I am looking to solve a Traveling Salesman type problem using a matrix in order to find the minimum time between transitions. The matrix looks something like this:
A = [inf 4 3 5;
1 inf 3 5;
4 5 inf 3;
6 7 1 inf]
The y-axis represents the "from" node and the x-axis represents the "to" node. I am trying to find the optimal time from node 1 to node 4. I was told that there is a Matlab function called "TravellingSalesman". Is that true, and if not, how would I go about solving this matrix?
Thanks!
Here's an outline of the brute-force algorithm to solve TSP for paths from node 1 to node n:
C = inf
P = zeros(1,n-2)
for each permutation P of the nodes [2..n-1]
// paths always start from node 1 and end on node n
C = A(1,P(1)) + A(P(1),P(2)) + A(P(2),P(3)) + ... +
A(P(n-3),P(n-2)) + A(P(n-2),n)
if C < minCost
minCost = C
minPath = P
elseif C == minCost // you only need this part if you want
minPath = [minPath; P] // ALL paths with the shortest distance
end
end
Note that the first and last factors in the sum are different because you know beforehand what the first and last nodes are, so you don't have to include them in the permutations. So in the example given, with n=4, there are actually only 2!=2 possible paths.
The list of permutations can be precalculated using perms(2:n-1), but that might involve storing a large matrix (n! x n). Or you can calculate the cost as you generate each permutation. There are several files on the Mathworks file exchange with names like nextPerm that should work for you. Either way, as n grows you're going to be generating a very large number of permutations and your calculations will take a very long time.
Related
EDIT TO REQUIREMENT 6 AND NEW REQUIREMENT ADDED
6) Exactly 4 columns/rows must have degree 3.
7) No two vertices of degree 3 are adjacent to each other.
My goal:
To generate and save all matrices that meet specific requirements. Then compare each matrix to additional matrices that have been manually entered previously to check for specific similarities. I can add more detail if somebody thinks it would be helpful. I believe I have the comparison aspect of the code sorted out already, so I am waiting on the matrix generation portion. I need to do this for multiple sizes but I will focus this question on the 10x10 case.
Specific requirements:
1) Must be a 10x10 matrix (representing a graph on 10 vertices).
2) Must be symmetric (representing an adjacency matrix).
3) Have a diagonal of 0s (no loops).
4) Only 1s and 0s (simple graph).
5) The entire matrix must have exactly 48 1s (the graph has 24 edges).
6) Each column/row must have either 3 or 6 1s (each node as degree 3 or 6).
Application:
I am investigating a conjecture and believe I have come up with a possible solution which could break down the conjecture into smaller pieces and possibly prove some aspects. I want to use brute force to show if my idea works for a small specific case. Also having a base code in place could allow for future modifications to test other possible cases or ideas.
Ideas and thought process:
I used the edges of a graph to manually input my comparison set. For example:
G9=graph([1 1 1 2 2 3 4 4 4 5 5 6 6 6 6 3 3 9 2 2 2 7 7 8],[2 3 4 3 4 4 5 6 7 6 7 7 3 9 10 9 10 10 7 8 9 8 9 9]);
I think this is the only graph, up to isomorphism, which meets the previously listed requirements.
My original thought was to create the possible matrices that satisfy the given conditions then compare them to my comparison set. I still think this is the best approach.
I foolishly attempted to generate random matrices, completely overlooking the massive number of possibilities. Using a while loop, I first generated a random matrix that satisfied the first four requirements. Then in separate nested for statements I checked for requirement 5 using numedes() and requirement 6 using all(mod(degree())). That was a bad approach for several fairly obvious reasons, but I learned a lot through the process and it led me to the code that should do my final comparisons.
This is the first time I have used Matlab so I am learning as I go. I have been working on this one code for nearly 2 weeks and do not know if what I have come up with is "good", but I am proud of what I have been able to do by myself. I have reached the point where I feel like I need some outside advice. I am open to any suggestions and any level of help. A reference to a source, a function suggestion, another approach, or a complete solution with a "plug and play" code would be appreciated. I do not shy away from putting forth the effort to achieve my goals.
I appreciate any feedback.
If you want to brute force it, you've got 3773655750150 possible configurations to test for 3-or-6-connectedness. I think you'll probably need more powerful math (Polya Enumeration Theorem? or some other combinatoric theorem I probably forgot) to solve this one.
edit: This recursive solution is much more constrained and is likely to finish in the next century.
E = containers.Map('KeyType', 'int32', 'ValueType', 'any');
for k = 0:9
E(k) = [];
end
foo(E, 3, 0);
foo(E, 6, 0);
function E = foo(E, D, n)
% E : graph edges (map)
% D : degree (3 or 6)
% n : current node
if (n == 9)
e_degree = cellfun(#length,E.values);
if all(e_degree) && all(~mod(e_degree,3))
print_E(E)
end
return
end
e = E(n); % existing edges
m = setdiff((n+1:9), e); % candidate new edges
K = D - length(e);
% if too many edges, return early
if (K < 0)
return
end
C = combnk(m, K);
N = size(C, 1);
for k = 1:N
c = C(k,:);
E(n) = unique([e, c]);
for kv = 1:K
v = c(kv);
E(v) = unique([E(v), n]);
end
% down the rabbit hole
E = foo(E, D, n + 1);
for D = 3:3:6
E = foo(E, D, n + 1);
end
% remove edges added in this loop
E(n) = setdiff(E(n), c);
for kv = 1:K
v = c(kv);
E(v) = setdiff(E(v), n);
end
end
end
function print_E(E)
for k = 0:9
fprintf('%i: ',k);
fprintf('%i ', E(k));
fprintf('\n');
end
fprintf('\n');
end
In order to use the stiffness method for trusses, I need to extract certain elements from a large global stiffness matrix.
Say I have a 9 x 9 matrix K representing a three-member truss. This means that the first 3 rows and columns correspond to the first node, the second set of three rows and columns with the second node, and the third with the third node. In the code is a vector zDisp that corresponds to each node that has zero displacement. On paper, a zero displacement of a node means you would cross out the rows and columns corresponding to that displacement, leaving you with a smaller and easier to work with K matrix. So if the first and third nodes have zero displacement, you would be left with a 3 x 3 matrix corresponding to the intersection of the middle three rows and the middle three columns.
I thought I could accomplish this one node at a time with a function like so:
function [ B ] = deleteNode( B, node )
%deleteNode removes the corresponding rows and vectors to a node that has
% zero deflection from the global stiffness matrix
% --- Problem line - this gets the first location in the matrix corresponding to the node
start = 3*node- 2;
for i = 0 : 2
B(start+i,:) = [];
B(:,start+i) = [];
end
end
So my main project would go something like
% Zero displacement nodes
zDisp = [1;
3;
];
% --- Create 9 x 9 global matrix Kg ---
% Make a copy of the global matrix
S = Kg;
for(i = 1 : length(zDisp))
S = deleteNode(S, zDisp(i));
end
This does not work because once the loop executes for node 1 and removes the first 3 rows and columns, the problem line in the function no longer works to find the correct location in the smaller matrix to find the node.
So I think this step needs to be executed all at once. I am thinking I may need to instead input which nodes are NOT zero displacement, and create a submatrix based off of that. Any tips on this? Been thinking on it awhile. Thanks all.
In your example, you want to remove rows/columns 1, 2, 3, 7, 8, and 9, so if zDisp=[1;3],
remCols=bsxfun(#plus,1:3,3*(zDisp-1))
If I understand correctly, you should just be able to first remove the columns given by zDisp:
S(remCols(:),:)=[]
then remove the rows:
S(:,remCols(:))=[]
I have a 3x3 matrix, A. I also compute a value, g, as the maximum eigen value of A. I am trying to change the element A(3,3) = 0 for all values from zero to one in 0.10 increments and then update g for each of the values. I'd like all of the other matrix elements to remain the same.
I thought a for loop would be the way to do this, but I do not know how to update only one element in a matrix without storing this update as one increasingly larger matrix. If I call the element at A(3,3) = p (thereby creating a new matrix Atry) I am able (below) to get all of the values from 0 to 1 that I desired. I do not know how to update Atry to get all of the values of g that I desire. The state of the code now will give me the same value of g for all iterations, as expected, as I do not know how to to update Atry with the different values of p to then compute the values for g.
Any suggestions on how to do this or suggestions for jargon or phrases for me to web search would be appreciated.
A = [1 1 1; 2 2 2; 3 3 0];
g = max(eig(A));
% This below is what I attempted to achieve my solution
clear all
p(1) = 0;
Atry = [1 1 1; 2 2 2; 3 3 p];
g(1) = max(eig(Atry));
for i=1:100;
p(i+1) = p(i)+ 0.01;
% this makes a one giant matrix, not many
%Atry(:,i+1) = Atry(:,i);
g(i+1) = max(eig(Atry));
end
This will also accomplish what you want to do:
A = #(x) [1 1 1; 2 2 2; 3 3 x];
p = 0:0.01:1;
g = arrayfun(#(x) eigs(A(x),1), p);
Breakdown:
Define A as an anonymous function. This means that the command A(x) will return your matrix A with the (3,3) element equal to x.
Define all steps you want to take in vector p
Then "loop" through all elements in p by using arrayfun instead of an actual loop.
The function looped over by arrayfun is not max(eig(A)) but eigs(A,1), i.e., the 1 largest eigenvalue. The result will be the same, but the algorithm used by eigs is more suited for your type of problem -- instead of computing all eigenvalues and then only using the maximum one, you only compute the maximum one. Needless to say, this is much faster.
First, you say 0.1 increments in the text of your question, but your code suggests you are actually interested in 0.01 increments? I'm going to operate under the assumption you mean 0.01 increments.
Now, with that out of the way, let me state what I believe you are after given my interpretation of your question. You want to iterate over the matrix A, where for each iteration you increase A(3, 3) by 0.01. Given that you want all values from 0 to 1, this implies 101 iterations. For each iteration, you want to calculate the maximum eigenvalue of A, and store all these eigenvalues in some vector (which I will call gVec). If this is correct, then I believe you just want the following:
% Specify the "Current" A
CurA = [1 1 1; 2 2 2; 3 3 0];
% Pre-allocate the values we want to iterate over for element (3, 3)
A33Vec = (0:0.01:1)';
% Pre-allocate a vector to store the maximum eigenvalues
gVec = NaN * ones(length(A33Vec), 1);
% Loop over A33Vec
for i = 1:1:length(A33Vec)
% Obtain the version of A that we want for the current i
CurA(3, 3) = A33Vec(i);
% Obtain the maximum eigen value of the current A, and store in gVec
gVec(i, 1) = max(eig(CurA));
end
EDIT: Probably best to paste this code into your matlab editor. The stack-overflow automatic text highlighting hasn't done it any favors :-)
EDIT: Go with Rody's solution (+1) - it is much better!
I am using 64 bit matlab with 32g of RAM (just so you know).
I have a file (vector) of 1.3 million numbers (integers). I want to make another vector of the same length, where each point is a weighted average of the entire first vector, weighted by the inverse distance from that position (actually it's position ^-0.1, not ^-1, but for example purposes). I can't use matlab's 'filter' function, because it can only average things before the current point, right? To explain more clearly, here's an example of 3 elements
data = [ 2 6 9 ]
weights = [ 1 1/2 1/3; 1/2 1 1/2; 1/3 1/2 1 ]
results=data*weights= [ 8 11.5 12.666 ]
i.e.
8 = 2*1 + 6*1/2 + 9*1/3
11.5 = 2*1/2 + 6*1 + 9*1/2
12.666 = 2*1/3 + 6*1/2 + 9*1
So each point in the new vector is the weighted average of the entire first vector, weighting by 1/(distance from that position+1).
I could just remake the weight vector for each point, then calculate the results vector element by element, but this requires 1.3 million iterations of a for loop, each of which contains 1.3million multiplications. I would rather use straight matrix multiplication, multiplying a 1x1.3mil by a 1.3milx1.3mil, which works in theory, but I can't load a matrix that large.
I am then trying to make the matrix using a shell script and index it in matlab so only the relevant column of the matrix is called at a time, but that is also taking a very long time.
I don't have to do this in matlab, so any advice people have about utilizing such large numbers and getting averages would be appreciated. Since I am using a weight of ^-0.1, and not ^-1, it does not drop off that fast - the millionth point is still weighted at 0.25 compared to the original points weighting of 1, so I can't just cut it off as it gets big either.
Hope this was clear enough?
Here is the code for the answer below (so it can be formatted?):
data = load('/Users/mmanary/Documents/test/insertion.txt');
data=data.';
total=length(data);
x=1:total;
datapad=[zeros(1,total) data];
weights = ([(total+1):-1:2 1:total]).^(-.4);
weights = weights/sum(weights);
Fdata = fft(datapad);
Fweights = fft(weights);
Fresults = Fdata .* Fweights;
results = ifft(Fresults);
results = results(1:total);
plot(x,results)
The only sensible way to do this is with FFT convolution, as underpins the filter function and similar. It is very easy to do manually:
% Simulate some data
n = 10^6;
x = randi(10,1,n);
xpad = [zeros(1,n) x];
% Setup smoothing kernel
k = 1 ./ [(n+1):-1:2 1:n];
% FFT convolution
Fx = fft(xpad);
Fk = fft(k);
Fxk = Fx .* Fk;
xk = ifft(Fxk);
xk = xk(1:n);
Takes less than half a second for n=10^6!
This is probably not the best way to do it, but with lots of memory you could definitely parallelize the process.
You can construct sparse matrices consisting of entries of your original matrix which have value i^(-1) (where i = 1 .. 1.3 million), multiply them with your original vector, and sum all the results together.
So for your example the product would be essentially:
a = rand(3,1);
b1 = [1 0 0;
0 1 0;
0 0 1];
b2 = [0 1 0;
1 0 1;
0 1 0] / 2;
b3 = [0 0 1;
0 0 0;
1 0 0] / 3;
c = sparse(b1) * a + sparse(b2) * a + sparse(b3) * a;
Of course, you wouldn't construct the sparse matrices this way. If you wanted to have less iterations of the inside loop, you could have more than one of the i's in each matrix.
Look into the parfor loop in MATLAB: http://www.mathworks.com/help/toolbox/distcomp/parfor.html
I can't use matlab's 'filter' function, because it can only average
things before the current point, right?
That is not correct. You can always add samples (i.e, adding or removing zeros) from your data or from the filtered data. Since filtering with filter (you can also use conv by the way) is a linear action, it won't change the result (it's like adding and removing zeros, which does nothing, and then filtering. Then linearity allows you to swap the order to add samples -> filter -> remove sample).
Anyway, in your example, you can take the averaging kernel to be:
weights = 1 ./ [3 2 1 2 3]; % this kernel introduces a delay of 2 samples
and then simply:
result = filter(w,1,[data, zeros(1,3)]); % or conv (data, w)
% removing the delay introduced by the kernel
result = result (3:end-1);
You considered only 2 options:
Multiplying 1.3M*1.3M matrix with a vector once or multiplying 2 1.3M vectors 1.3M times.
But you can divide your weight matrix to as many sub-matrices as you wish and do a multiplication of n*1.3M matrix with the vector 1.3M/n times.
I assume that the fastest will be when there will be the smallest number of iterations and n is such that creates the largest sub-matrix that fits in your memory, without making your computer start swapping pages to your hard drive.
with your memory size you should start with n=5000.
you can also make it faster by using parfor (with n divided by the number of processors).
The brute force way will probably work for you, with one minor optimisation in the mix.
The ^-0.1 operations to create the weights will take a lot longer than the + and * operations to compute the weighted-means, but you re-use the weights across all the million weighted-mean operations. The algorithm becomes:
Create a weightings vector with all the weights any computation would need:
weights = (-n:n).^-0.1
For each element in the vector:
Index the relevent portion of the weights vector to consider the current element as the 'centre'.
Perform the weighted-mean with the weights portion and the entire vector. This can be done with a fast vector dot-multiply followed by a scalar division.
The main loop does n^2 additions and subractions. With n equal to 1.3 million that's 3.4 trillion operations. A single core of a modern 3GHz CPU can do say 6 billion additions/multiplications a second, so that comes out to around 10 minutes. Add time for indexing the weights vector and overheads, and I still estimate you could come in under half an hour.
I'm desperately trying to avoid a for loop in Matlab, but I cannot figure out how to do it. Here's the situation:
I have two m x n matrices A and B and two vectors v and w of length d. I want to outer multiply A and v so that I get an m x n x d matrix where the (i,j,k) entry is A_(i,j) * v_k, and similarly for B and w.
Afterward, I want to add the resulting m x n x d matrices, and then take the mean along the last dimension to get back an m x n matrix.
I'm pretty sure I could handle the latter part, but the first part has me completely stuck. I tried using bsxfun to no avail. Anyone know an efficient way to do this? Thanks very much!
EDIT: This revision comes after the three great answers below. gnovice has the best answer to the question I asked without a doubt. However,the question that I meant to ask involves squaring each entry before taking the mean. I forgot to mention this part originally. Given this annoyance, both of the other answers work well, but the clever trick of doing algebra before coding doesn't help this time. Thanks for the help, everyone!
EDIT:
Even though the problem in the question has been updated, an algebraic approach can still be used to simplify matters. You still don't have to bother with 3-D matrices. Your result is just going to be this:
output = mean(v.^2).*A.^2 + 2.*mean(v.*w).*A.*B + mean(w.^2).*B.^2;
If your matrices and vectors are large, this solution will give you much better performance due to the reduced amount of memory required as compared to solutions using BSXFUN or REPMAT.
Explanation:
Assuming M is the m-by-n-by-d matrix that you get as a result before taking the mean along the third dimension, this is what a span along the third dimension will contain:
M(i,j,:) = A(i,j).*v + B(i,j).*w;
In other words, the vector v scaled by A(i,j) plus the vector w scaled by B(i,j). And this is what you get when you apply an element-wise squaring:
M(i,j,:).^2 = (A(i,j).*v + B(i,j).*w).^2;
= (A(i,j).*v).^2 + ...
2.*A(i,j).*B(i,j).*v.*w + ...
(B(i,j).*w).^2;
Now, when you take the mean across the third dimension, the result for each element output(i,j) will be the following:
output(i,j) = mean(M(i,j,:).^2);
= mean((A(i,j).*v).^2 + ...
2.*A(i,j).*B(i,j).*v.*w + ...
(B(i,j).*w).^2);
= sum((A(i,j).*v).^2 + ...
2.*A(i,j).*B(i,j).*v.*w + ...
(B(i,j).*w).^2)/d;
= sum((A(i,j).*v).^2)/d + ...
sum(2.*A(i,j).*B(i,j).*v.*w)/d + ...
sum((B(i,j).*w).^2)/d;
= A(i,j).^2.*mean(v.^2) + ...
2.*A(i,j).*B(i,j).*mean(v.*w) + ...
B(i,j).^2.*mean(w.^2);
Try reshaping the vectors v and w to be 1 x 1 x d:
mean (bsxfun(#times, A, reshape(v, 1, 1, [])) ...
+ bsxfun(#times, B, reshape(w, 1, 1, [])), 3)
Here I am using [] in the argument to reshape to tell it to fill that dimension in based on the product of all the other dimensions and the total number of elements in the vector.
Use repmat to tile the matrix in the third dimension.
A =
1 2 3
4 5 6
>> repmat(A, [1 1 10])
ans(:,:,1) =
1 2 3
4 5 6
ans(:,:,2) =
1 2 3
4 5 6
etc.
You still don't have to resort to any explicit loops or indirect looping using bsxfun et al. for your updated requirements. You can achieve what you want by a simple vectorized solution as follows
output = reshape(mean((v(:)*A(:)'+w(:)*B(:)').^2),size(A));
Since OP only says that v and w are vectors of length d, the above solution should work for both row and column vectors. If they are known to be column vectors, v(:) can be replaced by v and likewise for w.
You can check if this matches Lambdageek's answer (modified to square the terms) as follows
outputLG = mean ((bsxfun(#times, A, reshape(v, 1, 1, [])) ...
+ bsxfun(#times, B, reshape(w, 1, 1, []))).^2, 3);
isequal(output,outputLG)
ans =
1