関数 solvesdp のオプション shift を利用し，連立 PDLMI
（ただし，
）の解を求めるよう，gs.m を修正する（線形目的関数 
を最小化する凸最適化問題）．
>> gs_shift
SeDuMi 1.3 by AdvOL, 2005-2008 and Jos F. Sturm, 1998-2003.
Alg = 2: xz-corrector, theta = 0.250, beta = 0.500
eqs m = 29, order n = 97, dim = 809, blocks = 13
nnz(A) = 748 + 0, nnz(ADA) = 825, nnz(L) = 427
it : b*y gap delta rate t/tP* t/tD* feas cg cg prec
0 : 3.23E+003 0.000
1 : -1.35E+000 1.11E+003 0.000 0.3437 0.9000 0.9000 3.09 1 1 2.5E+002
2 : -3.08E+000 4.85E+002 0.000 0.4362 0.9000 0.9000 0.84 1 1 1.4E+002
3 : -5.23E+000 2.44E+002 0.000 0.5032 0.9000 0.9000 0.57 1 1 9.1E+001
4 : -7.64E+000 1.25E+002 0.000 0.5106 0.9000 0.9000 0.41 1 1 6.1E+001
5 : -9.87E+000 7.31E+001 0.000 0.5872 0.9000 0.9000 0.27 1 1 4.7E+001
6 : -1.30E+001 4.04E+001 0.000 0.5523 0.9000 0.9000 0.19 1 1 3.6E+001
7 : -1.65E+001 2.27E+001 0.000 0.5632 0.9000 0.9000 0.14 1 1 2.7E+001
8 : -2.01E+001 1.35E+001 0.000 0.5933 0.9000 0.9000 0.09 1 1 2.2E+001
9 : -2.48E+001 7.72E+000 0.000 0.5722 0.9000 0.9000 0.07 1 1 1.7E+001
10 : -2.97E+001 4.58E+000 0.000 0.5928 0.9000 0.9000 0.07 1 1 1.4E+001
11 : -3.55E+001 2.72E+000 0.000 0.5934 0.9000 0.9000 0.07 1 1 1.1E+001
12 : -4.17E+001 1.67E+000 0.000 0.6132 0.9000 0.9000 0.09 1 1 9.1E+000
13 : -4.89E+001 1.02E+000 0.000 0.6151 0.9000 0.9000 0.11 1 1 7.3E+000
14 : -5.67E+001 6.43E-001 0.000 0.6280 0.9000 0.9000 0.14 1 1 6.0E+000
15 : -6.55E+001 4.04E-001 0.000 0.6281 0.9000 0.9000 0.17 1 1 4.8E+000
16 : -7.48E+001 2.58E-001 0.000 0.6374 0.9000 0.9000 0.21 1 1 3.9E+000
17 : -8.52E+001 1.64E-001 0.000 0.6360 0.9000 0.9000 0.25 1 1 3.1E+000
18 : -9.61E+001 1.05E-001 0.000 0.6402 0.9000 0.9000 0.30 1 1 2.5E+000
19 : -1.08E+002 6.69E-002 0.000 0.6375 0.9000 0.9000 0.34 1 1 2.0E+000
20 : -1.19E+002 4.26E-002 0.000 0.6377 0.9000 0.9000 0.39 1 1 1.5E+000
21 : -1.31E+002 2.70E-002 0.000 0.6340 0.9000 0.9000 0.44 1 1 1.2E+000
22 : -1.43E+002 1.70E-002 0.000 0.6301 0.9000 0.9000 0.50 1 1 8.7E-001
23 : -1.54E+002 1.06E-002 0.000 0.6228 0.9000 0.9000 0.56 1 1 6.3E-001
24 : -1.64E+002 6.49E-003 0.000 0.6118 0.9000 0.9000 0.63 1 1 4.4E-001
25 : -1.73E+002 3.86E-003 0.000 0.5941 0.9000 0.9000 0.70 1 1 2.9E-001
26 : -1.80E+002 2.18E-003 0.000 0.5665 0.9000 0.9000 0.77 1 2 1.8E-001
27 : -1.86E+002 1.14E-003 0.000 0.5242 0.9000 0.9000 0.84 1 2 1.0E-001
28 : -1.90E+002 5.31E-004 0.000 0.4637 0.9000 0.9000 0.90 1 2 5.0E-002
29 : -1.92E+002 2.00E-004 0.000 0.3758 0.9000 0.9000 0.94 2 2 1.9E-002
30 : -1.93E+002 5.20E-005 0.000 0.2609 0.9000 0.9000 0.96 4 4 5.2E-003
31 : -1.94E+002 1.17E-005 0.000 0.2255 0.9000 0.9000 0.96 7 7 1.2E-003
32 : -1.94E+002 3.05E-006 0.000 0.2599 0.9000 0.9000 0.94 7 7 3.2E-004
33 : -1.94E+002 1.14E-007 0.000 0.0373 0.9900 0.9900 0.99 7 8 1.2E-005
34 : -1.94E+002 5.41E-009 0.000 0.0476 0.9900 0.9900 1.00 9 13 5.7E-007
Run into numerical problems.
iter seconds digits c*x b*y
34 1.7 8.5 -1.9388608254e+002 -1.9388608189e+002
|Ax-b| = 6.6e-007, [Ay-c]_+ = 1.3E-009, |x|= 3.8e+004, |y|= 3.8e+002
Detailed timing (sec)
Pre IPM Post
7.996E-003 1.348E+000 2.997E-003
Max-norms: ||b||=1, ||c|| = 2,
Cholesky |add|=0, |skip| = 4, ||L.L|| = 36019.4.
sol =
yalmiptime: 0.1570
solvertime: 1.3600
info: 'Numerical problems (SeDuMi-1.3)'
problem: 4
dimacs: [3.2775e-007 0 0 6.6175e-010 -1.6638e-009 3.3633e-008]
gamma_opt =
193.8861
X_0_opt =
0.9762 -0.0721 -3.4455 1.1962
-0.0721 4.5644 0.1768 -28.6062
-3.4455 0.1768 27.7753 -11.8619
1.1962 -28.6062 -11.8619 184.3916
X_1_opt =
-0.1514 0.1396 1.4519 -1.5444
0.1396 -6.3823 -0.2883 39.9948
1.4519 -0.2883 -20.8206 14.9115
-1.5444 39.9948 14.9115 -258.5054
F_0_opt =
-1.3316 0.2786 6.4606 -3.2160
F_1_opt =
0.5817 -0.1155 -8.3416 5.9741
pres =
7.5018e-003
6.1004e-004
9.9113e-008
1.0483e-007
9.9256e-008
1.0131e-007
1.0283e-007
1.0178e-007
9.9481e-008
9.9162e-008
2.3440e-003
9.8677e-008