tmtd.cpp
Go to the documentation of this file.
00001 
00002 
00003 
00006 
00007 
00008 //#define WANT_STREAM
00009 
00010 #include "include.h"
00011 #include "newmatap.h"
00012 
00013 #include "tmt.h"
00014 
00015 #ifdef use_namespace
00016 using namespace NEWMAT;
00017 #endif
00018 
00019 ReturnMatrix Inverter1(const CroutMatrix& X)
00020 {
00021    Matrix Y = X.i();
00022    Y.Release();
00023    return Y.ForReturn();
00024 }
00025 
00026 // this version forces a copy
00027 ReturnMatrix Inverter2(CroutMatrix X)
00028 {
00029    Matrix Y = X.i();
00030    Y.Release();
00031    return Y.ForReturn();
00032 }
00033 
00034 ReturnMatrix Inverter1(const BandLUMatrix& X)
00035 {
00036    Matrix Y = X.i();
00037    Y.Release();
00038    return Y.ForReturn();
00039 }
00040 
00041 // this version forces a copy
00042 ReturnMatrix Inverter2(BandLUMatrix X)
00043 {
00044    Matrix Y = X.i();
00045    Y.Release();
00046    return Y.ForReturn();
00047 }
00048 
00049 ReturnMatrix LU1(const Matrix& A)
00050 {
00051    Tracer et1("LU1 - Crout");
00052    CroutMatrix X = A;
00053    return X.for_return();
00054 }
00055 
00056 ReturnMatrix LU2(const Matrix& A)
00057 {
00058    Tracer et1("LU2 - Crout");
00059    CroutMatrix X = A; X.release();
00060    return X.for_return();
00061 }
00062 
00063 ReturnMatrix LU3(const Matrix& A)
00064 {
00065    Tracer et1("LU3 - Crout");
00066    CroutMatrix* X = new CroutMatrix(A); MatrixErrorNoSpace(X);
00067    X->release_and_delete();
00068    return X->for_return();
00069 }
00070 
00071 ReturnMatrix LU1(const BandMatrix& A)
00072 {
00073    Tracer et1("LU1 - BandLU");
00074    BandLUMatrix X = A;
00075    return X.for_return();
00076 }
00077 
00078 ReturnMatrix LU2(const BandMatrix& A)
00079 {
00080    Tracer et1("LU2 - BandLU");
00081    BandLUMatrix X = A; X.release();
00082    return X.for_return();
00083 }
00084 
00085 ReturnMatrix LU3(const BandMatrix& A)
00086 {
00087    Tracer et1("LU3 - BandLU");
00088    BandLUMatrix* X = new BandLUMatrix(A); MatrixErrorNoSpace(X);
00089    X->release_and_delete();
00090    return X->for_return();
00091 }
00092 
00093 
00094 
00095 void CircularShift(const Matrix& X1, int first, int last)
00096 {
00097       Matrix X; UpperTriangularMatrix U1, U2;
00098       int n = X1.Ncols();
00099 
00100       // Try right circular shift of columns
00101       X = X1; QRZ(X, U1);
00102       RightCircularUpdateCholesky(U1, first, last);
00103       X = X1.Columns(1,first-1) | X1.Column(last)
00104          | X1.Columns(first,last-1) | X1.Columns(last+1,n);
00105       QRZ(X, U2);
00106       X = U1 - U2; Clean(X, 0.000000001); Print(X);
00107 
00108       // Try left circular shift of columns
00109       X = X1; QRZ(X, U1);
00110       LeftCircularUpdateCholesky(U1, first, last);
00111       X = X1.Columns(1,first-1) | X1.Columns(first+1,last)
00112          | X1.Column(first) | X1.Columns(last+1,n);
00113       QRZ(X, U2);
00114       X = U1 - U2; Clean(X, 0.000000001); Print(X);
00115 }
00116 
00117 class TestUpdateQRZ
00118 {
00119    int m,n1,n2,n3;
00120    Matrix X1, X2, X3;
00121    MultWithCarry mwc;   // Uniform random number generator
00122 public:
00123    void Reset();
00124    TestUpdateQRZ(int mx, int n1x, int n2x=0, int n3x=0)
00125       : m(mx), n1(n1x), n2(n2x), n3(n3x) { Reset(); }
00126    void DoTest();
00127    void ClearRow(int i)       { X1.Row(i) = 0.0; }
00128    void SetRow(int i, int j)  { X1.Row(i) = X1.Row(j); }
00129 };
00130 
00131 
00132 
00133 void trymatd()
00134 {
00135    Tracer et("Thirteenth test of Matrix package");
00136    Tracer::PrintTrace();
00137    Matrix X(5,20);
00138    int i,j;
00139    for (j=1;j<=20;j++) X(1,j) = j+1;
00140    for (i=2;i<=5;i++) for (j=1;j<=20; j++) X(i,j) = (long)X(i-1,j) * j % 1001;
00141    SymmetricMatrix S; S << X * X.t();
00142    Matrix SM = X * X.t() - S;
00143    Print(SM);
00144    LowerTriangularMatrix L = Cholesky(S);
00145    Matrix Diff = L*L.t()-S; Clean(Diff, 0.000000001);
00146    Print(Diff);
00147    {
00148       Tracer et1("Stage 1");
00149       LowerTriangularMatrix L1(5);
00150       Matrix Xt = X.t(); Matrix Xt2 = Xt;
00151       QRZT(X,L1);
00152       Diff = L - L1; Clean(Diff,0.000000001); Print(Diff);
00153       UpperTriangularMatrix Ut(5);
00154       QRZ(Xt,Ut);
00155       Diff = L - Ut.t(); Clean(Diff,0.000000001); Print(Diff);
00156       Matrix Y(3,20);
00157       for (j=1;j<=20;j++) Y(1,j) = 22-j;
00158       for (i=2;i<=3;i++) for (j=1;j<=20; j++)
00159          Y(i,j) = (long)Y(i-1,j) * j % 101;
00160       Matrix Yt = Y.t(); Matrix M,Mt; Matrix Y2=Y;
00161       QRZT(X,Y,M); QRZ(Xt,Yt,Mt);
00162       Diff = Xt - X.t(); Clean(Diff,0.000000001); Print(Diff);
00163       Diff = Yt - Y.t(); Clean(Diff,0.000000001); Print(Diff);
00164       Diff = Mt - M.t(); Clean(Diff,0.000000001); Print(Diff);
00165       Diff = Y2 * Xt2 * S.i() - M * L.i();
00166       Clean(Diff,0.000000001); Print(Diff);
00167    }
00168 
00169    ColumnVector C1(5);
00170    {
00171       Tracer et1("Stage 2");
00172       X.ReSize(5,5);
00173       for (j=1;j<=5;j++) X(1,j) = j+1;
00174       for (i=2;i<=5;i++) for (j=1;j<=5; j++)
00175          X(i,j) = (long)X(i-1,j) * j % 1001;
00176       for (i=1;i<=5;i++) C1(i) = i*i;
00177       CroutMatrix A = X;
00178       ColumnVector C2 = A.i() * C1; C1 = X.i()  * C1;
00179       X = C1 - C2; Clean(X,0.000000001); Print(X);
00180    }
00181 
00182    {
00183       Tracer et1("Stage 3");
00184       X.ReSize(7,7);
00185       for (j=1;j<=7;j++) X(1,j) = j+1;
00186       for (i=2;i<=7;i++) for (j=1;j<=7; j++)
00187          X(i,j) = (long)X(i-1,j) * j % 1001;
00188       C1.ReSize(7);
00189       for (i=1;i<=7;i++) C1(i) = i*i;
00190       RowVector R1 = C1.t();
00191       Diff = R1 * X.i() - ( X.t().i() * R1.t() ).t(); Clean(Diff,0.000000001);
00192       Print(Diff);
00193    }
00194 
00195    {
00196       Tracer et1("Stage 4");
00197       X.ReSize(5,5);
00198       for (j=1;j<=5;j++) X(1,j) = j+1;
00199       for (i=2;i<=5;i++) for (j=1;j<=5; j++)
00200          X(i,j) = (long)X(i-1,j) * j % 1001;
00201       C1.ReSize(5);
00202       for (i=1;i<=5;i++) C1(i) = i*i;
00203       CroutMatrix A1 = X*X;
00204       ColumnVector C2 = A1.i() * C1; C1 = X.i()  * C1; C1 = X.i()  * C1;
00205       X = C1 - C2; Clean(X,0.000000001); Print(X);
00206    }
00207 
00208 
00209    {
00210       Tracer et1("Stage 5");
00211       int n = 40;
00212       SymmetricBandMatrix B(n,2); B = 0.0;
00213       for (i=1; i<=n; i++)
00214       {
00215          B(i,i) = 6;
00216          if (i<=n-1) B(i,i+1) = -4;
00217          if (i<=n-2) B(i,i+2) = 1;
00218       }
00219       B(1,1) = 5; B(n,n) = 5;
00220       SymmetricMatrix A = B;
00221       ColumnVector X(n);
00222       X(1) = 429;
00223       for (i=2;i<=n;i++) X(i) = (long)X(i-1) * 31 % 1001;
00224       X = X / 100000L;
00225       // the matrix B is rather ill-conditioned so the difficulty is getting
00226       // good agreement (we have chosen X very small) may not be surprising;
00227       // maximum element size in B.i() is around 1400
00228       ColumnVector Y1 = A.i() * X;
00229       LowerTriangularMatrix C1 = Cholesky(A);
00230       ColumnVector Y2 = C1.t().i() * (C1.i() * X) - Y1;
00231       Clean(Y2, 0.000000001); Print(Y2);
00232       UpperTriangularMatrix CU = C1.t().i();
00233       LowerTriangularMatrix CL = C1.i();
00234       Y2 = CU * (CL * X) - Y1;
00235       Clean(Y2, 0.000000001); Print(Y2);
00236       Y2 = B.i() * X - Y1; Clean(Y2, 0.000000001); Print(Y2);
00237 
00238       LowerBandMatrix C2 = Cholesky(B);
00239       Matrix M = C2 - C1; Clean(M, 0.000000001); Print(M);
00240       ColumnVector Y3 = C2.t().i() * (C2.i() * X) - Y1;
00241       Clean(Y3, 0.000000001); Print(Y3);
00242       CU = C1.t().i();
00243       CL = C1.i();
00244       Y3 = CU * (CL * X) - Y1;
00245       Clean(Y3, 0.000000001); Print(Y3);
00246 
00247       Y3 = B.i() * X - Y1; Clean(Y3, 0.000000001); Print(Y3);
00248 
00249       SymmetricMatrix AI = A.i();
00250       Y2 = AI*X - Y1; Clean(Y2, 0.000000001); Print(Y2);
00251       SymmetricMatrix BI = B.i();
00252       BandMatrix C = B; Matrix CI = C.i();
00253       M = A.i() - CI; Clean(M, 0.000000001); Print(M);
00254       M = B.i() - CI; Clean(M, 0.000000001); Print(M);
00255       M = AI-BI; Clean(M, 0.000000001); Print(M);
00256       M = AI-CI; Clean(M, 0.000000001); Print(M);
00257 
00258       M = A; AI << M; M = AI-A; Clean(M, 0.000000001); Print(M);
00259       C = B; BI << C; M = BI-B; Clean(M, 0.000000001); Print(M);
00260    }
00261 
00262    {
00263       Tracer et1("Stage 5");
00264       SymmetricMatrix A(4), B(4);
00265       A << 5
00266         << 1 << 4
00267         << 2 << 1 << 6
00268         << 1 << 0 << 1 << 7;
00269       B << 8
00270         << 1 << 5
00271         << 1 << 0 << 9
00272         << 2 << 1 << 0 << 6;
00273       LowerTriangularMatrix AB = Cholesky(A) * Cholesky(B);
00274       Matrix M = Cholesky(A) * B * Cholesky(A).t() - AB*AB.t();
00275       Clean(M, 0.000000001); Print(M);
00276       M = A * Cholesky(B); M = M * M.t() - A * B * A;
00277       Clean(M, 0.000000001); Print(M);
00278    }
00279    
00280    {
00281       Tracer et1("Stage 6");
00282       int N=49;
00283       int i;
00284       SymmetricBandMatrix S(N,1);
00285       Matrix B(N,N+1); B=0;
00286       for (i=1;i<=N;i++) { S(i,i)=1; B(i,i)=1; B(i,i+1)=-1; }
00287       for (i=1;i<N; i++) S(i,i+1)=-.5;
00288       DiagonalMatrix D(N+1); D = 1;
00289       B = B.t()*S.i()*B - (D-1.0/(N+1))*2.0;
00290       Clean(B, 0.000000001); Print(B);
00291    }
00292 
00293    {
00294       Tracer et1("Stage 7");
00295       // Copying and moving CroutMatrix
00296       Matrix A(7,7);
00297       A.Row(1) <<  3 <<  2 << -1 <<  4 << -3 <<  5 <<  9;
00298       A.Row(2) << -8 <<  7 <<  2 <<  0 <<  7 <<  0 << -1;
00299       A.Row(3) <<  2 << -2 <<  3 <<  1 <<  9 <<  0 <<  3;
00300       A.Row(4) << -1 <<  5 <<  2 <<  2 <<  5 << -1 <<  2;
00301       A.Row(5) <<  4 << -4 <<  1 <<  9 << -8 <<  7 <<  5;
00302       A.Row(6) <<  1 << -2 <<  5 << -1 << -2 <<  5 <<  1;
00303       A.Row(7) << -6 <<  3 << -1 <<  8 << -1 <<  2 <<  2;
00304       RowVector D(30); D = 0;
00305       Real x = determinant(A);
00306       CroutMatrix B = A;
00307       D(1) = determinant(B) / x - 1;
00308       Matrix C = A * Inverter1(B) - IdentityMatrix(7);
00309       Clean(C, 0.000000001); Print(C);
00310       // Test copy constructor (in Inverter2 and ordinary copy)
00311       CroutMatrix B1; B1 = B;
00312       D(2) = determinant(B1) / x - 1;
00313       C = A * Inverter2(B1) - IdentityMatrix(7);
00314       Clean(C, 0.000000001); Print(C);
00315       // Do it again with release
00316       B.release(); B1 = B;
00317       D(2) = B.nrows(); D(3) = B.ncols(); D(4) = B.size();
00318       D(5) = determinant(B1) / x - 1;
00319       B1.release();
00320       C = A * Inverter2(B1) - IdentityMatrix(7);
00321       D(6) = B1.nrows(); D(7) = B1.ncols(); D(8) = B1.size();
00322       Clean(C, 0.000000001); Print(C);
00323       // see if we get an implicit invert
00324       B1 = -A; 
00325       D(9) = determinant(B1) / x + 1; // odd number of rows - sign will change 
00326       C = -A * Inverter2(B1) - IdentityMatrix(7);
00327       Clean(C, 0.000000001); Print(C);
00328       // check for_return
00329       B = LU1(A); B1 = LU2(A); CroutMatrix B2 = LU3(A);
00330       C = A * B.i() - IdentityMatrix(7); Clean(C, 0.000000001); Print(C);
00331       D(10) = (B == B1 ? 0 : 1) + (B == B2 ? 0 : 1);
00332       // check lengths
00333       D(13) = B.size()-49;
00334       // check release(2)
00335       B1.release(2);
00336       B2 = B1; D(15) = B == B2 ? 0 : 1;
00337       CroutMatrix B3 = B1; D(16) = B == B3 ? 0 : 1;
00338       D(17) = B1.size();
00339       // some oddments
00340       B1 = B; B1 = B1.i(); C = A - B1.i(); Clean(C, 0.000000001); Print(C);
00341       B1 = B; B1.release(); B1 = B1; B2 = B1;
00342       D(19) = B == B1 ? 0 : 1; D(20) = B == B2 ? 0 : 1;
00343       B1.cleanup(); B2 = B1; D(21) = B1.size(); D(22) = B2.size();
00344       GenericMatrix GM = B; C = A.i() - GM.i(); Clean(C, 0.000000001); Print(C);
00345       B1 = GM; D(23) = B == B1 ? 0 : 1;
00346       B1 = A * 0; B2 = B1; D(24) = B2.is_singular() ? 0 : 1;
00347       // check release again - see if memory moves
00348       const Real* d = B.const_data();
00349       const int* i = B.const_data_indx();
00350       B1 = B;
00351       const Real* d1 = B1.const_data();
00352       const int* i1 = B1.const_data_indx();
00353       B1.release(); B2 = B1;
00354       const Real* d2 = B2.const_data();
00355       const int* i2 = B2.const_data_indx();
00356       D(25) = (d != d1 ? 0 : 1) + (d1 == d2 ? 0 : 1)
00357          + (i != i1 ? 0 : 1) + (i1 == i2 ? 0 : 1);
00358   
00359       Clean(D, 0.000000001); Print(D);
00360    }
00361 
00362    {
00363       Tracer et1("Stage 8");
00364       // Same for BandLUMatrix
00365       BandMatrix A(7,3,2);
00366       A.Row(1) <<  3 <<  2 << -1;
00367       A.Row(2) << -8 <<  7 <<  2 <<  0;
00368       A.Row(3) <<  2 << -2 <<  3 <<  1 <<  9;
00369       A.Row(4) << -1 <<  5 <<  2 <<  2 <<  5 << -1;
00370       A.Row(5)       << -4 <<  1 <<  9 << -8 <<  7 <<  5;
00371       A.Row(6)             <<  5 << -1 << -2 <<  5 <<  1;
00372       A.Row(7)                   <<  8 << -1 <<  2 <<  2;
00373       RowVector D(30); D = 0;
00374       Real x = determinant(A);
00375       BandLUMatrix B = A;
00376       D(1) = determinant(B) / x - 1;
00377       Matrix C = A * Inverter1(B) - IdentityMatrix(7);
00378       Clean(C, 0.000000001); Print(C);
00379       // Test copy constructor (in Inverter2 and ordinary copy)
00380       BandLUMatrix B1; B1 = B;
00381       D(2) = determinant(B1) / x - 1;
00382       C = A * Inverter2(B1) - IdentityMatrix(7);
00383       Clean(C, 0.000000001); Print(C);
00384       // Do it again with release
00385       B.release(); B1 = B;
00386       D(2) = B.nrows(); D(3) = B.ncols(); D(4) = B.size();
00387       D(5) = determinant(B1) / x - 1;
00388       B1.release();
00389       C = A * Inverter2(B1) - IdentityMatrix(7);
00390       D(6) = B1.nrows(); D(7) = B1.ncols(); D(8) = B1.size();
00391       Clean(C, 0.000000001); Print(C);
00392       // see if we get an implicit invert
00393       B1 = -A; 
00394       D(9) = determinant(B1) / x + 1; // odd number of rows - sign will change 
00395       C = -A * Inverter2(B1) - IdentityMatrix(7);
00396       Clean(C, 0.000000001); Print(C);
00397       // check for_return
00398       B = LU1(A); B1 = LU2(A); BandLUMatrix B2 = LU3(A);
00399       C = A * B.i() - IdentityMatrix(7); Clean(C, 0.000000001); Print(C);
00400       D(10) = (B == B1 ? 0 : 1) + (B == B2 ? 0 : 1);
00401       // check lengths
00402       D(11) = B.bandwidth().lower()-3;
00403       D(12) = B.bandwidth().upper()-2;
00404       D(13) = B.size()-42;
00405       D(14) = B.size2()-21;
00406       // check release(2)
00407       B1.release(2);
00408       B2 = B1; D(15) = B == B2 ? 0 : 1;
00409       BandLUMatrix B3 = B1; D(16) = B == B3 ? 0 : 1;
00410       D(17) = B1.size();
00411       // Compare with CroutMatrix
00412       CroutMatrix CM = A;
00413       C = CM.i() - B.i(); Clean(C, 0.000000001); Print(C);
00414       D(18) = determinant(CM) / x - 1;
00415       // some oddments
00416       B1 = B; CM = B1.i(); C = A - CM.i(); Clean(C, 0.000000001); Print(C);
00417       B1 = B; B1.release(); B1 = B1; B2 = B1;
00418       D(19) = B == B1 ? 0 : 1; D(20) = B == B2 ? 0 : 1;
00419       B1.cleanup(); B2 = B1; D(21) = B1.size(); D(22) = B2.size();
00420       GenericMatrix GM = B; C = A.i() - GM.i(); Clean(C, 0.000000001); Print(C);
00421       B1 = GM; D(23) = B == B1 ? 0 : 1;
00422       B1 = A * 0; B2 = B1; D(24) = B2.is_singular() ? 0 : 1;
00423       // check release again - see if memory moves
00424       const Real* d = B.const_data(); const Real* dd = B.const_data();
00425       const int* i = B.const_data_indx();
00426       B1 = B;
00427       const Real* d1 = B1.const_data(); const Real* dd1 = B1.const_data();
00428       const int* i1 = B1.const_data_indx();
00429       B1.release(); B2 = B1;
00430       const Real* d2 = B2.const_data(); const Real* dd2 = B2.const_data();
00431       const int* i2 = B2.const_data_indx();
00432       D(25) = (d != d1 ? 0 : 1) + (d1 == d2 ? 0 : 1)
00433          + (dd != dd1 ? 0 : 1) + (dd1 == dd2 ? 0 : 1)
00434          + (i != i1 ? 0 : 1) + (i1 == i2 ? 0 : 1);
00435 
00436       Clean(D, 0.000000001); Print(D);
00437    }
00438 
00439    {
00440       Tracer et1("Stage 9");
00441       // Modification of Cholesky decomposition
00442 
00443       int i, j;
00444 
00445       // Build test matrix
00446       Matrix X(100, 10);
00447       MultWithCarry mwc;   // Uniform random number generator
00448       for (i = 1; i <= 100; ++i) for (j = 1; j <= 10; ++j)
00449          X(i, j) = 2.0 * (mwc.Next() - 0.5);
00450       Matrix X1 = X;     // save copy
00451 
00452       // Form sums of squares and products matrix and Cholesky decompose
00453       SymmetricMatrix A; A << X.t() * X;
00454       UpperTriangularMatrix U1 = Cholesky(A).t();
00455 
00456       // Do QR decomposition of X and check we get same triangular matrix
00457       UpperTriangularMatrix U2;
00458       QRZ(X, U2);
00459       Matrix Diff = U1 - U2; Clean(Diff, 0.000000001); Print(Diff);
00460 
00461       // Try adding new row to X and updating triangular matrix 
00462       RowVector NewRow(10);
00463       for (j = 1; j <= 10; ++j) NewRow(j) = 2.0 * (mwc.Next() - 0.5);
00464       UpdateCholesky(U2, NewRow);
00465       X = X1 & NewRow; QRZ(X, U1);
00466       Diff = U1 - U2; Clean(Diff, 0.000000001); Print(Diff);
00467 
00468       // Try removing two rows and updating triangular matrix
00469       DowndateCholesky(U2, X1.Row(20));
00470       DowndateCholesky(U2, X1.Row(35));
00471       X = X1.Rows(1,19) & X1.Rows(21,34) & X1.Rows(36,100) & NewRow; QRZ(X, U1);
00472       Diff = U1 - U2; Clean(Diff, 0.000000001); Print(Diff);
00473 
00474       // Circular shifts
00475 
00476       CircularShift(X, 3,6);
00477       CircularShift(X, 5,5);
00478       CircularShift(X, 4,5);
00479       CircularShift(X, 1,6);
00480       CircularShift(X, 6,10);
00481    }
00482    
00483    {
00484       Tracer et1("Stage 10");
00485       // Try updating QRZ, QRZT decomposition
00486       TestUpdateQRZ tuqrz1(10, 100, 50, 25); tuqrz1.DoTest();
00487       tuqrz1.Reset(); tuqrz1.ClearRow(1); tuqrz1.DoTest();
00488       tuqrz1.Reset(); tuqrz1.ClearRow(1); tuqrz1.ClearRow(2); tuqrz1.DoTest();
00489       tuqrz1.Reset(); tuqrz1.ClearRow(5); tuqrz1.ClearRow(6); tuqrz1.DoTest();
00490       tuqrz1.Reset(); tuqrz1.ClearRow(10); tuqrz1.DoTest();
00491       TestUpdateQRZ tuqrz2(15, 100, 0, 0); tuqrz2.DoTest();
00492       tuqrz2.Reset(); tuqrz2.ClearRow(1); tuqrz2.DoTest();
00493       tuqrz2.Reset(); tuqrz2.ClearRow(1); tuqrz2.ClearRow(2); tuqrz2.DoTest();
00494       tuqrz2.Reset(); tuqrz2.ClearRow(5); tuqrz2.ClearRow(6); tuqrz2.DoTest();
00495       tuqrz2.Reset(); tuqrz2.ClearRow(15); tuqrz2.DoTest();
00496       TestUpdateQRZ tuqrz3(5, 0, 10, 0); tuqrz3.DoTest();
00497       
00498    }
00499    
00500 //   cout << "\nEnd of Thirteenth test\n";
00501 }
00502 
00503 void TestUpdateQRZ::Reset()
00504 {
00505    int i, j;
00506    // set up Matrices
00507    X1.ReSize(m, n1); X2.ReSize(m, n2); X3.ReSize(m, n3);
00508    for (i = 1; i <= m; ++i) for (j = 1; j <= n1; ++j)
00509       X1(i, j) = 2.0 * (mwc.Next() - 0.5);
00510    for (i = 1; i <= m; ++i) for (j = 1; j <= n2; ++j)
00511       X2(i, j) = 2.0 * (mwc.Next() - 0.5);
00512    for (i = 1; i <= m; ++i) for (j = 1; j <= n3; ++j)
00513       X3(i, j) = 2.0 * (mwc.Next() - 0.5);
00514 }
00515 
00516 void TestUpdateQRZ::DoTest()
00517 {
00518    Matrix XT1 = X1.t(), XT2 = X2.t(), XT3 = X3.t();
00519    Matrix X = X1 | X2 | X3;
00520    SymmetricMatrix SM; SM << X * X.t();
00521    LowerTriangularMatrix L, LX, LY, L0;
00522    QRZT(X, L);
00523    X = X1 | X2 | X3;
00524    LY.ReSize(m); LY = 0.0;
00525    UpdateQRZT(X, LY);
00526    QRZT(X1, LX); UpdateQRZT(X2, LX); UpdateQRZT(X3, LX);
00527    Matrix Diff = L * L.t() - SM; Clean(Diff, 0.000000001); Print(Diff);
00528    Diff = SM - LX * LX.t(); Clean(Diff, 0.000000001); Print(Diff);
00529    Diff = SM - LY * LY.t(); Clean(Diff, 0.000000001); Print(Diff);
00530    UpperTriangularMatrix U, UX, UY;
00531    X = XT1 & XT2 & XT3;
00532    QRZ(X, U);
00533    Diff = U.t() * U - SM; Clean(Diff, 0.000000001); Print(Diff);
00534    UY.ReSize(m); UY = 0.0;
00535    X = XT1 & XT2 & XT3;
00536    UpdateQRZ(X, UY);
00537    Diff = SM - UY.t() * UY; Clean(Diff, 0.000000001); Print(Diff);
00538    QRZ(XT1, UX); UpdateQRZ(XT2, UX); UpdateQRZ(XT3, UX);
00539    Diff = SM - UX.t() * UX; Clean(Diff, 0.000000001); Print(Diff);
00540 }   
00541 


kni
Author(s): Martin Günther
autogenerated on Mon Aug 14 2017 02:44:13