Introduction
I need to split an array filled with a certain type (let's take water buckets for example) with two values set (in this case weight and volume), while keeping the difference between the total of the weight to a minimum (preferred) and the difference between the total of the volumes less than 1000 (required). This doesn't need to be a full-fetched genetic algorithm or something similar, but it should be better than what I currently have...
Current Implementation
Due to not knowing how to do it better, I started by splitting the array in two same-length arrays (the array can be filled with an uneven number of items), replacing a possibly void spot with an item with both values being 0. The sides don't need to have the same amount of items, I just didn't knew how to handle it otherwise.
After having these distributed, I'm trying to optimize them like this:
func (main *Main) Optimize() {
for {
difference := main.Difference(WEIGHT)
for i := 0; i < len(main.left); i++ {
for j := 0; j < len(main.right); j++ {
if main.DifferenceAfter(i, j, WEIGHT) < main.Difference(WEIGHT) {
main.left[i], main.right[j] = main.right[j], main.left[i]
}
}
}
if difference == main.Difference(WEIGHT) {
break
}
}
for main.Difference(CAPACITY) > 1000 {
leftIndex := 0
rightIndex := 0
liters := 0
weight := 100
for i := 0; i < len(main.left); i++ {
for j := 0; j < len(main.right); j++ {
if main.DifferenceAfter(i, j, CAPACITY) < main.Difference(CAPACITY) {
newLiters := main.Difference(CAPACITY) - main.DifferenceAfter(i, j, CAPACITY)
newWeight := main.Difference(WEIGHT) - main.DifferenceAfter(i, j, WEIGHT)
if newLiters > liters && newWeight <= weight || newLiters == liters && newWeight < weight {
leftIndex = i
rightIndex = j
liters = newLiters
weight = newWeight
}
}
}
}
main.left[leftIndex], main.right[rightIndex] = main.right[rightIndex], main.left[leftIndex]
}
}
Functions:
main.Difference(const) calculates the absolute difference between the two sides, the constant taken as an argument decides the value to calculate the difference for
main.DifferenceAfter(i, j, const) simulates a swap between the two buckets, i being the left one and j being the right one, and calculates the resulting absolute difference then, the constant again determines the value to check
Explanation:
Basically this starts by optimizing the weight, which is what the first for-loop does. On every iteration, it tries every possible combination of buckets that can be switched and if the difference after that is less than the current difference (resulting in better distribution) it switches them. If the weight doesn't change anymore, it breaks out of the for-loop. While not perfect, this works quite well, and I consider this acceptable for what I'm trying to accomplish.
Then it's supposed to optimize the distribution based on the volume, so the total difference is less than 1000. Here I tried to be more careful and search for the best combination in a run before switching it. Thus it searches for the bucket switch resulting in the biggest capacity change and is also supposed to search for a tradeoff between this, though I see the flaw that the first bucket combination tried will set the liters and weight variables, resulting in the next possible combinations being reduced by a big a amount.
Conclusion
I think I need to include some more math here, but I'm honestly stuck here and don't know how to continue here, so I'd like to get some help from you, basically that can help me here is welcome.
As previously said, your problem is actually a constrained optimisation problem with a constraint on your difference of volumes.
Mathematically, this would be minimise the difference of volumes under constraint that the difference of volumes is less than 1000. The simplest way to express it as a linear optimisation problem would be:
min weights . x
subject to volumes . x < 1000.0
for all i, x[i] = +1 or -1
Where a . b is the vector dot product. Once this problem is solved, all indices where x = +1 correspond to your first array, all indices where x = -1 correspond to your second array.
Unfortunately, 0-1 integer programming is known to be NP-hard. The simplest way of solving it is to perform exhaustive brute force exploring of the space, but it requires testing all 2^n possible vectors x (where n is the length of your original weights and volumes vectors), which can quickly get out of hands. There is a lot of literature on this topic, with more efficient algorithms, but they are often highly specific to a particular set of problems and/or constraints. You can google "linear integer programming" to see what has been done on this topic.
I think the simplest might be to perform a heuristic-based brute force search, where you prune your search tree early when it would get you out of your volume constraint, and stay close to your constraint (as a general rule, the solution of linear optimisation problems are on the edge of the feasible space).
Here are a couple of articles you might want to read on this kind of optimisations:
UCLA Linear integer programming
MIT course on Integer programming
Carleton course on Binary programming
Articles on combinatorial optimisation & linear integer programming
If you are not familiar with optimisation articles or math in general, the wikipedia articles provides a good introduction, but most articles on this topic quickly show some (pseudo)code you can adapt right away.
If your n is large, I think at some point you will have to make a trade off between how optimal your solution is and how fast it can be computed. Your solution is probably suboptimal, but it is much faster than the exhaustive search. There might be a better trade off, depending on the exact configuration of your problem.
It seems that in your case, difference of weight is objective, while difference of volume is just a constraint, which means that you are seeking for solutions that optimize difference of weight attribute (as small as possible), and satisfy the condition on difference of volume attribute (total < 1000). In this case, it's a single objective constrained optimization problem.
Whereas, if you are interested in multi-objective optimization, maybe you wanna look at the concept of Pareto Frontier: http://en.wikipedia.org/wiki/Pareto_efficiency . It's good for keeping multiple good solutions with advantages in different objective, i.e., not losing diversity.
Related
Why is it that when I see example code for binary search there is never an if statement to check if the start of the array or end is the target?
import java.util.Arrays;
public class App {
public static int binary_search(int[] arr, int left, int right, int target) {
if (left > right) {
return -1;
}
int mid = (left + right) / 2;
if (target == arr[mid]) {
return mid;
}
if (target < arr[mid]) {
return binary_search(arr, left, mid - 1, target);
}
return binary_search(arr, mid + 1, right, target);
}
public static void main(String[] args) {
int[] arr = { 3, 2, 4, -1, 0, 1, 10, 20, 9, 7 };
Arrays.sort(arr);
for (int i = 0; i < arr.length; i++) {
System.out.println("Index: " + i + " value: " + arr[i]);
}
System.out.println(binary_search(arr, arr[0], arr.length - 1, -1));
}
}
in this example if the target was -1 or 20 the search would enter recursion. But it added an if statement to check if target is mid, so why not add two more statements also checking if its left or right?
EDIT:
As pointed out in the comments, I may have misinterpreted the initial question. The answer below assumes that OP meant having the start/end checks as part of each step of the recursion, as opposed to checking once before the recursion even starts.
Since I don't know for sure which interpretation was intended, I'm leaving this post here for now.
Original post:
You seem to be under the impression that "they added an extra check for mid, so surely they should also add an extra check for start and end".
The check "Is mid the target?" is in fact not a mere optimization they added. Recursively checking "mid" is the whole point of a binary search.
When you have a sorted array of elements, a binary search works like this:
Compare the middle element to the target
If the middle element is smaller, throw away the first half
If the middle element is larger, throw away the second half
Otherwise, we found it!
Repeat until we either find the target or there are no more elements.
The act of checking the middle is fundamental to determining which half of the array to continue searching through.
Now, let's say we also add a check for start and end. What does this gain us? Well, if at any point the target happens to be at the very start or end of a segment, we skip a few steps and end slightly sooner. Is this a likely event?
For small toy examples with a few elements, yeah, maybe.
For a massive real-world dataset with billions of entries? Hm, let's think about it. For the sake of simplicity, we assume that we know the target is in the array.
We start with the whole array. Is the first element the target? The odds of that is one in a billion. Pretty unlikely. Is the last element the target? The odds of that is also one in a billion. Pretty unlikely too. You've wasted two extra comparisons to speed up an extremely unlikely case.
We limit ourselves to, say, the first half. We do the same thing again. Is the first element the target? Probably not since the odds are one in half a billion.
...and so on.
The bigger the dataset, the more useless the start/end "optimization" becomes. In fact, in terms of (maximally optimized) comparisons, each step of the algorithm has three comparisons instead of the usual one. VERY roughly estimated, that suggests that the algorithm on average becomes three times slower.
Even for smaller datasets, it is of dubious use since it basically becomes a quasi-linear search instead of a binary search. Yes, the odds are higher, but on average, we can expect a larger amount of comparisons before we reach our target.
The whole point of a binary search is to reach the target with as few wasted comparisons as possible. Adding more unlikely-to-succeed comparisons is typically not the way to improve that.
Edit:
The implementation as posted by OP may also confuse the issue slightly. The implementation chooses to make two comparisons between target and mid. A more optimal implementation would instead make a single three-way comparison (i.e. determine ">", "=" or "<" as a single step instead of two separate ones). This is, for instance, how Java's compareTo or C++'s <=> normally works.
BambooleanLogic's answer is correct and comprehensive. I was curious about how much slower this 'optimization' made binary search, so I wrote a short script to test the change in how many comparisons are performed on average:
Given an array of integers 0, ... , N
do a binary search for every integer in the array,
and count the total number of array accesses made.
To be fair to the optimization, I made it so that after checking arr[left] against target, we increase left by 1, and similarly for right, so that every comparison is as useful as possible. You can try this yourself at Try it online
Results:
Binary search on size 10: Standard 29 Optimized 43 Ratio 1.4828
Binary search on size 100: Standard 580 Optimized 1180 Ratio 2.0345
Binary search on size 1000: Standard 8987 Optimized 21247 Ratio 2.3642
Binary search on size 10000: Standard 123631 Optimized 311205 Ratio 2.5172
Binary search on size 100000: Standard 1568946 Optimized 4108630 Ratio 2.6187
Binary search on size 1000000: Standard 18951445 Optimized 51068017 Ratio 2.6947
Binary search on size 10000000: Standard 223222809 Optimized 610154319 Ratio 2.7334
so the total comparisons does seem to tend to triple the standard number, implying the optimization becomes increasingly unhelpful for larger arrays. I'd be curious whether the limiting ratio is exactly 3.
To add some extra check for start and end along with the mid value is not impressive.
In any algorithm design the main concerned is moving around it's complexity either it is time complexity or space complexity. Most of the time the time complexity is taken as more important aspect.
To learn more about Binary Search Algorithm in different use case like -
If Array is not containing any repeated
If Array has repeated element in this case -
a) return leftmost index/value
b) return rightmost index/value
and many more point
I have created a model for solving the graph coloring problem in MiniZinc:
include "globals.mzn";
int: n_nodes; % Number of nodes
int: n_edges; % Number of edges
int: domain_ub; % Number of colors
array[int] of int: edges; % All edges of graph as a 1D array
array[1..n_edges, 1..2] of int: edges2d = array2d(1..n_edges, 1..2, edges);
array[1..n_nodes] of var 1..domain_ub: colors;
constraint forall (i in 1..n_edges) (colors[edges2d[i,1]] != colors[edges2d[i,2]]);
solve :: int_search(colors, dom_w_deg, indomain_random)
satisfy;
In order to tackle big problems (around 400-500 nodes), I start with an upper bound of the number of colors and solve successive satisfaction problems decrementing the number by one till it becomes unsatisfiable or times out. This method gives me decent results.
In order to improve my results, I added symmetry breaking constraints to the above model:
constraint colors[1] = 1;
constraint forall (i in 2..n_nodes) ( colors[i] in 1..max(colors[1..i-1])+1 );
This, however, brings down my results both speed-wise and quality-wise.
Why is my model performing badly after adding the additional constraints? How should I go about adding the symmetry breaking constraints?
For symmetry breaking for cases where the values are fully symmetric, I would recommend the seq_precede_chain constraint, which breaks that symmetry. As commented by #hakank, using indomain_random is probably not a good idea when used with symmetry breaking, indomain_min is a safer choice.
For graph coloring in general, it may help performance to run a clique-finding algorithm, and post all_different constraints over each cliques found. That would have to be done when generating a minizinc program for each instance. For comparison, see the Gecode graph coloring example which uses pre-computed cliques.
I know this is an old question, but I was working on the same problem and I wanted to write what I found about this topic that maybe it will be useful to someone in the future.
To improve the model the solution is to use symmetry breaking constraint, as you did, but in Minizinc there is a global constraint called value_precede which can be used in this case.
% A new color J is only allowed to appear after colors 0..J-1 have been seen before (in any order)
constraint forall(j in 1..n-1)(value_precede(j, j+1, map));
Changing the search heuristics the result does not improve much, I have tried different configurations and the best results are obtained using dom_w_deg and indomain_min (compared to my data files).
Another way to improve the results is to accept any good enough solution that's less than the number of colours in the domain.
But this model does not always lead to obtaining the optimal result.
include "globals.mzn";
int: n; % Number of nodes
int: e; % Number of edges
int: maxcolors = 17; % Domain of colors
array[1..e,1..2] of int: E; % 2d array, rows = edges, 2 cols = nodes per edge
array[0..n-1] of var 0..maxccolors: c; % Color of node n
constraint forall(i in 1..e)(c[E[i,1]] != c[E[i,2]] ); % Two linked nodes have diff color
constraint c[0] == 0; % Break Symmetry, force fist color == 0
% Big symmetry breaker. A new color J is only allowed to appear after colors
% 0..J-1 have been seen before (in any order)
constraint forall(i in 0..n-2)( value_precede(i,i+1, c) );
% Ideally solve would minimize(max(c)), but that's too slow, so we accept any good
% enough solution that's less equal our heuristic "maxcolors"
constraint max(c) <= maxcolors;
solve :: int_search(c, dom_w_deg, indomain_min, complete) satisfy;
output [ show(max(c)+1), "\n", show(c)]
A clear and complete explanation can be found here:
https://maxpowerwastaken.gitlab.io/model-idiot/posts/graph_coloring_and_minizinc/
I was reading book about competitive programming and was encountered to problem where we have to count all possible paths in the n*n matrix.
Now the conditions are :
`
1. All cells must be visited for once (cells must not be unvisited or visited more than once)
2. Path should start from (1,1) and end at (n,n)
3. Possible moves are right, left, up, down from current cell
4. You cannot go out of the grid
Now this my code for the problem :
typedef long long ll;
ll path_count(ll n,vector<vector<bool>>& done,ll r,ll c){
ll count=0;
done[r][c] = true;
if(r==(n-1) && c==(n-1)){
for(ll i=0;i<n;i++){
for(ll j=0;j<n;j++) if(!done[i][j]) {
done[r][c]=false;
return 0;
}
}
count++;
}
else {
if((r+1)<n && !done[r+1][c]) count+=path_count(n,done,r+1,c);
if((r-1)>=0 && !done[r-1][c]) count+=path_count(n,done,r-1,c);
if((c+1)<n && !done[r][c+1]) count+=path_count(n,done,r,c+1);
if((c-1)>=0 && !done[r][c-1]) count+=path_count(n,done,r,c-1);
}
done[r][c] = false;
return count;
}
Here if we define recurrence relation then it can be like: T(n) = 4T(n-1)+n2
Is this recurrence relation true? I don't think so because if we use masters theorem then it would give us result as O(4n*n2) and I don't think it can be of this order.
The reason, why I am telling, is this because when I use it for 7*7 matrix it takes around 110.09 seconds and I don't think for n=7 O(4n*n2) should take that much time.
If we calculate it for n=7 the approx instructions can be 47*77 = 802816 ~ 106. For such amount of instruction it should not take that much time. So here I conclude that my recurrene relation is false.
This code generates output as 111712 for 7 and it is same as the book's output. So code is right.
So what is the correct time complexity??
No, the complexity is not O(4^n * n^2).
Consider the 4^n in your notation. This means, going to a depth of at most n - or 7 in your case, and having 4 choices at each level. But this is not the case. In the 8th, level you still have multiple choices where to go next. In fact, you are branching until you find the path, which is of depth n^2.
So, a non tight bound will give us O(4^(n^2) * n^2). This bound however is far from being tight, as it assumes you have 4 valid choices from each of your recursive calls. This is not the case.
I am not sure how much tighter it can be, but a first attempt will drop it to O(3^(n^2) * n^2), since you cannot go from the node you came from. This bound is still far from optimal.
I was recently asked an interview question about testing the validity of a Sudoku board. A basic answer involves for loops. Essentially:
for(int x = 0; x != 9; ++x)
for(int y = 0; y != 9; ++y)
// ...
Do this nested for loops to check the rows. Do it again to check the columns. Do one more for the sub-squares but that one is more funky because we're dividing the suoku board into sub-boards so we end end up more than two nested loops, maybe three or four.
I was later asked the complexity of this code. Frankly, as far as I'm concerned, all the cells of the board are visited exactly three times so O(3n). To me, the fact that we have nested loops doesn't mean this code is automatically O(n^2) or even O(n^highest-nesting-level-of-loops). But I have suspicion that that's the answer the interviewer expected...
Posed another way, what is the complexity of these two pieces of code:
for(int i = 0; i != n; ++i)
// ...
and:
for(int i = 0; i != sqrt(n); ++i)
for(int j = 0; j != sqrt(n); ++j)
// ...
Your general intuition is correct. Let's clarify a bit about Big-O notation:
Big-O gives you an upper bound for the worst-case (time) complexity for your algorithm, in relation to n - the size of your input. In essence, it is a measurement of how the amount of work changes in relation to the size of the input.
When you say something like
all the cells of the board are visited exactly three times so O(3n).
you are implying that n (the size of your input) is the the number of cells in the board and therefore visiting all cells three times would indeed be an O(3n) (which is O(n)) operation. If this is the case you would be correct.
However usually when referring to Sudoku problems (or problems involving a grid in general), n is taken to be the number of cells in each row/column (an n x n board). In this case, the runtime complexity would be O(3n²) (which is indeed equal to O(n²)).
In the future, it is perfectly valid to ask your interviewer what n is.
As for the question in the title (Is a nested for loop automatically O(n^2)?) the short answer is no.
Consider this example:
for(int i = 0 ; i < n ; i++) {
for(int j = 0 ; j < n ; j * 2) {
... // some constant time operation
}
}
The outer loops makes n iterations while the inner loop makes log2(n) iterations - therefore the time complexity will be O(nlogn).
In your examples, in the first one you have a single for-loop making n iterations, therefore a complexity of (at least) O(n) (the operation is performed an order of n times).
In the second one you two nested for loops, each making sqrt(n) iterations, therefore a total runtime complexity of (at least) O(n) as well. The second function isn't automatically O(n^2) simply because it contains a nested loop. The amount of operations being made is still of the same order (n) therefore these two examples have the same complexity - since we assume n is the same for both examples.
This is the most crucial point to sail home. To compare between the performance of two algorithms, you must be using the same input to make the comparison. In your sudoku problem you could have defined n in a few different ways, and the way you did would directly affect the complexity calculation of the problem - even if the amount of work is all the same.
*NOTE - this is unrelated to your question, but in the future avoid using != in loop conditions. In your second example, if log(n) is not a whole number, the loop could run forever, depending on the language and how it is defined. It is therefore recommended to use < instead.
It depends on how you define the so-called N.
If the size of the board is N-by-N, then yes, the complexity is O(N^2).
But if you say, the total number of grids is N (i.e., the board id sqrt(N)-by-sqrt(N)), then the complexity is O(N), or 3O(N) if you mind the constant.
For the problem of producing a bit-pattern with exactly n set bits, I know of two practical methods, but they both have limitations I'm not happy with.
First, you can enumerate all of the possible word values which have that many bits set in a pre-computed table, and then generate a random index into that table to pick out a possible result. This has the problem that as the output size grows the list of candidate outputs eventually becomes impractically large.
Alternatively, you can pick n non-overlapping bit positions at random (for example, by using a partial Fisher-Yates shuffle) and set those bits only. This approach, however, computes a random state in a much larger space than the number of possible results. For example, it may choose the first and second bits out of three, or it might, separately, choose the second and first bits.
This second approach must consume more bits from the random number source than are strictly required. Since it is choosing n bits in a specific order when their order is unimportant, this means that it is making an arbitrary distinction between n! different ways of producing the same result, and consuming at least floor(log_2(n!)) more bits than are necessary.
Can this be avoided?
There is obviously a third approach of iteratively computing and counting off the legal permutations until a random index is reached, but that's simply a space-for-time trade-off on the first approach, and isn't directly helpful unless there is an efficient way to count off those n permutations.
clarification
The first approach requires picking a single random number between zero and (where w is the output size), as this is the number of possible solutions.
The second approach requires picking n random values between zero and w-1, zero and w-2, etc., and these have a product of , which is times larger than the first approach.
This means that the random number source has been forced to produce bits to distinguish n! different results which are all equivalent. I'd like to know if there's an efficient method to avoid relying on this superfluous randomness. Perhaps by using an algorithm which produces an un-ordered list of bit positions, or by directly computing the nth unique permutation of bits.
Seems like you want a variant of Floyd's algorithm:
Algorithm to select a single, random combination of values?
Should be especially useful in your case, because the containment test is a simple bitmask operation. This will require only k calls to the RNG. In the code below, I assume you have randint(limit) which produces a uniform random from 0 to limit-1, and that you want k bits set in a 32-bit int:
mask = 0;
for (j = 32 - k; j < 32; ++j) {
r = randint(j+1);
b = 1 << r;
if (mask & b) mask |= (1 << j);
else mask |= b;
}
How many bits of entropy you need here depends on how randint() is implemented. If k > 16, set it to 32 - k and negate the result.
Your alternative suggestion of generating a single random number representing one combination among the set (mathematicians would call this a rank of the combination) is simpler if you use colex order rather than lexicographic rank. This code, for example:
for (i = k; i >= 1; --i) {
while ((b = binomial(n, i)) > r) --n;
buf[i-1] = n;
r -= b;
}
will fill the array buf[] with indices from 0 to n-1 for the k-combination at colex rank r. In your case, you'd replace buf[i-1] = n with mask |= (1 << n). The binomial() function is binomial coefficient, which I do with a lookup table (see this). That would make the most efficient use of entropy, but I still think Floyd's algorithm would be a better compromise.
[Expanding my comment:] If you only have a little raw entropy available, then use a PRNG to stretch it further. You only need enough raw entropy to seed a PRNG. Use the PRNG to do the actual shuffle, not the raw entropy. For the next shuffle reseed the PRNG with some more raw entropy. That spreads out the raw entropy and makes less of a demand on your entropy source.
If you know exactly the range of numbers you need out of the PRNG, then you can, carefully, set up your own LCG PRNG to cover the appropriate range while needing the minimum entropy to seed it.
ETA: In C++there is a next_permutation() method. Try using that. See std::next_permutation Implementation Explanation for more.
Is this a theory problem or a practical problem?
You could still do the partial shuffle, but keep track of the order of the ones and forget the zeroes. There are log(k!) bits of unused entropy in their final order for your future consumption.
You could also just use the recurrence (n choose k) = (n-1 choose k-1) + (n-1 choose k) directly. Generate a random number between 0 and (n choose k)-1. Call it r. Iterate over all of the bits from the nth to the first. If we have to set j of the i remaining bits, set the ith if r < (i-1 choose j-1) and clear it, subtracting (i-1 choose j-1), otherwise.
Practically, I wouldn't worry about the couple of words of wasted entropy from the partial shuffle; generating a random 32-bit word with 16 bits set costs somewhere between 64 and 80 bits of entropy, and that's entirely acceptable. The growth rate of the required entropy is asymptotically worse than the theoretical bound, so I'd do something different for really big words.
For really big words, you might generate n independent bits that are 1 with probability k/n. This immediately blows your entropy budget (and then some), but it only uses linearly many bits. The number of set bits is tightly concentrated around k, though. For a further expected linear entropy cost, I can fix it up. This approach has much better memory locality than the partial shuffle approach, so I'd probably prefer it in practice.
I would use solution number 3, generate the i-th permutation.
But do you need to generate the first i-1 ones?
You can do it a bit faster than that with kind of divide and conquer method proposed here: Returning i-th combination of a bit array and maybe you can improve the solution a bit
Background
From the formula you have given - w! / ((w-n)! * n!) it looks like your problem set has to do with the binomial coefficient which deals with calculating the number of unique combinations and not permutations which deals with duplicates in different positions.
You said:
"There is obviously a third approach of iteratively computing and counting off the legal permutations until a random index is reached, but that's simply a space-for-time trade-off on the first approach, and isn't directly helpful unless there is an efficient way to count off those n permutations.
...
This means that the random number source has been forced to produce bits to distinguish n! different results which are all equivalent. I'd like to know if there's an efficient method to avoid relying on this superfluous randomness. Perhaps by using an algorithm which produces an un-ordered list of bit positions, or by directly computing the nth unique permutation of bits."
So, there is a way to efficiently compute the nth unique combination, or rank, from the k-indexes. The k-indexes refers to a unique combination. For example, lets say that the n choose k case of 4 choose 3 is taken. This means that there are a total of 4 numbers that can be selected (0, 1, 2, 3), which is represented by n, and they are taken in groups of 3, which is represented by k. The total number of unique combinations can be calculated as n! / ((k! * (n-k)!). The rank of zero corresponds to the k-index of (2, 1, 0). Rank one is represented by the k-index group of (3, 1, 0), and so forth.
Solution
There is a formula that can be used to very efficiently translate between a k-index group and the corresponding rank without iteration. Likewise, there is a formula for translating between the rank and corresponding k-index group.
I have written a paper on this formula and how it can be seen from Pascal's Triangle. The paper is called Tablizing The Binomial Coeffieicent.
I have written a C# class which is in the public domain that implements the formula described in the paper. It uses very little memory and can be downloaded from the site. It performs the following tasks:
Outputs all the k-indexes in a nice format for any N choose K to a file. The K-indexes can be substituted with more descriptive strings or letters.
Converts the k-index to the proper lexicographic index or rank of an entry in the sorted binomial coefficient table. This technique is much faster than older published techniques that rely on iteration. It does this by using a mathematical property inherent in Pascal's Triangle and is very efficient compared to iterating over the entire set.
Converts the index in a sorted binomial coefficient table to the corresponding k-index. The technique used is also much faster than older iterative solutions.
Uses Mark Dominus method to calculate the binomial coefficient, which is much less likely to overflow and works with larger numbers. This version returns a long value. There is at least one other method that returns an int. Make sure that you use the method that returns a long value.
The class is written in .NET C# and provides a way to manage the objects related to the problem (if any) by using a generic list. The constructor of this class takes a bool value called InitTable that when true will create a generic list to hold the objects to be managed. If this value is false, then it will not create the table. The table does not need to be created in order to use the 4 above methods. Accessor methods are provided to access the table.
There is an associated test class which shows how to use the class and its methods. It has been extensively tested with at least 2 cases and there are no known bugs.
The following tested example code demonstrates how to use the class and will iterate through each unique combination:
public void Test10Choose5()
{
String S;
int Loop;
int N = 10; // Total number of elements in the set.
int K = 5; // Total number of elements in each group.
// Create the bin coeff object required to get all
// the combos for this N choose K combination.
BinCoeff<int> BC = new BinCoeff<int>(N, K, false);
int NumCombos = BinCoeff<int>.GetBinCoeff(N, K);
// The Kindexes array specifies the indexes for a lexigraphic element.
int[] KIndexes = new int[K];
StringBuilder SB = new StringBuilder();
// Loop thru all the combinations for this N choose K case.
for (int Combo = 0; Combo < NumCombos; Combo++)
{
// Get the k-indexes for this combination.
BC.GetKIndexes(Combo, KIndexes);
// Verify that the Kindexes returned can be used to retrive the
// rank or lexigraphic order of the KIndexes in the table.
int Val = BC.GetIndex(true, KIndexes);
if (Val != Combo)
{
S = "Val of " + Val.ToString() + " != Combo Value of " + Combo.ToString();
Console.WriteLine(S);
}
SB.Remove(0, SB.Length);
for (Loop = 0; Loop < K; Loop++)
{
SB.Append(KIndexes[Loop].ToString());
if (Loop < K - 1)
SB.Append(" ");
}
S = "KIndexes = " + SB.ToString();
Console.WriteLine(S);
}
}
So, the way to apply the class to your problem is by considering each bit in the word size as the total number of items. This would be n in the n!/((k! (n - k)!) formula. To obtain k, or the group size, simply count the number of bits set to 1. You would have to create a list or array of the class objects for each possible k, which in this case would be 32. Note that the class does not handle N choose N, N choose 0, or N choose 1 so the code would have to check for those cases and return 1 for both the 32 choose 0 case and 32 choose 32 case. For 32 choose 1, it would need to return 32.
If you need to use values not much larger than 32 choose 16 (the worst case for 32 items - yields 601,080,390 unique combinations), then you can use 32 bit integers, which is how the class is currently implemented. If you need to use 64 bit integers, then you will have to convert the class to use 64 bit longs. The largest value that a long can hold is 18,446,744,073,709,551,616 which is 2 ^ 64. The worst case for n choose k when n is 64 is 64 choose 32. 64 choose 32 is 1,832,624,140,942,590,534 - so a long value will work for all 64 choose k cases. If you need numbers bigger than that, then you will probably want to look into using some sort of big integer class. In C#, the .NET framework has a BigInteger class. If you are working in a different language, it should not be hard to port.
If you are looking for a very good PRNG, one of the fastest, lightweight, and high quality output is the Tiny Mersenne Twister or TinyMT for short . I ported the code over to C++ and C#. it can be found here, along with a link to the original author's C code.
Rather than using a shuffling algorithm like Fisher-Yates, you might consider doing something like the following example instead:
// Get 7 random cards.
ulong Card;
ulong SevenCardHand = 0;
for (int CardLoop = 0; CardLoop < 7; CardLoop++)
{
do
{
// The card has a value of between 0 and 51. So, get a random value and
// left shift it into the proper bit position.
Card = (1UL << RandObj.Next(CardsInDeck));
} while ((SevenCardHand & Card) != 0);
SevenCardHand |= Card;
}
The above code is faster than any shuffling algorithm (at least for obtaining a subset of random cards) since it only works on 7 cards instead of 52. It also packs the cards into individual bits within a single 64 bit word. It makes evaluating poker hands much more efficient as well.
As a side, note, the best binomial coefficient calculator I have found that works with very large numbers (it accurately calculated a case that yielded over 15,000 digits in the result) can be found here.