Related
I need to find an efficient (pseudo)code to solve the following problem:
Given two sequences of (not necessarily distinct) integers (a[1], a[2], ..., a[n]) and (b[1], b[2], ..., b[n]), find the maximum d such that a[n-d+1] == b[1], a[n-d+2] == b[2], ..., and a[n] == b[d].
This is not homework, I actually came up with this when trying to contract two tensors along as many dimensions as possible. I suspect an efficient algorithm exists (maybe O(n)?), but I cannot come up with something that is not O(n^2). The O(n^2) approach would be the obvious loop on d and then an inner loop on the items to check the required condition until hitting the maximum d. But I suspect something better than this is possible.
You can utilize the z algorithm, a linear time (O(n)) algorithm that:
Given a string S of length n, the Z Algorithm produces an array Z
where Z[i] is the length of the longest substring starting from S[i]
which is also a prefix of S
You need to concatenate your arrays (b+a) and run the algorithm on the resulting constructed array till the first i such that Z[i]+i == m+n.
For example, for a = [1, 2, 3, 6, 2, 3] & b = [2, 3, 6, 2, 1, 0], the concatenation would be [2, 3, 6, 2, 1, 0, 1, 2, 3, 6, 2, 3] which would yield Z[10] = 2 fulfilling Z[i] + i = 12 = m + n.
For O(n) time/space complexity, the trick is to evaluate hashes for each subsequence. Consider the array b:
[b1 b2 b3 ... bn]
Using Horner's method, you can evaluate all the possible hashes for each subsequence. Pick a base value B (bigger than any value in both of your arrays):
from b1 to b1 = b1 * B^1
from b1 to b2 = b1 * B^1 + b2 * B^2
from b1 to b3 = b1 * B^1 + b2 * B^2 + b3 * B^3
...
from b1 to bn = b1 * B^1 + b2 * B^2 + b3 * B^3 + ... + bn * B^n
Note that you can evaluate each sequence in O(1) time, using the result of the previous sequence, hence all the job costs O(n).
Now you have an array Hb = [h(b1), h(b2), ... , h(bn)], where Hb[i] is the hash from b1 until bi.
Do the same thing for the array a, but with a little trick:
from an to an = (an * B^1)
from an-1 to an = (an-1 * B^1) + (an * B^2)
from an-2 to an = (an-2 * B^1) + (an-1 * B^2) + (an * B^3)
...
from a1 to an = (a1 * B^1) + (a2 * B^2) + (a3 * B^3) + ... + (an * B^n)
You must note that, when you step from one sequence to another, you multiply the whole previous sequence by B and add the new value multiplied by B. For example:
from an to an = (an * B^1)
for the next sequence, multiply the previous by B: (an * B^1) * B = (an * B^2)
now sum with the new value multiplied by B: (an-1 * B^1) + (an * B^2)
hence:
from an-1 to an = (an-1 * B^1) + (an * B^2)
Now you have an array Ha = [h(an), h(an-1), ... , h(a1)], where Ha[i] is the hash from ai until an.
Now, you can compare Ha[d] == Hb[d] for all d values from n to 1, if they match, you have your answer.
ATTENTION: this is a hash method, the values can be large and you may have to use a fast exponentiation method and modular arithmetics, which may (hardly) give you collisions, making this method not totally safe. A good practice is to pick a base B as a really big prime number (at least bigger than the biggest value in your arrays). You should also be careful as the limits of the numbers may overflow at each step, so you'll have to use (modulo K) in each operation (where K can be a prime bigger than B).
This means that two different sequences might have the same hash, but two equal sequences will always have the same hash.
This can indeed be done in linear time, O(n), and O(n) extra space. I will assume the input arrays are character strings, but this is not essential.
A naive method would -- after matching k characters that are equal -- find a character that does not match, and go back k-1 units in a, reset the index in b, and then start the matching process from there. This clearly represents a O(n²) worst case.
To avoid this backtracking process, we can observe that going back is not useful if we have not encountered the b[0] character while scanning the last k-1 characters. If we did find that character, then backtracking to that position would only be useful, if in that k sized substring we had a periodic repetition.
For instance, if we look at substring "abcabc" somewhere in a, and b is "abcabd", and we find that the final character of b does not match, we must consider that a successful match might start at the second "a" in the substring, and we should move our current index in b back accordingly before continuing the comparison.
The idea is then to do some preprocessing based on string b to log back-references in b that are useful to check when there is a mismatch. So for instance, if b is "acaacaacd", we could identify these 0-based backreferences (put below each character):
index: 0 1 2 3 4 5 6 7 8
b: a c a a c a a c d
ref: 0 0 0 1 0 0 1 0 5
For example, if we have a equal to "acaacaaca" the first mismatch happens on the final character. The above information then tells the algorithm to go back in b to index 5, since "acaac" is common. And then with only changing the current index in b we can continue the matching at the current index of a. In this example the match of the final character then succeeds.
With this we can optimise the search and make sure that the index in a can always progress forwards.
Here is an implementation of that idea in JavaScript, using the most basic syntax of that language only:
function overlapCount(a, b) {
// Deal with cases where the strings differ in length
let startA = 0;
if (a.length > b.length) startA = a.length - b.length;
let endB = b.length;
if (a.length < b.length) endB = a.length;
// Create a back-reference for each index
// that should be followed in case of a mismatch.
// We only need B to make these references:
let map = Array(endB);
let k = 0; // Index that lags behind j
map[0] = 0;
for (let j = 1; j < endB; j++) {
if (b[j] == b[k]) {
map[j] = map[k]; // skip over the same character (optional optimisation)
} else {
map[j] = k;
}
while (k > 0 && b[j] != b[k]) k = map[k];
if (b[j] == b[k]) k++;
}
// Phase 2: use these references while iterating over A
k = 0;
for (let i = startA; i < a.length; i++) {
while (k > 0 && a[i] != b[k]) k = map[k];
if (a[i] == b[k]) k++;
}
return k;
}
console.log(overlapCount("ababaaaabaabab", "abaababaaz")); // 7
Although there are nested while loops, these do not have more iterations in total than n. This is because the value of k strictly decreases in the while body, and cannot become negative. This can only happen when k++ was executed that many times to give enough room for such decreases. So all in all, there cannot be more executions of the while body than there are k++ executions, and the latter is clearly O(n).
To complete, here you can find the same code as above, but in an interactive snippet: you can input your own strings and see the result interactively:
function overlapCount(a, b) {
// Deal with cases where the strings differ in length
let startA = 0;
if (a.length > b.length) startA = a.length - b.length;
let endB = b.length;
if (a.length < b.length) endB = a.length;
// Create a back-reference for each index
// that should be followed in case of a mismatch.
// We only need B to make these references:
let map = Array(endB);
let k = 0; // Index that lags behind j
map[0] = 0;
for (let j = 1; j < endB; j++) {
if (b[j] == b[k]) {
map[j] = map[k]; // skip over the same character (optional optimisation)
} else {
map[j] = k;
}
while (k > 0 && b[j] != b[k]) k = map[k];
if (b[j] == b[k]) k++;
}
// Phase 2: use these references while iterating over A
k = 0;
for (let i = startA; i < a.length; i++) {
while (k > 0 && a[i] != b[k]) k = map[k];
if (a[i] == b[k]) k++;
}
return k;
}
// I/O handling
let [inputA, inputB] = document.querySelectorAll("input");
let output = document.querySelector("pre");
function refresh() {
let a = inputA.value;
let b = inputB.value;
let count = overlapCount(a, b);
let padding = a.length - count;
// Apply some HTML formatting to highlight the overlap:
if (count) {
a = a.slice(0, -count) + "<b>" + a.slice(-count) + "</b>";
b = "<b>" + b.slice(0, count) + "</b>" + b.slice(count);
}
output.innerHTML = count + " overlapping characters:\n" +
a + "\n" +
" ".repeat(padding) + b;
}
document.addEventListener("input", refresh);
refresh();
body { font-family: monospace }
b { background:yellow }
input { width: 90% }
a: <input value="acacaacaa"><br>
b: <input value="acaacaacd"><br>
<pre></pre>
There are two random functions f1(),f2().
f1() returns 1 with probability p1, and 0 with probability 1-p1.
f2() returns 1 with probability p2, and 0 with probability 1-p2.
I want to implement a new function f3() which returns 1 with probability p3(a given probability), and returns 0 with probability 1-p3. In the implemetion of function f3(), we can use function f1() and f2(), but you can't use any other random function.
If p3=0.5, an example of implemention:
int f3()
{
do
{
int a = f1();
int b = f1();
if (a==b) continue;
// when reachs here
// a==1 with probability p1(1-p1)
// b==1 with probability (1-p1)p1
if (a==1) return 1;//now returns 1 with probability 0.5
if (b==1) return 0;
}while(1)
}
This implemention of f3() will give a random function returns 1 with probability 0.5, and 0 with probability 0.5. But how to implement the f3() with p3=0.4? I have no idea.
I wonder, is that task possible? And how to implement f3()?
Thanks in advance.
p1 = 0.77 -- arbitrary value between 0 and 1
function f1()
if math.random() < p1 then
return 1
else
return 0
end
end
-- f1() is enough. We don't need f2()
p3 = 0.4 -- arbitrary value between 0 and 1
--------------------------
function f3()
left = 0
rigth = 1
repeat
middle = left + (right - left) * p1
if f1() == 1 then
right = middle
else
left = middle
end
if right < p3 then -- completely below
return 1
elseif left >= p3 then -- completely above
return 0
end
until false -- loop forever
end
This can be solved if p3 is a rational number.
We should use conditional probabilities for this.
For example, if you want to make this for p3=0.4, the method is the following:
Calculate the fractional form of p3. In our case it is p3=0.4=2/5.
Now generate as many random variables from the same distribution (let's say, from f1, we won't use f2 anyway) as the denominator, call them X1, X2, X3, X4, X5.
We should regenerate all these random X variables until their sum equals the numerator in the fractional form of p3.
Once this is achieved then we just return X1 (or any other Xn, where n was chosen independently of the values of the X variables). Since there are 2 1s among the 5 X variables (because their sum equals the numerator), the probability of X1 being 1 is exactly p3.
For irrational p3, the problem cannot be solved by using only f1. I'm not sure now, but I think, it can be solved for p3 of the form p1*q+p2*(1-q), where q is rational with a similar method, generating the appropriate amount of Xs with distribution f1 and Ys with distribution f2, until they have a specific predefined sum, and returning one of them. This still needs to be detailed.
First to say, that's a nice problem to tweak one's brain. I managed to solve the problem for p3 = 0.4, for what you just asked for! And I think, generalisation of such problem, is not so trivial. :D
Here is how, you can solve it for p3 = 0.4:
The intuition comes from your example. If we generate a number from f1() five times in an iteration, (see the code bellow), we can have 32 types of results like bellow:
1: 00000
2: 00001
3: 00010
4: 00011
.....
.....
32: 11111
Among these, there are 10 such results with exactly two 1's in it! After identifying this, the problem becomes simple. Just return 1 for any of the 4 combinations and return 0 for 6 others! (as probability 0.4 means getting 1, 4 times out of 10). You can do that like bellow:
int f3()
{
do{
int a[5];
int numberOfOneInA = 0;
for(int i = 0; i < 5; i++){
a[i] = f1();
if(a[i] == 1){
numberOfOneInA++;
}
}
if (numberOfOneInA != 2) continue;
else return a[0]; //out of 10 times, 4 times a[0] is 1!
}while(1)
}
Waiting to see a generalised solution.
Cheers!
Here is an idea that will work when p3 is of a form a/2^n (a rational number with a denominator that is a power of 2).
Generate n random numbers with probability distribution of 0.5:
x1, x2, ..., xn
Interpret this as a binary number in the range 0...2^n-1; each number in this range has equal probability. If this number is less than a, return 1, else return 0.
Now, since this question is in a context of computer science, it seems reasonable to assume that p3 is in a form of a/2^n (this a common representation of numbers in computers).
I implement the idea of anatolyg and Egor:
inline double random(void)
{
return static_cast<double>(rand()) / static_cast<double>(RAND_MAX);
}
const double p1 = 0.8;
int rand_P1(void)
{
return random() < p1;
}
int rand_P2(void)//return 0 with 0.5
{
int x, y; while (1)
{
mystep++;
x = rand_P1(); y = rand_P1();
if (x ^ y) return x;
}
}
double p3 = random();
int rand_P3(void)//anatolyg's idea
{
double tp = p3; int bit, x;
while (1)
{
if (tp * 2 >= 1) {bit = 1; tp = tp * 2 - 1;}
else {bit = 0; tp = tp * 2;}
x = rand_P2();
if (bit ^ x) return bit;
}
}
int rand2_P3(void)//Egor's idea
{
double left = 0, right = 1, mid;
while (1)
{
dashenstep++;
mid = left + (right - left) * p1;
int x = rand_P1();
if (x) right = mid; else left = mid;
if (right < p3) return 1;
if (left > p3) return 0;
}
}
With massive math computings, I get, assuming P3 is uniformly distributed in [0,1), then the expectation of Egor is (1-p1^2-(1-p1)^2)^(-1). And anatolyg is 2(1-p1^2-(1-p1)^2)^(-1).
Speaking Algorithmically , Yes It is possible to do that task done .
Even Programmatically , It is possible , but a complex problem .
Lets take an example .
Let
F1(1) = .5 which means F1(0) =.5
F2(2) = .8 which means F1(0) =.2
Let Suppose You need a F3, such that F3(1)= .128
Lets try Decomposing it .
.128
= (2^7)*(10^-3) // decompose this into know values
= (8/10)*(8/10)*(2/10)
= F2(1)&F2(1)*(20/100) // as no Fi(1)==2/10
= F2(1)&F2(1)*(5/10)*(4/10)
= F2(1)&F2(1)&F1(1)*(40/100)
= F2(1)&F2(1)&F1(1)*(8/10)*(5/10)
= F2(1)&F2(1)&F1(1)&F2(1)&F1(1)
So F3(1)=.128 if we define F3()=F2()&F2()&F2()&F1()&F1()
Similarly if you want F4(1)=.9 ,
You give it as F4(0)=F1(0) | F2(0) =F1(0)F2(0)=.5.2 =.1 ,which mean F4(1)=1-0.1=0.9
Which means F4 is zero only when both are zero which happens .
So making use this ( & , | and , not(!) , xor(^) if you want ) operations with a combinational use of f1,f2 will surely give you the F3 which is made purely out of f1,f2,
Which may be NP hard problem to find the combination which gives you the exact probability.
So, Finally the answer to your question , whether it is possible or not ? is YES and this is one way of doing it, may be many hacks can be made into it this to optimize this, which gives you any optimal way .
I try to calculate Monte Carlo pi function in R. I have some problems in the code.
For now I write this code:
ploscinaKvadrata <- 0
ploscinaKroga <- 0
n = 1000
for (i in i:n) {
x <- runif(1000, min= -1, max= 1)
y <- runif(1000, min= -1, max= 1)
if ((x^2 + y^2) <= 1) {
ploscinaKroga <- ploscinaKroga + 1
} else {
ploscinaKvadrata <- ploscinaKvadrata + 1
}
izracunPi = 4* ploscinaKroga/ploscinaKvadrata
}
izracunPi
This is not working, but I don't know how to fix it.
I would also like to write a code to plot this (with circle inside square and with dots).
Here is a vectorized version (and there was also something wrong with your math)
N <- 1000000
R <- 1
x <- runif(N, min= -R, max= R)
y <- runif(N, min= -R, max= R)
is.inside <- (x^2 + y^2) <= R^2
pi.estimate <- 4 * sum(is.inside) / N
pi.estimate
# [1] 3.141472
As far as plotting the points, you can do something like this:
plot.new()
plot.window(xlim = 1.1 * R * c(-1, 1), ylim = 1.1 * R * c(-1, 1))
points(x[ is.inside], y[ is.inside], pch = '.', col = "blue")
points(x[!is.inside], y[!is.inside], pch = '.', col = "red")
but I'd recommend you use a smaller N value, maybe 10000.
This is a fun game -- and there are a number of versions of it floating around the web. Here's one I hacked from the named source (tho' his code was somewhat naive).
from http://giventhedata.blogspot.com/2012/09/estimating-pi-with-r-via-mcs-dart-very.html
est.pi <- function(n){
# drawing in [0,1] x [0,1] covers one quarter of square and circle
# draw random numbers for the coordinates of the "dart-hits"
a <- runif(n,0,1)
b <- runif(n,0,1)
# use the pythagorean theorem
c <- sqrt((a^2) + (b^2) )
inside <- sum(c<1)
#outside <- n-inside
pi.est <- inside/n*4
return(pi.est)
}
Typo 'nside' to 'inside'
I have a legend that I need to be color coordinated and have pretty much an infinite range of colors (infinite in the terms of 50-100 identifiers. I would like to use the color spectrum like a rainbow and evenly distribute colors from the number of items in the legend. For example.
I have pool A, B, and C, and would like them to be distinguised by color. A is red, B is orange, and C is green, etc. I would like to dynamically have this figured out using the hex color code going from 000000 to FFFFFF. So my pools would evenly pick three colors in that range. If I had 10 pools, it evenly pick 10 colors and so on. What would be the best algorithm to do this?
This will be done in javascript, but would love to have it done in SASS.
I strongly recommend to let the algorithm be a human being: Let somebody, who knows the trade, select 100 colors for you (with the most pleasing to the eye first), use this as a static array and just use index 0..N-1.
That said: A common algorithm is to use the PAL/NTSC color model:
R=Y+V/0.88
G=Y-0.38*U-0.58*V
B=Y+U/0.49
With U/V being the coordinates of the color and Y being the brightness.
So you create a circle of radius 1 with center at 0/0 in the U/V-plane and mark N points on it - this can easily be done with U=cos(phi), V=sin(phi) and phi going 0..360 degrees in N steps. This gives you an array of N (U/V) tuples.
Using an eye-plesing Y (ca. 0.5) you calculate the RGB-Values like above, resulting in an array of N (R/G/B) tuples.
Not sure exactly what you're asking, but this may help. It spits out a color of the rainbow from w=0 (red) to w=1.0 (purple). If you wanted to divide it evenly between N colors, I would do this:
for(int i = 0; i<N; i++){
colors[i] = spectrum(i/((double)(N-1)));
}
Using this method:
public Color spectrum(double w) {
if (w>1)w=1;
if (w<0)w=0;
w=w*(645-380)+380;
double R,B,G;
if (w >= 380 && w < 440){
R = -(w - 440.) /(440. - 350.);
G = 0.0;
B = 1.0;
}
else if (w >= 440 && w < 490){
R = 0.0;
G = (w - 440.) /(490. - 440.);
B = 1.0;
}
else if (w >= 490 && w < 510){
R = 0.0;
G = 1.0;
B = (510-w) /(510. - 490.);
}
else if (w >= 510 && w < 580){
R = (w - 510.) /(580. - 510.);
G = 1.0;
B = 0.0;
}
else if (w >= 580 && w < 645){
R = 1.0;
G = -(w - 645.) /(645. - 580.);
B = 0.0;
}
else if (w >= 645 && w <= 780){
R = 1.0;
G = 0.0;
B = 0.0;
}
else{
R = 0.0;
G = 0.0;
B = 0.0;
}
return new Color(R,G,B);
}
Sorry it's not in javascript. I am much more fluent in Java. The numbers in there map to wavelengths in nanometers, so it's useful for all kinds of stuff. The R,G,B values can of course be converted to ints by multiplying by 255 and truncating decimals.
Problem: We have x checkboxes and we want to check y of them evenly.
Example 1: select 50 checkboxes of 100 total.
[-]
[x]
[-]
[x]
...
Example 2: select 33 checkboxes of 100 total.
[-]
[-]
[x]
[-]
[-]
[x]
...
Example 3: select 66 checkboxes of 100 total:
[-]
[x]
[x]
[-]
[x]
[x]
...
But we're having trouble to come up with a formula to check them in code, especially once you go 11/111 or something similar. Anyone has an idea?
Let's first assume y is divisible by x. Then we denote p = y/x and the solution is simple. Go through the list, every p elements, mark 1 of them.
Now, let's say r = y%x is non zero. Still p = y/x where / is integer devision. So, you need to:
In the first p-r elements, mark 1 elements
In the last r elements, mark 2 elements
Note: This depends on how you define evenly distributed. You might want to spread the r sections withx+1 elements in between p-r sections with x elements, which indeed is again the same problem and could be solved recursively.
Alright so it wasn't actually correct. I think this would do though:
Regardless of divisibility:
if y > 2*x, then mark 1 element every p = y/x elements, x times.
if y < 2*x, then mark all, and do the previous step unmarking y-x out of y checkboxes (so like in the previous case, but x is replaced by y-x)
Note: This depends on how you define evenly distributed. You might want to change between p and p+1 elements for example to distribute them better.
Here's a straightforward solution using integer arithmetic:
void check(char boxes[], int total_count, int check_count)
{
int i;
for (i = 0; i < total_count; i++)
boxes[i] = '-';
for (i = 0; i < check_count; i++)
boxes[i * total_count / check_count] = 'x';
}
total_count is the total number of boxes, and check_count is the number of boxes to check.
First, it sets every box to unchecked. Then, it checks check_count boxes, scaling the counter to the number of boxes.
Caveat: this is left-biased rather than right-biased like in your examples. That is, it prints x--x-- rather than --x--x. You can turn it around by replacing
boxes[i * total_count / check_count] = 'x';
with:
boxes[total_count - (i * total_count / check_count) - 1] = 'x';
Correctness
Assuming 0 <= check_count <= total_count, and that boxes has space for at least total_count items, we can prove that:
No check marks will overlap. i * total_count / check_count increments by at least one on every iteration, because total_count >= check_count.
This will not overflow the buffer. The subscript i * total_count / check_count
Will be >= 0. i, total_count, and check_count will all be >= 0.
Will be < total_count. When n > 0 and d > 0:
(n * d - 1) / d < n
In other words, if we take n * d / d, and nudge the numerator down, the quotient will go down, too.
Therefore, (check_count - 1) * total_count / check_count will be less than total_count, with the assumptions made above. A division by zero won't happen because if check_count is 0, the loop in question will have zero iterations.
Say number of checkboxes is C and the number of Xes is N.
You example states that having C=111 and N=11 is your most troublesome case.
Try this: divide C/N. Call it D. Have index in the array as double number I. Have another variable as counter, M.
double D = (double)C / (double)N;
double I = 0.0;
int M = N;
while (M > 0) {
if (checkboxes[Round(I)].Checked) { // if we selected it, skip to next
I += 1.0;
continue;
}
checkboxes[Round(I)].Checked = true;
M --;
I += D;
if (Round(I) >= C) { // wrap around the end
I -= C;
}
}
Please note that Round(x) should return nearest integer value for x.
This one could work for you.
I think the key is to keep count of how many boxes you expect to have per check.
Say you want 33 checks in 100 boxes. 100 / 33 = 3.030303..., so you expect to have one check every 3.030303... boxes. That means every 3.030303... boxes, you need to add a check. 66 checks in 100 boxes would mean one check every 1.51515... boxes, 11 checks in 111 boxes would mean one check every 10.090909... boxes, and so on.
double count = 0;
for (int i = 0; i < boxes; i++) {
count += 1;
if (count >= boxes/checks) {
checkboxes[i] = true;
count -= count.truncate(); // so 1.6 becomes 0.6 - resetting the count but keeping the decimal part to keep track of "partial boxes" so far
}
}
You might rather use decimal as opposed to double for count, or there's a slight chance the last box will get skipped due to rounding errors.
Bresenham-like algorithm is suitable to distribute checkboxes evenly. Output of 'x' corresponds to Y-coordinate change. It is possible to choose initial err as random value in range [0..places) to avoid biasing.
def Distribute(places, stars):
err = places // 2
res = ''
for i in range(0, places):
err = err - stars
if err < 0 :
res = res + 'x'
err = err + places
else:
res = res + '-'
print(res)
Distribute(24,17)
Distribute(24,12)
Distribute(24,5)
output:
x-xxx-xx-xx-xxx-xx-xxx-x
-x-x-x-x-x-x-x-x-x-x-x-x
--x----x----x---x----x--
Quick html/javascript solution:
<html>
<body>
<div id='container'></div>
<script>
var cbCount = 111;
var cbCheckCount = 11;
var cbRatio = cbCount / cbCheckCount;
var buildCheckCount = 0;
var c = document.getElementById('container');
for (var i=1; i <= cbCount; i++) {
// make a checkbox
var cb = document.createElement('input');
cb.type = 'checkbox';
test = i / cbRatio - buildCheckCount;
if (test >= 1) {
// check the checkbox we just made
cb.checked = 'checked';
buildCheckCount++;
}
c.appendChild(cb);
c.appendChild(document.createElement('br'));
}
</script>
</body></html>
Adapt code from one question's answer or another answer from earlier this month. Set N = x = number of checkboxes and M = y = number to be checked and apply formula (N*i+N)/M - (N*i)/M for section sizes. (Also see Joey Adams' answer.)
In python, the adapted code is:
N=100; M=33; p=0;
for i in range(M):
k = (N+N*i)/M
for j in range(p,k-1): print "-",
print "x",
p=k
which produces
- - x - - x - - x - - x - - [...] x - - x - - - x where [...] represents 25 --x repetitions.
With M=66 the code gives
x - x x - x x - x x - x x - [...] x x - x x - x - x where [...] represents mostly xx- repetitions, with one x- in the middle.
Note, in C or java: Substitute for (i=0; i<M; ++i) in place of for i in range(M):. Substitute for (j=p; j<k-1; ++j) in place of for j in range(p,k-1):.
Correctness: Note that M = x boxes get checked because print "x", is executed M times.
What about using Fisher–Yates shuffle ?
Make array, shuffle and pick first n elements. You do not need to shuffle all of them, just first n of array. Shuffling can be find in most language libraries.