Problem: Topcoder SRM 170 500
Consider a sequence {x0, x1, x2, ...}. A relation that defines some term xn in terms of previous terms is called a recurrence relation. A linear recurrence relation is one where the recurrence is of the form xn = c(k-1) * x(n-1) + c(k-2) * x(n-2) + ... + c(0) * x(n-k)
where all the c(i) are real-valued constants, k is the length of the recurrence relation, and n is an arbitrary positive integer which is greater than or equal to k.
You will be given a int[] coefficients, indicating, in order, c(0), c(1), ..., c(k-1). You will also be given a int[] initial, giving the values of x(0), x(1), ..., x(k-1), and an int N. Your method should return xN modulo 10.
More specifically, if coefficients is of size k, then the recurrence relation will be
xn = coefficients[k - 1] * xn-1 + coefficients[k - 2] * xn-2 + ... + coefficients[0] * xn-k.
For example, if coefficients = {2,1}, initial = {9,7}, and N = 6, then our recurrence relation is xn = xn-1 + 2 * xn-2 and we have x0 = 9 and x1 = 7. Then x2 = x1 + 2 * x0 = 7 + 2 * 9 = 25, and similarly, x3 = 39, x4 = 89, x5 = 167, and x6 = 345, so your method would return (345 modulo 10) = 5.
Constraints:
- Code must run in less than or equal to 2 seconds
- Memory utilization must not exceed 64 MB
My attempted Solution:
class RecurrenceRelation
{
public int moduloTen(int[] coefficients, int[] initial, int N)
{
double xn = 0; int j = 0;
int K = coefficients.Length;
List<double> xs = new List<double>(Array.ConvertAll<int, double>(initial,
delegate(int i)
{
return (double)i;
}));
if (N < K)
return negativePositiveMod(xs[N]);
while (xs.Count <= N)
{
for (int i = xs.Count - 1; i >= j; i--)
{
xn += xs[i] * coefficients[K--];
}
K = coefficients.Length;
xs.Add(xn);
xn = 0;
j++;
}
return negativePositiveMod(xs[N]);
}
public int negativePositiveMod(double b)
{
while (b < 0)
{
b += 10;
}
return (int)(b % 10);
}
}
My problem with this solution is the precision of the double representation, and since I can't use a third party library or the BigInteger library in .NET for this SRM, i need to find a way of solving it without them. I suspect I could use recursion but I'm a little clueless on how to go about that.
Here is a test case that shows when my code works and when it doesn't
{2,1}, {9,7}, 6 - Successfully returns 5
{9,8,7,6,5,4,3,2,1,0}, {1,2,3,4,5,6,7,8,9,10}, 654 - Unsuccessfully returns 8 instead of 5 due to precision of double type
Can anyone help me figure this out? I was going to consider using arrays to store the values but it is a little bit beyond me especially on how to cater for multiplication and still be within the time and space complexity set out in the problem. Perhaps my entire approach is wrong? I'd appreciate some pointers and direction (not fully fleshed out answers) answers please.
Thanks
Notice that we only need to return the modulo 10 of xn.
We also need to know that if a = b + c, we have a % 10 = (b % 10 + c % 10) %10.
And a = b*c, so we also have a % 10 = (b %10 * c % 10) % 10;
So, for
xn = c(k-1) * x(n-1) + c(k-2) * x(n-2) + ... + c(0) * x(n-k)
= a0 + a1 + .... + an
(with a0 = c(k - 1)*x(n-1), a1 = ...)
we have xn % 10 = (a0 % 10 + a1 % 10 + ...)%10
And for each ai = ci*xi, so ai % 10 = (ci % 10 * xi % 10)% 10.
So by doing all of these math calculations, we can avoid to use double and keep the result in manageable size.
As Pham has answered, the trick is to realize that you only need to return a modulo, thereby bypassing the problem of overflow. Here is my quick attempt. I use a queue to put in the last result xN, and evict the oldest one.
static int solve(int[] coefficients, int[] seed, int n)
{
int k = coefficients.Count();
var queue = new Queue<int>(seed.Reverse().Take(k).Reverse());
for (int i = k; i <= n; i++)
{
var xn = coefficients.Zip(queue, (x, y) => x * y % 10).Sum() % 10;
queue.Enqueue(xn);
queue.Dequeue();
}
return (int) (queue.Last() );
}
Edit:
Getting the same results as you expect, however I don't guarantee that there is no bug in this example.
Related
Update 01
Thanks to Caius, found the main problem, the logic on the "if" was wrong, now fixed and giving the correct results. The loop still create more positions than needed on the secondary List, an extra position for each number on the main List.
I've updated the code bellow for refence for the following question:
-001 I can figure out why it create positions that needed, the for loop should run only after the foreach finishes its loops correct?
-002 To kind of solving this issue, I've used a List.Remove() to remove all the 0's, so far no crashes, but, the fact that I'm creating the extra indexes, and than removing them, does means a big performance down if I have large list of numbers? Or is an acceptable solution?
Description
It supposed to read all numbers in a central List1 (numberList), and count how many numbers are inside a certain (0|-15 / 15|-20) range, for that I use another List, that each range is a position on the List2 (numberSubList), where each number on List2, tells how many numbers exists inside that range.
-The range changes as the numbers grows or decrease
Code:
void Frequency()
{
int minNumb = numberList.Min();
int maxNumb = numberList.Max();
int size = numberList.Count();
numberSubList.Clear();
dGrdVFrequency.Rows.Clear();
dGrdVFrequency.Refresh();
double k = (1 + 3.3 * Math.Log10(size));
double h = (maxNumb - minNumb) / k;
lblH.Text = $"H: {Math.Round(h, 2)} / Rounded = {Math.Round(h / 5) * 5}";
lblK.Text = $"K: {Math.Round(k, 4)}";
if (h <= 5) { h = 5; }
else { h = Math.Round(h / 5) * 5; }
int counter = 1;
for (int i = 0; i < size; i++)
{
numberSubList.Add(0); // 001 HERE, creating more positions than needed, each per number.
foreach (int number in numberList)
{
if (number >= (h * i) + minNumb && number < (h * (i + 1)) + minNumb)
{
numberSubList[i] = counter++;
}
}
numberSubList.Remove(0); // 002-This to remove all the extra 0's that are created.
counter = 1;
}
txtBoxSubNum.Clear();
foreach (int number in numberSubList)
{
txtBoxSubNum.AppendText($"{number.ToString()} , ");
}
lblSubTotalIndex.Text = $"Total in List: {numberSubList.Count()}";
lblSubSumIndex.Text = $"Sum of List: {numberSubList.Sum()}";
int inc = 0;
int sum = 0;
foreach (int number in numberSubList)
{
sum = sum + number;
int n = dGrdVFrequency.Rows.Add();
dGrdVFrequency.Rows[n].Cells[0].Value = $"{(h * inc) + minNumb} |- {(h * (1 + inc)) + minNumb}";
dGrdVFrequency.Rows[n].Cells[1].Value = $"{number}";
dGrdVFrequency.Rows[n].Cells[2].Value = $"{sum}";
dGrdVFrequency.Rows[n].Cells[3].Value = $"{(number * 100) / size} %";
dGrdVFrequency.Rows[n].Cells[4].Value = $"{(sum * 100) / size} %";
inc++;
}
}
Screen shot showing the updated version.
I think, if your aim is to only store eg 17 in the "15 to 25" slot, this is wonky:
if (number <= (h * i) + minNumb) // Check if number is smaller than the range limit
Because it's found inside a loop that will move on to the next range, "25 to 35" and it only asks if the number 17 is less than the upper limit (and 17 is less than 35) so 17 is accorded to the 25-35 range too
FWIW the range a number should be in can be derived from the number, with (number - min) / number_of_ranges - at the moment you create your eg 10 ranges and then you visit each number 10 times looking to put it in a range, so you do 9 times more operations than you really need to
In this case,this is the array which serves as coefficients and degrees which first value having no degree.
double[] arr = { 12, 2, 3 ,4};
I then made a method to print the above array in terms of polynomial equation.
It gives output in type string as follows :
2x^2 + 3x^3 + 4x^4 + 12
I want to a function which takes an argument x and then solve the above polynomial with respect to value of x.
How can I do that?
Any kind of help will be appreciated!.
Edit: Question Solved
To evaluate it you can simply sum the power values times the coefficients. Using LINQ, that's one line:
double result = arr.Select((c,i) => c * Math.Pow(x, i)).Sum();
Here i is the index into your array, it starts at zero, so x^0 = 1 * 12 == 12 etc.
You can also do it without LINQ like this:
List<string> debug = new List<string>();
double y = 1.0;
result = 0.0;
for (int i = 0; i < arr.Length; i++)
{
debug.Add($"{arr[i]} * x^{i}");
result = result + arr[i] * y;
y = y * x;
}
Console.WriteLine(string.Join(" + ", debug));
Console.WriteLine(result);
Which, for x=3 outputs:
12 * x^0 + 2 * x^1 + 3 * x^2 + 4 * x^3
153
Same result as LINQ.
This is what I created:
for (int i = 1; i < degree.Length; i++)
{
result_first += degree[i] * Math.Pow(x, degree[i]);
}
result_first += degree[0];
It works perfectly.
I have a task to find pairs of amicable numbers and I've already solved it. My solution is not efficient, so please help me to make my algorithm faster.
Amicable numbers are two different numbers so related that the sum of the proper divisors of each is equal to the other number. The smallest pair of amicable numbers is (220, 284). They are amicable because the proper divisors of 220 are 1, 2, 4, 5, 10, 11, 20, 22, 44, 55 and 110, of which the sum is 284; and the proper divisors of 284 are 1, 2, 4, 71 and 142, of which the sum is 220.
Task: two long numbers and find the first amicable numbers between them. Let s(n) be the sum of the proper divisors of n:
For example:
s(10) = 1 + 2 + 5 = 8
s(11) = 1
s(12) = 1 + 2 + 3 + 4 + 6 = 16
If s(firstlong) == s(secondLong) they are amicable numbers
My code:
public static IEnumerable<long> Ranger(long length) {
for (long i = 1; i <= length; i++) {
yield return i;
}
}
public static IEnumerable<long> GetDivisors(long num) {
return from a in Ranger(num/2)
where num % a == 0
select a;
}
public static string FindAmicable(long start, long limit) {
long numberN = 0;
long numberM = 0;
for (long n = start; n <= limit; n++) {
long sumN = GetDivisors(n).Sum();
long m = sumN;
long sumM = GetDivisors(m).Sum();
if (n == sumM ) {
numberN = n;
numberM = m;
break;
}
}
return $"First amicable numbers: {numberN} and {numberM}";
}
I generally don't write C#, so rather than stumble through some incoherent C# spaghetti, I'll describe an improvement in C#-madeup-psuedo-code.
The problem seems to be in your GetDivisors function. This is linear O(n) time with respect to each divisor n, when it could be O(sqrt(n)). The trick is to only divide up to the square root, and infer the rest of the factors from that.
GetDivisors(num) {
// same as before, but up to sqrt(num), plus a bit for floating point error
yield return a in Ranger((long)sqrt(num + 0.5)) where num % a == 0
if ((long)sqrt(num + 0.5) ** 2 == num) { // perfect square, exists
num -= 1 // don't count it twice
}
// repeat, but return the "other half"- num / a instead of a
yield return num/a in Ranger((long)sqrt(num + 0.5)) where num % a == 0
}
This will reduce your complexity of that portion from O(n) to O(sqrt(n)), which should provide a noticeable speedup.
There is a simple formula giving the sum of divisors of a number knowing its prime decomposition:
let x = p1^a1 * ... * pn^an, where pi is a prime for all i
sum of divisors = (1+p1+...+p1^a1) * ... (1+pn+...+pn^an)
= (1-p1^(a1+1))/(1-p1) * ... ((1-pn^(an+1))/(1-pn)
In order to do a prime decomposition you must compute all prime numbers up to the square root of the maximal value in your search range. This is easily done using the sieve of Erathostenes.
I have a lot of dislikes on this post, I'm not sure why, but for letting you guys help me out with this question, I will give you this script as a gift. This script converts experience to level and from level to experience points given an exponential expression. those constants ensure that level 100 will equal 10 million experience. In Runescape, their level 99 equals 13,032,xxx which is a strange number.
using System.IO;
using System;
class Program
{
const float salt = 2.82842712474619f;
const float factor = 0.64977928f;
const int lvl_100_XP = 10000000;
static void Main()
{
int xp = 9717096;// lvl 99
int lvl = ExperienceToLevel(xp);
Console.WriteLine("LVL: " + lvl.ToString()+ " XP: " + LevelToExperience(lvl).ToString());
}
static public int ExperienceToLevel(int xp){
int lvl = 0;
if (xp == lvl_100_XP){//9999987 is lvl 100 due to roundoff issues so it is fixed to 10mill
lvl = 100;
}
else{
lvl = (int)((1f / salt) * (float)Math.Pow((float)xp, (1f - factor)));
if (lvl == 0){
//lvl = 1;
}
}
if (lvl == 100 && xp < lvl_100_XP){
lvl = 99;
}
return lvl;
}
static public int LevelToExperience(int lvl){
int xp = 0;
if (lvl == 100){//9999987 is lvl 100 due to roundoff issues so it is fixed to 10mill
xp = lvl_100_XP;
}
else{
xp = (int)Math.Exp((float)Math.Log(salt * (float)lvl) / (1f - factor))+1;
if (xp <= 1){
xp= 0;
}
if (lvl == 100){
xp = lvl_100_XP;
}
}
return xp;
}
}
Let's work it out.
Let x be the experience, a and c are constants. L is the level. We notate exponentiation as ^; note that C# does not have such an operator. ^ in C# is XOR.
You have
b = x / a
d = x ^ c
L = b / d
so that's
L = x / (a * x ^ c)
which is
L = (1 / a) * (x / x ^ c)
which is
L = (1 / a) * x ^ (1 - c)
You wish to solve for x. So multiply both sides by 'a':
a * L = x ^ (1 - c)
Take the ln of both sides. (Or whatever logarithm you like best.)
ln (a * L) = (1 - c) ln (x)
Divide both sides by 1 - c
ln (a * L) / (1 - c) = ln x
And eliminate the ln; remember that exp is the inverse of ln. If you used some other logarithm, then use some other exponent.
exp (ln (a * L) / (1 - c)) = x
And we're done.
First of all, exp to level is a many to one relationship. The reverse is a one to many, so what exp are you aiming to return in lvl2exp? The minimum for that level? The maximum? A mean value?
I think your better off just computing once a table with experience brackets for each level (even cache it to a file if its not going to change) and simply doing a binary search to find the corresponding bracket for any given level.
I have this function wrote in C# to calc the sin(x). But when I try with x = 3.14, the printed result of sin X is NaN (not a number),
but when debugging, its is very near to 0.001592653
The value is not too big, neither too small. So how could the NaN appear here?
static double pow(double x, int mu)
{
if (mu == 0)
return 1;
if (mu == 1)
return x;
return x * pow(x, mu - 1);
}
static double fact(int n)
{
if (n == 1 || n == 0)
return 1;
return n * fact(n - 1);
}
static double sin(double x)
{
var s = x;
for (int i = 1; i < 1000; i++)
{
s += pow(-1, i) * pow(x, 2 * i + 1) / fact(2 * i + 1);
}
return s;
}
public static void Main(String[] param)
{
try
{
while (true)
{
Console.WriteLine("Enter x value: ");
double x = double.Parse(Console.ReadLine());
var sinX = sin(x);
Console.WriteLine("Sin of {0} is {1}: " , x , sinX);
Console.ReadLine();
}
}
catch (Exception ex)
{
Console.WriteLine(ex.Message);
}
}
It fails because both pow(x, 2 * i + 1) and fact(2 * i + 1) eventually return Infinity.
In my case, it's when x = 4, i = 256.
Note that pow(x, 2 * i + 1) = 4 ^ (2 * 257) = 2.8763090157797054523668883052624395737887631663 × 10^309 - a stupidly large number which is just over the max value of a double, which is approximately 1.79769313486232 x 10 ^ 308.
You might be interested in just using Math.Sin(x)
Also note that fact(2 * i + 1) = 513! =an even more ridiculously large number which is more than 10^1000 times larger than the estimated number of atoms in the observable universe.
When x == 3.14 and i == 314 then you get Infinity:
?pow(-1, 314)
1.0
?pow(x, 2 * 314 + 1)
Infinity
? fact(2 * 314 + 1)
Infinity
The problem here is an understanding of floating point representation of 'real' numbers.
Double numbers while allowing a large range of values only has a precision of 15 to 17 decimal digits.
In this example we are calculating a value between -1 and 1.
We calculate the value of the sin function by using the series expansion of it which is basically a the sum of terms. In that expansion the terms become smaller and smaller as we go along.
When the terms have reached a value less than 1e-17 adding them to what is already there will not make any difference. This is so because we only have 52 bit of precision which are used up by the time we get to a term of less than 1e-17.
So instead of doing a constant 1000 loops you should do something like this:
static double sin(double x)
{
var s = x;
for (int i = 1; i < 1000; i++)
{
var term = pow(x, 2 * i + 1) / fact(2 * i + 1);
if (term < 1e-17)
break;
s += pow(-1, i) * term;
}
return s;
}