OpenOF: Framework for Sparse Non-linear Least Squares Optimization on a GPU

Version 0.1.1

Hardware requirements:

• Nvidia GPU with compute capability >= 2.0

Software requirements:

The framework was tested using Ubuntu 12.04, 64 bit and libraries with the following version number.

• CUDA 5

• Python 2.7

• sympy 0.7.1 or 0.7.3

  It is recommended to use this version as the sympy.cse function in version 0.7.2 does not work for some expression (http://www.mail-archive.com/[email protected]/msg15986.html).

• Thrust 1.6

• CUSP 0.3

• numpy 1.6.1

• swig 2.0.4

Cite:

If you use OpenOF for your research please cite:

Cornelius Wefelscheid and Olaf Hellwich OpenOF: Framework for sparse non-linear least squares optimization on a gpu. VISAPP 2013 [pdf]

License:

OpenOF - Open Optimization Framework Copyright (C) 2012 C. Wefelscheid

OpenOF is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

OpenOF is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with OpenOF. If not, see http://www.gnu.org/licenses/.

Getting started:

OpenOF comes with two examples, similarity transformation and bundle adjustement. To get started we recommend looking into the simpler model of the similarity transformation. To get the similarity transformation example running you need to execute the following commands.

• Move to OpenOF folder

---

$ cd Example

$ python similarityTransform.py

$ cd ..

$ cmake .

$ make

---

A C++ library as well as a Python library was created.

An example how to use the Python library can be found under ./Example. To run the Example proceed as follows:

---

$ cd Example

$ python runSimilarityTransformObj.py

---

or

---

$ python runSimilarityTransform.py

---

A more complex example is under Example/bundle.py (design) and Example/runBundle.py (runtime).

Trouble Shooting:

• In case you have version 0.7.2 of sympy you can downgrade with the following command

---

$ sudo pip install sympy==0.7.1

---

or upgrade to version 0.7.3.

• Cuda 5 does not come with Thrust 1.6, please exchange the thrust version usually under /usr/local/cuda/include/thrust.

• Cuda 5 is not working with gcc version 4.7 or higher. Try using an older gcc version.

• Compiler error: /bin/sh: 1: Syntax error: "(" unexpected.

  Make sure the path of the library doesn't contain any brackets.