# [Numpy-tickets] [NumPy] #434: Incomplete docstring linalg.eigh

NumPy numpy-tickets at scipy.net
Wed Jan 24 09:01:36 CST 2007

```#434: Incomplete docstring linalg.eigh
--------------------------+-------------------------------------------------
Reporter:  nils          |        Owner:  somebody
Type:  enhancement   |       Status:  new
Priority:  low           |    Milestone:
Component:  numpy.linalg  |      Version:
Severity:  trivial       |   Resolution:
Keywords:                |
--------------------------+-------------------------------------------------
Comment (by nils):

I mean the difference between the docstrings numpy.linalg.eigh and
scipy.linalg.eigh.
Sorry for the confusion.

{{{
Help on function eigh in module numpy.linalg.linalg:

eigh(a, UPLO='L')
Compute eigenvalues for a Hermitian-symmetric matrix.
}}}

versus

{{{
Help on function eigh in module scipy.linalg.decomp:

eigh(a, lower=True, eigvals_only=False, overwrite_a=False)
Solve real symmetric or complex hermitian eigenvalue problem.

Inputs:

a            -- A hermitian N x N matrix.
lower        -- values in a are read from lower triangle
[True: UPLO='L' (default) / False: UPLO='U']
eigvals_only -- don't compute eigenvectors.
overwrite_a  -- content of a may be destroyed

Outputs:

For eigvals_only == False (the default),
w,v     -- w: eigenvalues, v: eigenvectors
For eigvals_only == True,
w       -- eigenvalues

Definitions:

a * v[:,i] = w[i] * vr[:,i]
v.H * v = identity
}}}

--
Ticket URL: <http://projects.scipy.org/scipy/numpy/ticket/434#comment:2>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.
```