[SciPy-user] ANN: PyAMG v1.0 (Algebraic Multigrid Solvers in Python)

[Nathan Bell]

We are pleased to announce the first release of PyAMG: Algebraic
Multigrid Solvers in Python. With a user-friendly interface and
efficient implementation, PyAMG addresses the growing demand for
scalable solvers by non-experts. PyAMG features implementations of
* Ruge-Stuben (RS) or Classical AMG
* AMG based on Smoothed Aggregation (SA)
and experimental support for
* Adaptive Smoothed Aggregation (=E1SA)
* Compatible Relaxation (CR)
along with many tunable options for coarsening, interpolation,
relaxation, prolongator smoothing.


Our goal with the PyAMG project is to provide a framework for existing
AMG methods and to allow for quick testing and prototyping of
additional functionality and algorithms. Specifically, our objectives
are:
* ease of use
o interface is accessible to non-experts
o extensive documentation and references
* speed
o solves problems with millions of unknowns efficiently
o core multigrid algorithms are implemented in C++
o sparse matrix support is provided by scipy.sparse
* readability
o source code is organized into intuitive components
* extensibility
o core components can be reused to implement additional techniques
o new features are easy integrated
* experimentation
o facilitates rapid prototyping and analysis of multigrid methods
* portability
o tested on several platforms (Linux, Windows, Mac)
o relies only on Python, NumPy, and SciPy


PyAMG relies extensively on the NumPy and SciPy libraries for
scientific programming with Python. We thank the NumPy/SciPy
community for their support and continued efforts.


For more information see
http://www.pyamg.org


PyAMG developers:
Nathan Bell (http://graphics.cs.uiuc.edu/~wnbell/)
Luke Olson (http://www.cs.uiuc.edu/homes/lukeo/)
Jacob Schroder (www.cse.uiuc.edu/~jschrod3/)

[Robert Cimrman]

Hi Nathan,

thank you for PyAMG - it's great to have multigrid solvers in Python.

Best regards,
r.

Post by Nathan Bell
We are pleased to announce the first release of PyAMG: Algebraic
Multigrid Solvers in Python. With a user-friendly interface and
efficient implementation, PyAMG addresses the growing demand for
scalable solvers by non-experts. PyAMG features implementations of
* Ruge-Stuben (RS) or Classical AMG
* AMG based on Smoothed Aggregation (SA)
and experimental support for
* Adaptive Smoothed Aggregation (=E1SA)
* Compatible Relaxation (CR)
along with many tunable options for coarsening, interpolation,
relaxation, prolongator smoothing.
Our goal with the PyAMG project is to provide a framework for existing
AMG methods and to allow for quick testing and prototyping of
additional functionality and algorithms. Specifically, our objectives
* ease of use
o interface is accessible to non-experts
o extensive documentation and references
* speed
o solves problems with millions of unknowns efficiently
o core multigrid algorithms are implemented in C++
o sparse matrix support is provided by scipy.sparse
* readability
o source code is organized into intuitive components
* extensibility
o core components can be reused to implement additional techniques
o new features are easy integrated
* experimentation
o facilitates rapid prototyping and analysis of multigrid methods
* portability
o tested on several platforms (Linux, Windows, Mac)
o relies only on Python, NumPy, and SciPy
PyAMG relies extensively on the NumPy and SciPy libraries for
scientific programming with Python. We thank the NumPy/SciPy
community for their support and continued efforts.
For more information see
http://www.pyamg.org
Nathan Bell (http://graphics.cs.uiuc.edu/~wnbell/)
Luke Olson (http://www.cs.uiuc.edu/homes/lukeo/)
Jacob Schroder (www.cse.uiuc.edu/~jschrod3/)

[Fernando Perez]

Post by Robert Cimrman
Hi Nathan,
thank you for PyAMG - it's great to have multigrid solvers in Python.

Seconded, many thanks! This is great to have... We just finished the
SIAM CSE09 python panel (I'll post a little writeup soon and will get
the slides from the speakers), it would have been great to know you
were working on this to have you give a talk here :) Next time...

When you feel the time is right, it would be worth listing it in the
canonical 'yellow pages' for scipy:

http://www.scipy.org/Topical_Software

Cheers,

f

[Nathan Bell]

Post by Fernando Perez
Seconded, many thanks! This is great to have... We just finished the
SIAM CSE09 python panel (I'll post a little writeup soon and will get
the slides from the speakers), it would have been great to know you
were working on this to have you give a talk here :)  Next time...

Absolutely!

Post by Fernando Perez
When you feel the time is right, it would be worth listing it in the
http://www.scipy.org/Topical_Software

Done. For lack of a better category I put it under "Science: basic tools".

[Ondrej Certik]

Post by Fernando Perez

Post by Robert Cimrman
Hi Nathan,
thank you for PyAMG - it's great to have multigrid solvers in Python.

Seconded, many thanks! This is great to have... We just finished the

Thanks for this. Fernando told me at the conference that you released
it, but I haven't found time until now to read about it.

Post by Fernando Perez
SIAM CSE09 python panel (I'll post a little writeup soon and will get
the slides from the speakers), it would have been great to know you

Please post it, so that I can link to it from my blog. :)

Ondrej

[Robert Cimrman]

Post by Fernando Perez

Post by Robert Cimrman
Hi Nathan,
thank you for PyAMG - it's great to have multigrid solvers in Python.

Seconded, many thanks! This is great to have... We just finished the
SIAM CSE09 python panel (I'll post a little writeup soon and will get
the slides from the speakers), it would have been great to know you
were working on this to have you give a talk here :) Next time...
When you feel the time is right, it would be worth listing it in the
http://www.scipy.org/Topical_Software

Ah, Topical_Software page - can I add a finite element codes section? I
know some based on scipy :)

r.

[Fernando Perez]

Post by Robert Cimrman
Ah, Topical_Software page - can I add a finite element codes section? I
know some based on scipy :)

Of course! You don't need to ask for permission, it's a wiki :)

Cheers,

f

[Robert Cimrman]

Post by Fernando Perez

Post by Robert Cimrman
Ah, Topical_Software page - can I add a finite element codes section? I
know some based on scipy :)

Of course! You don't need to ask for permission, it's a wiki :)

But it is a very visible one. :)

I have added a new section (Partial differential equation (PDE) solvers)
and listed there FiPy and SfePy. I might have overlooked other PDE
solvers listed on the Topical Software page, so, please, fill yourselves in.

r.

[Ondrej Certik]

Post by Robert Cimrman

Post by Robert Cimrman
Ah, Topical_Software page - can I add a finite element codes section? I
know some based on scipy :)

Of course!  You don't need to ask for permission, it's a wiki :)

But it is a very visible one. :)
I have added a new section (Partial differential equation (PDE) solvers)
and listed there FiPy and SfePy. I might have overlooked other PDE
solvers listed on the Topical Software page, so, please, fill yourselves in.

I think our hermes hp-FEM solver counts there too, as it can be used
from Python and the sage notebook (at least the 2d one).

http://hpfem.org/

so I added it in.

Ondrej