The entire project contains matrices holding the coefficients of (in)equality-systems or affine functions.
All matrices have the following structure:
<coeff. of indices> | <coeff. of structure-par.> | <coeff. of pseudo-par.> | <offset>
Therefore the following line of a matrix
1 5 -1 3 7 2
describes the expression
1 * i + 5 * j -1 * k + 3 * m + 7 * loopo_lambda_dep1 + 2
assuming that (i,j,k) is the vector of all indices, m is a structure
parameter and loopo_lambda_dep1 is a pseudo-parameter introduced by this
tool. (The appearance of pseudo parameters is optional and depends on
the context, and it is explained at the place where they appear.)
In order to denote matrices with rational coefficients, we prefix the matrix with a global denominator, i.e., all matrix entries must be divided by this denominator.
If an (in)equality system is described the implicit meaning
is:
expression <op> 0, where <op> evaluates to ">="
for inequalities or "=" for equations.
Format for matrices: