a fudge factor which happens to work for your test case.
You should have something like this:
public static void findRoots(double a, double b, double c)
{
if (b * b < 4 * a * c)
{
throw new IllegalArgumentException("Equation has no roots");
}
double tmp = Math.sqrt(b * b - 4 * a * c);
double firstRoot = (-b + tmp) / (2 * a);
double secondRoot = (-b - tmp) / (2 * a);
System.out.println("Roots: " + firstRoot + ", " + secondRoot);
}
EDIT: Your modulo method is currently going to recurse pretty chronically. Try this instead:
public static int modulo(int x)
{
return ((x % 11) + 11) % 11;
}
Basically the result of the first % 11 will be in the range [-10, 10] - so after adding another 11 and taking % 11 again, it'll be correct. No need to recurse.
At that point there's not much reason to have it as a separate method, so you can use:
public static void findRoots(double a, double b, double c)
{
int squareMod11 = (((b * b - 4 * a * c) % 11) + 11) % 11;
double tmp = Math.sqrt(squareMod11);
double firstRoot = (-b + tmp) / (2 * a);
double secondRoot = (-b - tmp) / (2 * a);
System.out.println("Roots: " + firstRoot + ", " + secondRoot);
}

You need to take the square root. Note that Q^2-4PR yields a negative number, and consequently you're going to have to handle complex numbers (or restrict input to avoid this scenario). Apache Math may help you here.

use Math.sqrt for the square root. Why do you cast i and j to ints? It is equation giving you roots of square function, so i and j can be any complex numbers. You shall limit the discriminant to positive-only values for real (double) roots, otherwise use complex numbers.
double test = Q*Q - 4*P*R;
if(Q < 0) throw new Exception("negative discriminant!");
else {
test = Math.sqrt(test);
double i = (-Q + test) / 2*P;
double i = (-Q - test) / 2*P;
}

Why are you doing modulo and not square root? Your code seems to be the way to get the roots of a quadratic equation ((a±sqrt(b^2-4ac))/2a), so the code should be:
double delta = Q*Q-4*P*R);
if(delta < 0.0) {
throw new Exception("no roots");
}
double d = Math.power(delta,0.5);
double r1 = (Q + d)/(2*P)
double r2 = (Q - d)/(2*P)

As pointed out by others, your use of mod isn't even wrong. Why are you making up mathematics like this?
It's well known that the naive solution to the quadratic equation can have problems if the value of b is very nearly equal to the discriminant.
A better way to do it is suggested in section 5.6 of "Numerical Recipes in C++": if we define
(source: equationsheet.com)
Then the two roots are:
and
Your code also needs to account for pathological cases (e.g., a = 0).
Let's substitute your values into these formulas and see what we get. If a = 10, b = 7, and c = 10, then :
(source: equationsheet.com)
Then the two roots are:
(source: equationsheet.com)
and
(source: equationsheet.com)
I think I have the signs right.
If your calculation is giving you trouble, it's likely due to the fact that you have complex roots that your method can't take into account properly. You'll need a complex number class.