Mercurial > hg > smallbox

comparison examples/SMALL_solver_test.m @ 1:7750624e0c73 version0.5

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
(none)
author idamnjanovic
date Thu, 05 Nov 2009 16:36:01 +0000
parents
children 2d7d8736ce6c
comparison
equal deleted inserted replaced
0:5181bee80bc1 1:7750624e0c73
1 function SMALL_solver_test
2 % Example test of solvers from different toolboxes on Sparco compressed
3 % sensing problems
4 %
5 % The main purpose of this example is to show how to use SMALL structure
6 % to solve SPARCO compressed sensing problems (1-11) and compare results
7 % from different solvers.
8 % To generate SMALL.Problem part of structure you can use generateProblem
9 % function from Sparco toolbox giving the problem number and any
10 % additional parameters you might want to change. Alternatively, you can
11 % might want to consult sparco documentation to write a problem by
12 % yourself. There are four fields the must be specified in SMALL.Problem
13 % - A, b, sizeA and reconstruct.
14 %
15 % To generate SMALL.solver part of the structure you must specify three
16 % fields:
17 %
18 % SMALL.solver.toolbox - string with toolbox name is needed because
19 % different toolboxes are calling solver
20 % functions in different ways.
21 % SMALL.solver.name - its string representing solver name (e.g.
22 % SolveOMP)
23 % SMALL.solver.param - string that contains optional parameters for
24 % particular solver (all parameters you want to
25 % specify except A, b and size of solution)
26 %
27 % Every call to SMALL_solve function will generate following output:
28 %
29 % SMALL.solver.solution - contains solution vector x
30 % SMALL.solver.reconstructed - vector containing signal reconstructed
31 % from the solution
32 % SMALL.solver.time - time that solver spent to find the solution
33 %
34 % SMALL_plot function plots the SMALL.solver.solution and reconstructed
35 % against original signal.
36 %
37 % In this particular example we are testing SMALL_cgp, SMALL_chol,
38 % SolveOMP form SparseLab and greed_pcgp form Sparsify against "PROB006
39 % Daubechies basis, Gaussian ensemble measurement basis, piecewise cubic
40 % polynomial signal" from Sparco.
41 %
42 %
43 % Ivan Damnjanovic 2009%
44 %
45 % SPARCO Copyright 2008, Ewout van den Berg and Michael P. Friedlander
46 % http://www.cs.ubc.ca/labs/scl/sparco
47 % $Id: exGPSR.m 1040 2008-06-26 20:29:02Z ewout78 $
48
49 fprintf('\n\nExample test of SMALL solver against their counterparts on Sparco problems.\n\n');
50
51 %%
52 % Generate SPARCO problem
53
54 global SMALL
55 SMALL.Problem = generateProblem(6, 'P', 6, 'm', 270,'n',1024, 'show');
56 %%
57
58 %%
59 % SMALL Conjugate Gradient test
60
61 SMALL.solver.toolbox='SMALL';
62 SMALL.solver.name='SMALL_cgp';
63
64 % In the following string all parameters except matrix, measurement vector
65 % and size of solution need to be specified. If you are not sure which
66 % parameters are needed for particular solver type "help <Solver name>" in
67 % MATLAB command line
68
69 SMALL.solver.param='200, 1e-14';
70
71 SMALL=SMALL_solve(SMALL);
72
73
74 SMALL_plot(SMALL);
75
76 %%
77 % SMALL OMP with Cholesky update test
78
79 SMALL.solver.toolbox='SMALL';
80 SMALL.solver.name='SMALL_chol';
81
82 % In the following string all parameters except matrix, measurement vector
83 % and size of solution need to be specified. If you are not sure which
84 % parameters are needed for particular solver type "help <Solver name>" in
85 % MATLAB command line
86
87 SMALL.solver.param='200, 1e-14';
88
89 SMALL=SMALL_solve(SMALL);
90
91 SMALL_plot(SMALL);
92 %%
93 % SolveOMP from SparseLab test
94
95 SMALL.solver.toolbox='SparseLab';
96 SMALL.solver.name='SolveOMP';
97
98 % In the following string all parameters except matrix, measurement vector
99 % and size of solution need to be specified. If you are not sure which
100 % parameters are needed for particular solver type "help <Solver name>" in
101 % MATLAB command line
102
103 SMALL.solver.param='200, 0, 0, 0, 1e-14';
104
105 SMALL=SMALL_solve(SMALL);
106
107 SMALL_plot(SMALL);
108
109 %%
110 % greed_pcgp from Sparsify test
111
112 SMALL.solver.toolbox='Sparsify';
113 SMALL.solver.name='greed_pcgp';
114
115 % In the following string all parameters except matrix, measurement vector
116 % and size of solution need to be specified. If you are not sure which
117 % parameters are needed for particular solver type "help <Solver name>" in
118 % MATLAB command line
119
120 SMALL.solver.param='''stopCrit'', ''M'', ''stopTol'', 200';
121
122 SMALL=SMALL_solve(SMALL);
123
124 SMALL_plot(SMALL);
125
126 %%
127
128
129 end % function SMALL_solver_test