Skip to content

Add missing trtrs, orgqr, ormqr docs #3720

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
soumith merged 2 commits into from
Nov 15, 2017
Merged

Add missing trtrs, orgqr, ormqr docs #3720

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
288cac5
trtrs docs
zou3519 Nov 15, 2017
1077353
orgqr and ormqr docs
zou3519 Nov 15, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
105 changes: 94 additions & 11 deletions torch/_torch_docs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3140,14 +3140,39 @@
[torch.FloatTensor of size 2x3]
""")

# TODO
# add_docstr(torch._C.orgqr,
# """
# """)
add_docstr(torch._C.orgqr,
"""
orgqr(a, tau) -> Tensor

Computes the orthogal matrix `Q` of a QR factorization, from the `(a, tau)` tuple
returned by :func:`torch.geqrf`.

This directly calls the underlying LAPACK function `?orgqr`.
See `?orgqr LAPACK documentation`_ for further details.

Args:
a (Tensor): The `a` from :func:`torch.geqrf`.
tau (Tensor): The `tau` from `torch.geqrf`.

.. _?orgqr LAPACK documentation:
https://software.intel.com/en-us/mkl-developer-reference-c-orgqr

""")

add_docstr(torch._C.ormqr,
"""
ormqr(a, tau, mat, left=True, transpose=False) -> (Tensor, Tensor)

# add_docstr(torch._C.ormqr,
# """
# """)
Multiplies `mat` by the orthogonal `Q` matrix of the QR factorization
formed by :func:`torch.geqrf` that is represented by `(a, tau)`.

This directly calls the underlying LAPACK function `?ormqr`.
See `?ormqr LAPACK documentation`_ for further details.

.. _?ormqr LAPACK documentation:
https://software.intel.com/en-us/mkl-developer-reference-c-ormqr

""")

add_docstr(torch._C.potrf,
"""
Expand Down Expand Up @@ -4723,10 +4748,68 @@

""")

# TODO
# add_docstr(torch._C.trtrs,
# """
# """)
add_docstr(torch._C.trtrs,
"""
trtrs(b, A, upper=True, transpose=False, unitriangular=False) -> (Tensor, Tensor)

Solves a system of equations with a triangular coefficient matrix `A`
and multiple right-hand sides `b`.

In particular, solves :math:`AX = b` and assumes `A` is upper-triangular
with the default keyword arguments.

This method is NOT implemented for CUDA tensors.

Args:
A (Tensor): the input triangular coefficient matrix.
b (Tensor): multiple right-hand sides. Each column of `b` is a
right-hand side for the system of equations.
upper (bool, optional): Solves the upper-triangular system
of equations if True, lower-triangular if False. Default: True.
transpose (bool, optional): If `A` should be transposed before
being sent into the solver. Default: False.
unitriangular (bool, optional): If `A` is unit triangular.
If True, the diagonal elements of `A` are assumed to be
1 and not referenced from `A`. Default: False.

Returns:
A tuple (X, M) where `M` is a clone of `A` and `X` is the solution to
`AX = b` (or whatever variant of the system of equations, depending on
the keyword arguments.)

Shape:
- A: :math:`(N, N)`
- b: :math:`(N, C)`
- output[0]: :math:`(N, C)`
- output[1]: :math:`(N, N)`

Examples::

>>> A = torch.randn(2,2).triu()
>>> A

-1.8793 0.1567
0.0000 -2.1972
[torch.FloatTensor of size 2x2]

>>> b = torch.randn(2,3)
>>> b

1.8776 -0.0759 1.6590
-0.5676 0.4771 0.7477
[torch.FloatTensor of size 2x3]

>>> torch.trtrs(b, A)

(
-0.9775 0.0223 -0.9112
0.2583 -0.2172 -0.3403
[torch.FloatTensor of size 2x3],
-1.8793 0.1567
0.0000 -2.1972
[torch.FloatTensor of size 2x2])

""")

add_docstr(torch._C.trunc,
"""
Expand Down