ColPack's Doxygen documentation is available here: http://cscapes.cs.purdue.edu/coloringpage/software.htm

ColPack's project home page: http://cscapes.cs.purdue.edu/coloringpage/

ColPack
Installation Guilds
2.1 Compile ColPack Without Install
2.2 Ubuntu Install
2.3 Windows Install
2.4 MacOS Install
2.5 Utilize the Installed Library
Usages
HowToCite

ColPack

ColPack is a package comprising of implementations of algorithms for the specialized vertex coloring problems discussed in the previous section as well as algorithms for a variety of related supporting tasks in derivative computation.

Coloring capabilities

the table below gives a quick summary of all the coloring problems (on general and bipartite graphs) supported by ColPack.

General Graph Coloring	Bipartite Graph one-sided coloring	Bipartite Graph Bicoloring
Distance 1 coloring	Partial distance-2 coloring	Star bicoloring
Distance 2 coloring	Partial distance-2 coloring
Star coloring
Acyclic coloring
Restricted star coloring
Triangular coloring

All of the coloring problems listed in the above table are NP-hard. Their corresponding algorithms in ColPack are greedy heuristics in the sense that the algorithms progressively extend a partial coloring by processing one vertex at a time, in some order, in each step assigning a vertex the smallest allowable color. Listed beneath each coloring problem in the table is the complexity of the corresponding algorithm in ColPack. In the cases where ColPack has multiple algorithms for a problem (these are designated by the superscript †), the complexity expression corresponds to that of the fastest algorithm. In the complexity expressions,

the complexity of the corresponding algorithm can be found here ColPack's project

Ordering techniques

The order in which vertices are processed in a greedy coloring algorithm determines the number of colors used by the algorithm. ColPack has implementations of various effective ordering techniques for each of the supported coloring problems. These are summarized in the table below.

General Graph Coloring	Bipartite Graph one-sided coloring	Bipartite Graph Bicoloring
Natural	Column Natural	Natural
Largest First	Column Largest First	Largest First
Smallest Last	Column Smallest Last	Smallest Last
Incidence Degree	Column Incidence Degree	Incidence Degree
Dynamic Largest First	Row Natural	Dynamic Largest First
Distance-2 Largest First	Row Largest First	Selective Largest First
Distance-2 Smallest Last	Row Smallest Last	Selective Smallest Last
Distance-2 Incidence Degree	Row Incidence Degree	Selective Incidence Degree
Distance-2 Dynamic Largest First

Recovery routines

Besides coloring and ordering capabilities, ColPack also has routines for recovering the numerical values of the entries of a derivative matrix from a compressed representation. In particular the following reconstruction routines are currently available:

Recovery routines for direct (via star coloring ) and substitution-based (via acyclic coloring) Hessian computation
Recovery routines for unidirectional, direct Jacobian computation (via column-wise or row-wise distance-2 coloring)
Recovery routines for bidirectional, direct Jacobian computation via star bicoloring

Graph construction routines

Finally, as a supporting functionality, ColPack has routines for constructing bipartite graphs (for Jacobians) and adjacency graphs (for Hessians) from files specifying matrix sparsity structures in various formats, including Matrix Market, Harwell-Boeing and MeTis.

ColPack : organization

ColPack is written in an object-oriented fashion in C++ heavily using the Standard Template Library (STL). It is designed to be simple, modular, extensible and efficient. Figure 1 below gives an overview of the structure of the major classes of ColPack.

Build and Compile ColPack Instructions

There are two ways to use ColPack, Try without Installiation and Build and Install. The former is fast and easy to use, but is vulnerable for various OS enviroments settings, thus it requires the user know how to modify the makefile if met some compiling issue. The later one is more robust and it will also collect the ColPack into a shared library which makes ColPack easy to cooperate with other applications. But it requires to pre-install automake(or CMake) software.

Try ColPack by Compile and Run without Installation

You can just try ColPack by download, compile and run it. This is the fastest and simplest way to use ColPack. Do the following instructions in terminals.

cd              
git clone https://github.com/CSCsw/ColPack.git   #Download ColPack
cd ColPack      # go to ColPack Root Directory
cd Example_Try  # go to Try ColPack folder
make            # compile the code

After all source codes been compiled, we will generate a executable file ColPack under current folder.
The above instruction are tested under Ubuntu system. You may need to modify the Makefile to fit the different OS environments and compilers.(delete -fopenmp for mac os. Replace -fopenmp to -Qopenmp )for intel compiler.)

Ubuntu Build Instructions

Install ColPack makes ColPack easy to use and it can also decreases the size of the execuable file. To install ColPack using autotools (requires that have installed automake on your machine.), follows the instructions below.:

cd   
git clone https://github.com/CSCsw/ColPack.git  #Download ColPack
cd ColPack             # ColPack Root Directory
autoreconf -vif                                
fullpath=$(pwd)        # modify fullpath to your destination folder if need
./configure --prefix=${fullpath}  
make -j 4              # Where "4" is the number of cores on your machine
make install           # install lib and include/ColPack to destination

Append --disable-openmp to ./configure above if you need to disable OpenMP.(MAC user and some Windows user)

ColPack also has experimental support for building with CMake, which you can do via the following:

mkdir build
cd build
cmake ..
make -j 4   #Where "4" is the number of cores on your machine
ctest       #Run the examples in SampleDrivers/Basic folder
make install

Use cmake -LH . or ccmake . in the build directory to see a list of options, such as ENABLE_EXAMPLES and ENABLE_OPENMP, which you can set by running the following from the build directory:

cmake .. -DENABLE_OPENMP=ON

Windows Build Instructions

You can build ColPack's static library on Windows using Visual Studio (tested with Visual Studio 2015) and CMake. Note, however, that you are not able to use OpenMP (Visual Studio supports only OpenMP 2.0), and cannot compile the ColPack executable (it depends on the POSIX getopt.h).

If you are using CMake 3.4 or greater, you can build and use ColPack's shared library. If you have an older CMake, we still build the shared library, but you will not be able to use it because none of the symbols will be exported (Visual Studio will not generate a .lib file).

On Windows, the examples link to the static library instead of the shared library.

Unlike on UNIX, the static library is named ColPack_static (ColPack_static.lib) to avoid a name conflict with the shared library's ColPack.lib.

Finally, some of the examples do not compile, seemingly because their filenames are too long.

MAC OS Build Instructions

To install ColPack on Mac, you first need to install Apple Xcode and automake. Since (it is well known that) Mac's default compiler clang doesn't support OpenMP well, you need either install OpenMP and gcc compiler or disable OpenMP by --disable-openmp .(It's a well known problem, MAC's default compiler clang doesn't support OpenMP well.)

cd   
git clone https://github.com/CSCsw/ColPack.git  #Download ColPack
cd ColPack             # ColPack Root Directory
autoreconf -vif                                
fullpath=$(pwd)        # modify fullpath to your destination folder if need
./configure --prefix=${fullpath} --disable-openmp
make -j 4              # Where "4" is the number of cores on your machine
make install           # install lib and include/ColPack to destination

Another recommend altinative way is to install an Ubuntu system on your MAC with VirtualBox (or any other virtual machine software), then install ColPack on your virtual machines.

After the Build, Use ColPack as Installed Library

After the build, we have already generate an executable file 'ColPack' under the colpack root directory. And you can use it. However if you want to write your own code and use ColPack as an shared library. Then follow the following ways:

export library's path to LD_LIBRARY_PATH
create your own code.
include the relative ColPack header files within your code.
added -ldl path/to/installed/library and -I /path/to/installed/include to the compiler
compile the code

We provide a template codes in Example_Use_Library

USAGE

After building (or try), you can run the following commands (from ColPack root directory if using autotools, or from the build directory if using CMake):

$./ColPack -f <graph_file_name> -o <ordering> -m <methods> [-v]

DISPLAY HELP

$./ColPack

OPTIONs

<gfile_name>:  Input file name
<ordering>  :  LARGEST_FIRST
               SMALLEST_LAST,
               DYNAMIC_LARGEST_FIRST,
               INCIDENCE_DEGREE,
               NATURAL,
               RANDOM
<methods>   :  DISTANCE_ONE
               ACYCLIC
               ACYCLIC_FOR_INDIRECT_RECOVERY
               STAR
               RESTRICTED_STAR
               DISTANCE_TWO
               --------------------
               DISTANCE_ONE_OMP    (automatic display: nThreads,num_colors,timall,conflicts,loops)
               --------------------
               IMPLICIT_COVERING__STAR_BICOLORING
               EXPLICIT_COVERING__STAR_BICOLORING
               EXPLICIT_COVERING__MODIFIED_STAR_BICOLORING
               IMPLICIT_COVERING__GREEDY_STAR_BICOLORING
               --------------------
               COLUMN_PARTIAL_DISTANCE_TWO
               ROW_PARTIAL_DISTANCE_TWO

-v          :  verbose for debug infomation

EXAMPLES:

./ColPack -f Graphs/bcsstk01.mtx -o LARGEST_FIRST -m DISTANCE_ONE -v
./ColPack -f Graphs/bcsstk01.mtx -o SMALLEST_LAST -m ACYCLIC -v
./ColPack -f Graphs/bcsstk01.mtx -o DYNAMIC_LARGEST_FIRST -m DISTANCE_ONE_OMP -v

EXAMPLE OUTPUT

ReadMatrixMarketAdjacencyGraph
Found file Graphs/bcsstk01.mtx
Graph of Market Market type: [matrix coordinate real symmetric]
		Graph structure and VALUES will be read

#DISTANCE_ONE Result: 
6  : (NATURAL)
6  : (LARGEST_FIRST)
6  : (DYNAMIC_LARGEST_FIRST)
6  : (SMALLEST_LAST)
6  : (INCIDENCE_DEGREE)
6  : (RANDOM)

#ACYCLIC Result: 
8  : (NATURAL)
8  : (LARGEST_FIRST)
8  : (DYNAMIC_LARGEST_FIRST)
8  : (SMALLEST_LAST)
8  : (INCIDENCE_DEGREE)
8  : (RANDOM)

#ACYCLIC_FOR_INDIRECT_RECOVERY Result: 
8  : (NATURAL)
8  : (LARGEST_FIRST)
8  : (DYNAMIC_LARGEST_FIRST)
8  : (SMALLEST_LAST)
8  : (INCIDENCE_DEGREE)
8  : (RANDOM)

#STAR Result: 
12  : (NATURAL)
12  : (LARGEST_FIRST)
12  : (DYNAMIC_LARGEST_FIRST)
12  : (SMALLEST_LAST)
12  : (INCIDENCE_DEGREE)
12  : (RANDOM)

#RESTRICTED_STAR Result: 
15  : (NATURAL)
15  : (LARGEST_FIRST)
15  : (DYNAMIC_LARGEST_FIRST)
15  : (SMALLEST_LAST)
15  : (INCIDENCE_DEGREE)
15  : (RANDOM)

#DISTANCE_TWO Result: 
15  : (NATURAL)
15  : (LARGEST_FIRST)
15  : (DYNAMIC_LARGEST_FIRST)
15  : (SMALLEST_LAST)
15  : (INCIDENCE_DEGREE)
15  : (RANDOM)

The best source for citing this work

Assefaw H. Gebremedhin, Duc Nguyen, Mostofa Ali Patwary, and Alex Pothen, ColPack: Graph coloring software for derivative computation and beyond, ACM Transactions on Mathematical Software, 40 (1), 30 pp., 2013.

barak / colpack

Table of Contents