OOF2 version 2.0.4 is now available! See the Change Log to learn what's new.

OOF2 version 2.0.4 can now be run on the nanoHUB. See the announcement for more details.

OOF2 version 2.0.5a6 is now available as well. It fixes some bugs, and includes preliminary versions of some new features. See the Change Log to learn what's new. It is an alpha release, meaning that the new features may yet change and have not been thoroughly tested. You should probably use it anyway, because it contains important bug fixes.

OOF2 retains (almost) all of the features of OOF1, although it does not read OOF1 data files. The latest versions of OOF1, however, can write OOF2 data files.

OOF2 is based on a new set of C++ classes for finite elements and material properties, tied together in a Python infrastructure. Python is an easy to use, high-level, object-oriented scripting language.

• OOF2 is much more flexible and expandable than OOF1. OOF2 can potentially handle any problem of the form:
  • "Flux = Modulus times gradient of Field"
  and
  • "divergence of Flux = Applied Force".
  Currently OOF2 can solve the heat equation, mechanical force balance, and the Coulomb equation. It includes material properties for linear elasticity, body forces (gravity), thermal conductivity, heat sources, dielectric permittivity, space charge (in an approximation in which charges interact only with the local polarization field, not with other charges at a distance), thermal expansion, and piezoelectricity.
• New Fields and Fluxes can be added with only a few lines of Python code. New material properties can be added with a few lines of Python code, or, if speed is an issue, in C++. (Defining new Properties is not quite as simple as defining new Fields, but is much simpler than the corresponding task in OOF1.) Instructions for adding features to OOF2 are given in the manual.
• Materials are built from a collection of Properties. Any combination of Properties is allowed, with reasonable constraints on completeness and unambiguity.
• OOF2 contains a more powerful set of finite elements than does OOF1. OOF2 has 3 noded triangles, 4 node quadrilaterals, 6 noded subparametric triangles, and 8 noded subparametric quadrilaterals. Adding new element types in C++ is easy.
• OOF2 generates and refines triangular, quadrilateral, and mixed meshes from image data. Element order is specified independently from element geometry.
• OOF2 incorporates nonlinear solvers. (But not many non-linear material properties are yet present.)
• OOF2 can refine meshes adaptively using a-posteriori error estimators.
• OOF2 is threaded, meaning that it can perform multiple calculations simultaneously, unlike OOF1. OOF2 is not yet fully parallel -- it doesn't yet use multiple processors to perform a single calculation.
• OOF2 is completely scriptable in Python, and can also be run interactively from a graphical user interface.
• OOF2 has more flexible graphical output than OOF1.
• OOF2 can export mesh geometry directly into abaqus input files.
• OOF2 (version 2.0.4 or later) can read EBSD orientation map data files. See the OOF2 Orientation Mapping page for more information.

• Expanding the scope of the problems solved to include
  • "Flux = Constitutive Relation Function of Field"
  • "divergence of Flux = lambda * Flux"
  • "divergence of Flux = gamma * time derivative of Field"
• Non-linear physics, including plasticity and possibly fracture.
• Time dependence, both for fields and material properties.
• Line elements, as well as elements with higher order continuity (for phase-field or strain-gradient models).
• Parallel execution.
• OOF2 will eventually become OOF3, which will solve 3 dimensional problems. The primary difficulties here will involve image based mesh generation and user interfaces. The finite element and material definition machinery will carry over from OOF2.

OOF2 will run on any computer running a variant of the Unix operating system, including Linux and Macintosh OS X.

In addition, OOF2 requires

• An X11 server. This is standard on most Unix systems. Macintosh users may have to install one, such as Apple's X11. Note that users of OS X 10.5 should install the latest version from Mac OS Forge instead.
• The Python scripting language, version 2.4 or 2.5. OOF2 might work with Python 2.6, and almost certainly does not yet work with Python 3.0.
• The Image Magick++ image processing library.
• The gtk+ graphics toolkit. OOF2 requires version 2.6 or later.
• The Gnome canvas library, libgnomecanvas2, version 2.6 or later. (Earlier versions may work too.)
• The pygtk Python bindings for the gtk+ library. OOF2 requires version 2.6 or later.
• The BLAS basic linear algebra subroutines and the lapack linear algebra library. These are provided with many computer systems. Check to see if you have a native version before downloading and installing the generic code from netlib. On Macintosh OS X, the blas routines are built in to the vecLib framework and do not have to be installed separately.

In addition, if you are going to be compiling OOF2 (which is the only way to get it at the moment), you will need

• a C++ compiler.

• the SWIG program for generating C++/Python interface code. OOF2 requires version 1.1 build 883.

NOTE: We're using an old version of swig because it was the stable version at the time we started working on OOF2. The current version will not work with OOF2. We will upgrade to a modern version of swig in a later release.

Please note the following Disclaimer and Copyright notice:

This software was produced by NIST, an agency of the U.S. government, and by statute is not subject to copyright in the United States. Recipients of this software assume all responsibilities associated with its operation, modification and maintenance. However, to facilitate maintenance we ask that before distributing modified versions of this software, you first contact the authors at oof_manager@ctcms.nist.gov.

• Download OOF2 version 2.0.4 (3336092 bytes, compressed).
• Download OOF2 version 2.0.5a6 (3471971 bytes, compressed).

• Generic installation instructions.
• Additional system-specific installation instructions.
• FAQ
• Change log. Learn what's new in the latest versions.
• Archive of old versions.

1. The downloaded file is a compressed tar archive. If your browser didn't unpack it for you, unpack it like this:
   % tar -xzf oof2-2.0.4.tar.gz
   
   This will create a directory called "oof2-2.0.4".
2. cd to that directory.
3. Read the README file and follow the directions in it. Additional system-specific installation notes are available here, and some frequently asked questions are answered in the FAQ.
4. If you like, run the test suites located in the TEST and TEST/GUI subdirectories of the OOF2 distribution. Read the README files in those directories for instructions.

After installing OOF2, you should have an executable file named oof2 in a bin directory in your execution path. You can now simply type oof2 at your shell prompt, and OOF2 will start up.

(Macintosh OS X users will have to start X11 before starting OOF2. Then OOF2 can be started in an xterm window, or in a Terminal window if the DISPLAY environment variable has been set correctly, probably to :0.0.)

OOF also has many options, and you can get a summary of them by typing oof2 --help.

By default, OOF runs in graphics mode, opening a couple of windows to get you started. If you don't want this, you can use the --text option to run it in command-line mode.

Most importantly, explore the Tutorials in the Help menu. (The tutorials are only available in graphics mode.)

If you run into trouble, please check the FAQ.

The OOF2 Manual is available at http://www.ctcms.nist.gov/~langer/oof2man/index.html. You can also download the html files as a 8.1 MB gzipped tar file. Use tar -xzf oof2man.tgz to unpack it.

Please check the FAQ and the Known Problems list below before submitting bug reports. Send bug reports via e-mail to oof_bugs@ctcms.nist.gov. Include the following information with your report:

1. The type of computer and operating system that you're using.
2. The version of OOF2 that you're using. Starting OOF2 with the --version flag will print the version number.
3. A complete description of the problem: what happened, and what did you do to make it happen?
4. If possible, an OOF2 script that reproduces the problem. Be sure to include any files that the script requires.

• Images drawn in the graphics window sometimes are drawn one screen pixel to the left and/or above where they should be. This appears to be due to round-off error within the gtk+ library. A work-around is to turn on antialiasing (in the graphics window's Settings menu), but this causes other problems.
• A number of operations need to be sped up.
• Some parts of the GUI test suite fail intermittently. We believe that this is due to timing and threading problems related to the test suite itself, and does not reflect errors in the OOF2 code. If a test fails, repeat it and see what happens. If it fails consistently, we'd like to know about it, but otherwise you can probably ignore it. Partially fixed in version 2.0.2.
• Since the introduction of subproblems in 2.0.2, and possibly prior to that, the Adaptive Mesh Refinement tool on the FEMesh page has caused segfaults. Partially fixed in version 2.0.4, fixed more in version 2.0.5a1.
• Adaptive Mesh Refinement is unreliable. Our algorithm is unstable, and results may be unreproducible.
• The "Save Image" menu item in the Graphics Window's "File" menu doesn't work if the window is displaying a Microstructure's Material colors. Fixed in version 2.0.5a1.
• If a Field is defined and initialized on a Subproblem and then defined on a different Subproblem that shares Nodes with the first one, the Field will be reinitialized on those shared Nodes, destroying any calculated values. Fixed in version 2.0.5a1.
• Modifying a Skeleton, creating a Mesh, unmodifying the Skeleton, and then modifying one of its boundaries causes an error. Fixed in version 2.0.5a1.
• Mesh data files containing initialized out-of-plane fields cannot be loaded. Fixed in version 2.0.5a2.
• If the pixel size isn't 1x1 in physical units, round-off error can cause crashes in various operations (for example, creating contour plots of orientation-map dependent quantities). Fixed in version 2.0.5a2.
• OOF.Skeleton.Auto can create illegal elements because of a poor choice of parameters in SnapRefine, which can lead to a divide-by-zero error in Rationalize. Fixed in version 2.0.5a2.

• Intersecting floating boundary conditions with continuum profiles don't work. Fixed in version 2.0.3.
• Switching Skeleton modifiers can sometimes cause an RWLock error. Fixed in version 2.0.2.
• The "File/Save/Python Log" dialog box's "Save" button is sometimes incorrectly disabled. A workaround is to switch directories within the dialog, and then switch back, if necessary. Fixed in version 2.0.2.
• The Nodes on the top and/or right edges of a Skeleton sometimes move off of the edge when the Skeleton is modified. Fixed in version 2.0.2.
• The stress-free strain property and the thermal expansion property with non-zero T0 do not work correctly in plane-stress. Fixed in 2.0.2.
• Material dependent output quantities (e.g, stress) are not computed correctly in some cases if a mesh is loaded from a data file. A workaround is to solve the mesh, even if the saved state has already been equilibrated. Fixed in 2.0.2.
• In versions 2.0.2 and 2.0.3, OOF lets you use the CG solver even for non-symmetric problems. The workaround is to choose another solver when using thermal expansion with an active temperature field. Fixed in version 2.0.4.
• OOF dumps core when quitting on some systems, including NetBSD. This is harmless (you were quitting anyway, right?), but annoying. Fixed in version 2.0.4b2.

Send comments to the OOF team.