# Difference between revisions of "SchmidtDecomposition"

 Other toolboxes required SchmidtDecomposition Computes the Schmidt decomposition of a bipartite vector none IsProductVectorOperatorSchmidtDecompositionSchmidtRank Entanglement and separability

SchmidtDecomposition is a function that computes the Schmidt decomposition of a bipartite vector. The user may specify how many terms in the Schmidt decomposition they wish to be returned.

## Syntax

• S = SchmidtDecomposition(VEC)
• S = SchmidtDecomposition(VEC,DIM)
• S = SchmidtDecomposition(VEC,DIM,K)
• [S,U,V] = SchmidtDecomposition(VEC,DIM,K)

## Argument descriptions

### Input arguments

• VEC: A bipartite vector (e.g., a pure quantum state) to have its Schmidt decomposition computed.
• DIM (optional, by default has both subsystems of equal dimension): A 1-by-2 vector containing the dimensions of the subsystems that VEC lives on.
• K (optional, default 0): A flag that determines how many terms in the Schmidt decomposition should be computed. If K = 0 then all terms with non-zero Schmidt coefficients are computed. If K = -1 then all terms (including zero Schmidt coefficients) are computed. If K > 0 then the K terms with largest Schmidt coefficients are computed.

### Output arguments

• S: A vector containing the Schmidt coefficients of VEC.
• U (optional): A matrix whose columns are the left Schmidt vectors of VEC.
• V (optional): A matrix whose columns are the right Schmidt vectors of VEC.

## Examples

The following code returns the Schmidt decomposition of the standard maximally-entangled pure state $\frac{1}{\sqrt{d}}\sum_j|j\rangle\otimes|j\rangle \in \mathbb{C}^d \otimes \mathbb{C}^d$ in the $d = 3$ case:

>> [s,u,v] = SchmidtDecomposition(MaxEntangled(3))

s =

0.5774
0.5774
0.5774

u =

1     0     0
0     1     0
0     0     1

v =

1     0     0
0     1     0
0     0     1

## Source code

Click on "expand" to the right to view the MATLAB source code for this function.

1. %%  SCHMIDTDECOMPOSITION   Computes the Schmidt decomposition of a bipartite vector
2. %   This function has one required argument:
3. %     VEC: a bipartite vector to have its Schmidt decomposition computed
4. %
5. %   S = SchmidtDecomposition(VEC) is a vector containing the non-zero
6. %   Schmidt coefficients of the bipartite vector VEC, where the two
7. %   subsystems are each of size sqrt(length(VEC)).
8. %
9. %   This function has two optional input arguments:
10. %     DIM (default [sqrt(length(VEC)),sqrt(length(VEC))])
11. %     K (default 0)
12. %
13. %   [S,U,V] = SchmidtDecomposition(VEC,DIM,K) gives the Schmidt
14. %   coefficients S of the vector VEC and the corresponding left and right
15. %   Schmidt vectors in the matrices U and V. DIM is a 1x2 vector containing
16. %   the dimensions of the subsystems that VEC lives on. K is a flag that
17. %   determines how many terms in the Schmidt decomposition should be
18. %   computed. If K = 0 then all terms with non-zero Schmidt coefficients
19. %   are computed. If K = -1 then all terms (including zero Schmidt
20. %   coefficients) are computed. If K > 0 then the K terms with largest
21. %   Schmidt coefficients are computed.
22. %
23. %   If DIM is a scalar instead of a vector, then it is assumed that the
24. %   first subsystem of size DIM and the second subsystem of size
25. %   length(VEC)/DIM.
26. %
27. %   URL: http://www.qetlab.com/SchmidtDecomposition
28. 
29. %   requires: opt_args.m
30. %   author: Nathaniel Johnston (nathaniel@njohnston.ca)
31. %   package: QETLAB
32. %   last updated: December 1, 2012
33. 
34. function [s,u,v] = SchmidtDecomposition(vec,varargin)
35. 
36. lv = length(vec);
37. 
38. % set optional argument defaults: dim=sqrt(length(vec)), k=0
39. [dim,k] = opt_args({ round(sqrt(lv)), 0 },varargin{:});
40. 
41. % allow the user to enter a single number for dim
42. if(length(dim) == 1)
43.     dim = [dim,lv/dim];
44.     if abs(dim(2) - round(dim(2))) >= 2*lv*eps
45.         error('SchmidtDecomposition:InvalidDim','The value of DIM must evenly divide length(VEC); please provide a DIM array containing the dimensions of the subsystems.');
46.     end
47.     dim(2) = round(dim(2));
48. end
49. 
50. % Try to guess whether svd or svds will be faster, and then perform the
51. % appropriate singular value decomposition.
52. adj = 20 + 1000*(~issparse(vec));
53. 
54. if(k > 0 && k <= ceil(min(dim)/adj)) % just a few Schmidt coefficients
55.     [v,s,u] = svds(reshape(vec,dim(end:-1:1)),k);
56. else % lots of Schmidt coefficients
57.     [v,s,u] = svd(reshape(full(vec),dim(end:-1:1)));
58.     if(k > 0)
59.         v = v(:,1:k);
60.         s = s(:,1:k);
61.         u = u(:,1:k);
62.     end
63. end
64. s = diag(s);
65. if(k == 0)
66.     r = sum(s > max(dim) * eps(s(1)));  % Schmidt rank (use same tolerance as MATLAB's rank function)
67.     s = s(1:r);   % Schmidt coefficients
68.     u = u(:,1:r);
69.     v = v(:,1:r);
70. end
71. u = conj(u);