Deron Eriksson created SYSTEMML-707:
---------------------------------------
Summary: diag not generate square matrix if given Nx1 matrix of
zeroes
Key: SYSTEMML-707
URL: https://issues.apache.org/jira/browse/SYSTEMML-707
Project: SystemML
Issue Type: Bug
Components: APIs
Reporter: Deron Eriksson
Thank you Matthew Plourde for finding this!
If an Nx1 matrix of 0's is given to the diag() function, an Nx1 matrix of 0's
is returned. However, if an Nx1 matrix consists of any values that aren't 0's,
an NxN diagonal matrix is returned. This is inconsistent and the Nx1 matrix of
0's to diag() should probably return an NxN matrix.
Example 1:
{code}
zeroes=matrix(0, 5, 1);
print(toString(zeroes));
print(toString(diag(zeroes)));
{code}
gives:
{code}
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
{code}
Example 2:
{code}
ones=matrix(1, 5, 1);
print(toString(ones));
print(toString(diag(ones)));
{code}
gives:
{code}
1.000
1.000
1.000
1.000
1.000
1.000 0.000 0.000 0.000 0.000
0.000 1.000 0.000 0.000 0.000
0.000 0.000 1.000 0.000 0.000
0.000 0.000 0.000 1.000 0.000
0.000 0.000 0.000 0.000 1.000
{code}
Example 3:
{code}
nums=matrix("0 1 2 3 4", 5, 1);
print(toString(nums));
print(toString(diag(nums)));
{code}
gives:
{code}
0.000
1.000
2.000
3.000
4.000
0.000 0.000 0.000 0.000 0.000
0.000 1.000 0.000 0.000 0.000
0.000 0.000 2.000 0.000 0.000
0.000 0.000 0.000 3.000 0.000
0.000 0.000 0.000 0.000 4.000
{code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
