I am reading some notes about tensor product of vector spaces (those in here) where the following sentence can be found:

Note that not every matrix on V ⊗ W can be written as a tensor product of a matrix on V and another on W.

Matrices in V ⊗ W are defined as a Kronecker product. So, how could this claim be possible? Has it to do with the problem of splitting the Kronecker product of two matrices?

More generally: does the general picture changes if we consider V and W infinite-dimensional vector spaces?


Two elements:

  1. If you read the sentence following the one you quote, there is an explanation based on dimensions.
  2. Consider a $4 \times 4$ matrix $C$ with only one zero entry. If you try to write $C = A \otimes B$, either none of the $A,B$ entries is equal to zero and then $C$ will have no zero entry. Or they have at least one, and then $C$ will have at least four zero entries.

The second argument can be extended for infinite dimensions.

| cite | improve this answer | |
  • $\begingroup$ Thank you for your clarity. Does this fact mean that not every basis of V⊗W is of the form of a tensor product of a basis of V and a basis of W? $\endgroup$ – gibarian Jun 2 at 13:33
  • $\begingroup$ You can refer here for some elements. $\endgroup$ – mathcounterexamples.net Jun 2 at 13:38

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.