commonsense team mailing list archive

[Ken Arnold <kenneth.arnold@xxxxxxxxx>]

** Summary changed:

- Why the heck do we have to use a : in indexes? 
+ Tensors should behave more like NumPy matrices

** Description changed:

- Why the heck do we have to use a : in indexes, anyways?
+ You should be able to get row k of a matrix with matrix[k].
  
- You should be able to get row k of a matrix with matrix[k].
+ You should be able to iterate rows with `for row in matrix` or perhaps
+ matrix.rows().
+ 
+ All of NumPy's broadcasting stuff should work.
+ 
+ In short, a Tensor should look like a NumPy matrix, except that
+ .iteritems() should return key-value pairs and sparse matrices can grow
+ in size. (Plus all the convenient .normalized, .svd, etc.)

** Changed in: divisi
       Status: New => Confirmed

** Changed in: divisi
   Importance: Undecided => Medium

-- 
Tensors should behave more like NumPy matrices
https://bugs.launchpad.net/bugs/413182
You received this bug notification because you are a member of
Commonsense Computing, which is the registrant for Divisi.

Status in Divisi: Confirmed

Bug description:
You should be able to get row k of a matrix with matrix[k].

You should be able to iterate rows with `for row in matrix` or perhaps matrix.rows().

All of NumPy's broadcasting stuff should work.

In short, a Tensor should look like a NumPy matrix, except that .iteritems() should return key-value pairs and sparse matrices can grow in size. (Plus all the convenient .normalized, .svd, etc.)