6
$\begingroup$

There is list containing some $(i,j)$ values of indices. Now i want to create a matrix of dimension n $\times$ m whose entries are 1 for the indices corresponding to the given list otherwise its zero.

For e.g if the list is as given below

list = {{1, 1}, {1, 2}, {2, 1}, {3, 3}};

And if $n=m=3$ then the output should be 

 {{1, 1, 0}, {1, 0, 0}, {0, 0, 1}}

I was able to do this as shown below but i believe this could be achieved with a shorter and a more elegant code. Maybe a one liner.

      cHeck[n1_, n2_] := Module[{flag = 0, list, i},
                         list = {{1, 1}, {1, 2}, {2, 1}, {3, 3}};
                         For[i = 1, i <= Length[list], i++,
                           Which[{n1, n2} === list[[i]], flag = 1]
                             ];
                        flag
                            ]
 A[n_, m_] := Table[cHeck[i, j], {i, 1, n}, {j, 1, m}]

 A[3, 3] 
 (*{{1, 1, 0}, {1, 0, 0}, {0, 0, 1}}*)
Improve this question
$\endgroup$
5
  • 5
    $\begingroup$ Try SparseArray[# -> 1 & /@ list, {n, m}, 0]. $\endgroup$ – b.gates.you.know.what Feb 7 '14 at 14:49
  • $\begingroup$ wow that worked. i knew there would be a one liner. Thanks. $\endgroup$ – Hubble07 Feb 7 '14 at 15:05
  • $\begingroup$ @rm-rf I disagree with closing this question as a duplicate of the one you chose. That one is considerably more complex and obscures the simplicity if what is needed here. I searched for a duplicate myself and found many examples with Band, as well as many uses of SparseArray as part of a larger answer, but I did not fine one that I felt was a duplicate. $\endgroup$ – Mr.Wizard Feb 7 '14 at 15:59
  • $\begingroup$ @Mr.Wizard Isn't the solution the same as your f1? Just with list -> 1 instead of individually setting them? Feel free to reopen it if you feel it isn't addressed by the other one. $\endgroup$ – rm -rf♦ Feb 7 '14 at 16:34
  • $\begingroup$ @rm-rf The syntax is different; the answers to this question show that both ReplacePart and SparseArray can accept the form positionlist -> value without the need for mapping or threading. The first comment above shows that people are not all aware of this. Closer are some of the answers I found using Band. I am going to reopen this for now; if you come across another Q&A that is not overly involved that clearly shows the syntax I describe please close it again. $\endgroup$ – Mr.Wizard Feb 7 '14 at 16:42
8
$\begingroup$

You can use SparseArray:

list = {{1, 1}, {1, 2}, {2, 1}, {3, 3}};

n = m = 3;

SparseArray[list -> 1, {n, m}] // MatrixForm

1 1 0
1 0 0
0 0 1

Timing compared to ReplacePart as proposed by bill s:

list = DeleteDuplicates @ RandomInteger[{1, 1000}, {75000, 2}];
n = m = 1000;

ReplacePart[ConstantArray[0, {n, m}], list -> 1] // AbsoluteTiming // First
SparseArray[list -> 1, {n, m}]                   // AbsoluteTiming // First

0.170

0.018

The SparseArray object can be converted to a standard array using Normal in negligible time.

Improve this answer
$\endgroup$
4
$\begingroup$

If the matrix is already defined, you can use ReplacePart:

list = {{1, 1}, {1, 2}, {2, 1}, {3, 3}};
q = ConstantArray[0, {3, 3}];

ReplacePart[q, list -> 1]
{{1, 1, 0}, {1, 0, 0}, {0, 0, 1}}

Of course, this is more lines than the SparseArray solution.

Improve this answer
$\endgroup$
1
  • $\begingroup$ This is faster than I remembered it being. +1. However it's still not nearly as fast as SparseArray, at least in v7. $\endgroup$ – Mr.Wizard Feb 7 '14 at 15:37
0
$\begingroup$

Also: MapAt

list = {{1, 1}, {1, 2}, {2, 1}, {3, 3}};
q = ConstantArray[0, {3, 3}];

MapAt[1 &, q, list]

{{1, 1, 0}, {1, 0, 0}, {0, 0, 1}}

Note: this is much slower than both SparseArray and ReplacePart.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.