How to Export/Write a SparseArray?
How to Import/Read a SparseArray?
Is it always stored as a normal array?
How can it be stored in a dense form?
Is a numerical SparseArray different to
Export/Import from a string one?

  • 12
    $\begingroup$ What a nice poem $\endgroup$
    – Rojo
    May 14, 2013 at 9:27
  • $\begingroup$ You can use the methods described in answer to this question: mathematica.stackexchange.com/q/1959/121 In fact I am inclined to close this question as I believe it has already been answered. $\endgroup$
    – Mr.Wizard
    May 14, 2013 at 11:08

1 Answer 1


The two file formats I am familiar with, with respect to sparse matrices, are the Harwell-Boeing format, and the Matrix Market format. I have linked to the docs on how you can Import[]/Export[] matrices in those formats, and you'll do well by reading up on them.

Sparse arrays are very much different from lists. For instance, consider these two representations of the $9\times 9$ identity matrix:

idDense = IdentityMatrix[9];
idSparse = SparseArray[Band[{1, 1}] -> 1, {9, 9}];

For the purposes of Equal[], they're the same:

idSparse == idDense

SameQ[] shows that they aren't:

idSparse === idDense

One can peek at heads to see the difference:

Head /@ {idSparse, idDense}
   {SparseArray, List}

If, for some reason, you need to convert your sparse array to a dense array, that's what Normal[] is for:

Normal[idSparse] === idDense

See the docs for SparseArray[] and Normal[] for more details.

  • $\begingroup$ OK. . .What if you have txt char data instead? . .As in your example idSparse above replace the 1 by "A"? . .It is still a SparseArry. . .How to export this one? . .Can't see it in Harwell-Boeing or Matrix Market. $\endgroup$ May 15, 2013 at 14:04
  • $\begingroup$ I don't believe any of those formats will work for data with text entries. You will have to use Normal[] on your sparse array and then export to your favorite format. $\endgroup$ May 15, 2013 at 14:07

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