provide overloads for operator* for matrix/vector multiplies

[Gunter]

Since many people a familiar with using operator* we should add the corresponding overloads to uBLAS. A good reference is the syntax used by MATLAB. The final syntax should be

    matrix<double> A;
    vector<double> x;
    vector<double> y;

    cout << 2.0 * y << endl;
    cout << y * 2 << endl;
    cout << A * 2 << endl;
    cout << 2 * A  << endl;
    cout << A * y << endl;
    cout << trans(y) * A  << endl;
    cout << A * A << endl;

Similar to trans() we should provide herm() for complex matrices.

the syntax for inner and outer products could be

    vector<double> x,a,b;
    x = trans(a) * b;     // inner prod
    x = a * trans(b);     // outer prod

the syntax for products with higher precision should be

  prec(A*b);
  prec(trans(a) * b);

a *= operator should also be provided

b *= A; 
A *= B;

which could be mapped to the correspondy axpy_prod() call.

further tasks:

• find a solution for the product of three matrices which is currently only possible by stating an explicit temporary matrix.

[Gunter]

(In [56045]) see #3397

vector_expression.hpp: added enable_if to expression scalar * vector and vector * scalar

matrix_expression.hpp: added enable_if to expression scalar * matrix and matrix * scalar

[Gunter]

(In [56625]) see #3397

• merged [56045] into release
• add enable_if to operator* for scalar*vector, vector*scalar, scalar*matrix, matrix*scalar in order to hinder unexpected instantiation.

[David Bellot]

I will explore eigen solution to the problem of 3 matrices

[anonymous]

a new ticket 4442 with a simple implementation has been opened and a patch proposed.

[maxim.yegorushkin@…]

When is this functionality expected to merged into a boost release please?