probability of sum of dice rolling n times

I have trouble with me homework.

so the problem is this: i have n fair dices with 6 sides that roll all together and i need to calculate the probability of sum of dices side.

i need to write script when:

function prob = sumDicePDF(n,k)
input: 1. number of dices(n >= 1) 2. k - integer number from n to 6n
output: probability

example: sumDicePDF(3,4) ans = 0.0139 sumDicePDF(8,20) ans = 0.0218

i need to use equalition: probability = sigma[1:6](P(sum(n-1) = k - i) * 1/6

and i forbiden to use recursion so my script run time must be near n^2

i understood the problem but i cant find a way to biuld the loop for counting probability so my answer be   prob=(1/6)^n*sumOfCounting;

so my quastion is how to make loop to find sumOfCounting.

i wrote:

vec = ones(1,n - 1);
sumOfCounting = 0;
for i = 1:6
    for j = 1:length(vec)
       vec(j) = i;
       sumOfvalue = sum(vec);
       if k - sumOfvalue < 7 && k - sumOfvalue > 0
           for m = 1:length(vec)
               sumOfCounting = sumOfCounting + length(vec) + 1 - m;
           end
       end
    end
end

but it not counting all options so where is the problem?

thank and best wishes.

Answer 1

[Image Analyst]

Nice try, but why not just use the sum() function - that's what it's there for:

% Roll the dice "numberOfRolls" times
numberOfRolls = 200; % Number of times you roll all 6 dice.
n = 10; % Number of dice.
maxFaceValue = 6;
rolls = randi(maxFaceValue, n, numberOfRolls)

% Sum up dice values for each roll.
columnSums = sum(rolls, 1)
% Find out how many times each sum occurred.
edges = min(columnSums):max(columnSums)
counts = histc(columnSums, edges)
% Normalize
grandTotalSum = sum(counts(:))
normalizedCountSums = counts / grandTotalSum
bar(edges, normalizedCountSums, 'BarWidth', 1);
grid on;
% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]);
% Give a name to the title bar.
set(gcf,'name','Demo by ImageAnalyst','numbertitle','off') 
title('Frequency of Roll Sums', 'FontSize', 40);

[Ivan]

Thanks for help i learned your code but it still dont do what i asked for maybe i didnt explain well

my input is number of dices i have in hand and sum of their values after they were rolled once...

i need to find out what the probability to get that sum with the number of dices i have...

that why i didnt use sum function..

Thank you for help.

[Image Analyst]

You can set numberOfRolls = 1, but then that's just for one roll. You can't get reliable trustworthy probability data after only one roll. That's why I did 200 rolls. If you want to do it inefficiently, you could set numberOfRolls = 1 and then put the whole code in a loop over the number of rolls (like you tried to do), but this would be a very inefficient, non-MATLAB-ish way of working.

[Ivan]

Thank i found the way to answer it. if its interesting the function i was trying to write is:

    % function:  sumDicePDF
    % purpose:  Calculate the probability function of the sum of dice 
    %           Given n fair dice. Let define X(i) the value 
    %           out in the cube (i).
    %           Define Y(n) = sum(X(i)). 
    %           Calculate the probability function of Y(n). 
    %           Means the P(Y(n) = k)
    %           for given k.
    % input: n - number of dices. n >= 1 (integer)
    %        k - integer betwen n and 6n
    % output: probability - P(Y(n) = k)
    % example: sumDicePDF(3,4) ans = 0.0139
    %          sumDicePDF(8,20) ans = 0.0218

function prob=sumDicePDF(n,k)
    %Check the input 
    if n < 1 || k < n || k > 6*n
        display('Bad Input! Try Again.');
        n = input('n = ');
        k = input('k = ');
        prob=sumDicePDF(n,k);
        return;
    end
    %If there is only one dice return probability 1/6
    if n == 1
        prob = 1/6;
        return;
    end

      %We will use equalition: f(n,k)=sigma(i = 1:6)[f(n-1,k-i)*1/6]. The
      %idea is to create matrix (f(1,1) f(1,2) ... f(1,6n))
      %                         (...... ...... ... .......)
      %                         (f(n,1) f(n,2) ... f(n,6n))
      %And use it to find  out f(n,k) that we need with equalition

      valuesMatix = zeros(n - 1,6*n);  %Define new matrix size n-1,n*6 
                                       %with zeros
      %Fill the first row for 1<x<6 with 1/6 and the rest leave with zeros
      for i = 1:6
          valuesMatix(1,i) = 1/6;
      end
      %Fill the rest of matrix with relivant values 
      for i = 2:n - 1
          for j = i:6*i               
              probForSpecPlace = 0;
              for l = 1:6
                  if j - l > 0 %Check if value out of range becouse 
                               %(k - i) must be positive 
                      %Sum the Probabilites for specific f(n,k)
                      probForSpecPlace = probForSpecPlace + valuesMatix(i - 1,j - l)*1/6;
                  end
              end
              %Put that value in its place in matrix for futher use
              valuesMatix(i,j) = probForSpecPlace;               
          end
      end
      prob = 0; %Define the probability to 0
      %The last module: find the exact value for wanted probability using 
      %the equalition  
      %by using the created matrix of values
      for i = 1:6
          if k - i > 0
              prob = prob + valuesMatix(n - 1,k - i)*1/6;
          end
      end
  end

so it work like i wanted and for prove:

feature accel off; tic; sumDicePDF(8,20) , toc; feature accel on;
ans =
0.0218
Elapsed time is 0.004089 seconds.

Thanks a lot for helping.

[Image Analyst]

With lots of nested for loops, it doesn't look as efficient as mine, which I timed at 0.001 seconds, but whatever ... it's only 3 milliseconds (not noticeable at all) and it's your homework and if you understand yours better that's fine. At least I introduced you to a vectorized, fast way of doing it.

[Ivan]

You write but, your code make use of randomize, and its note needed. Becouse player already rolled his dices and get the sum, and he asked what was the probability that would get it for example:

he rolled 5 dices and get 20, so he asked me what was probability to get 20 by rolling 5 dices, answer is: 0.0837.

List of probability's for 1 to 25 dices can be fount here: http://wizardofodds.com/gambling/dice/2/

Still worked on your code: (Made i little change):

function normalizedCountSums=testMine(n,k)
% Roll the dice "numberOfRolls" times
numberOfRolls = 200; % Number of times you roll all 6 dice.
%n = 10; % Number of dice.
maxFaceValue = 6;
rolls = randi(maxFaceValue, n, numberOfRolls);
% Sum up dice values for each roll.
columnSums = sum(rolls, 1);
% Find out how many times each sum occurred.
edges = min(columnSums):max(columnSums);
counts = histc(columnSums, edges);
% Normalize
grandTotalSum = sum(counts(:));
normalizedCountSums = counts(k) / grandTotalSum;
% bar(edges, normalizedCountSums, 'BarWidth', 1);
% grid on;
% % Enlarge figure to full screen.
% set(gcf, 'units','normalized','outerposition',[0 0 1 1]);
% % Give a name to the title bar.
% set(gcf,'name','Demo by ImageAnalyst','numbertitle','off') 
% title('Frequency of Roll Sums', 'FontSize', 40);
end

the output was:

    >>  feature accel off; tic; testMine(8,20) , toc; feature accel on;

ans =

      0.0250

Elapsed time is 0.003182 seconds.
>>  feature accel off; tic; testMine(8,20) , toc; feature accel on;

ans =

      0.0300

Elapsed time is 0.001315 seconds.

Its Fast but the output is little wrong, it must be the same all the time = 0.0218

Maybe i wrote the problem not so good. but still thanks used it for other little program. :)