Fix bug #314:

- remove most of the metaprogramming kung fu in MathFunctions.h (only keep functions that differs from the std)
- remove the overloads for array expression that were in the std namespace

unsupported/Eigen/src/IterativeSolvers/IncompleteCholesky.h

@@ -116,6 +116,7 @@ template<typename Scalar, int _UpLo, typename OrderingType>
 template<typename _MatrixType>
 void IncompleteCholesky<Scalar,_UpLo, OrderingType>::factorize(const _MatrixType& mat)
 {
+  using std::sqrt;
   eigen_assert(m_analysisIsOk && "analyzePattern() should be called first"); 
   
   // FIXME Stability: We should probably compute the scaling factors and the shifts that are needed to ensure a succesful LLT factorization and an efficient preconditioner. 
@@ -182,7 +183,7 @@ void IncompleteCholesky<Scalar,_UpLo, OrderingType>::factorize(const _MatrixType
        m_info = NumericalIssue; 
        return; 
      }
-     RealScalar rdiag = internal::sqrt(RealScalar(diag));
+     RealScalar rdiag = sqrt(RealScalar(diag));
      Scalar scal = Scalar(1)/rdiag; 
      vals[colPtr[j]] = rdiag;
      // Insert the largest p elements in the matrix and scale them meanwhile  

unsupported/Eigen/src/IterativeSolvers/IterationController.h

@@ -129,7 +129,8 @@ class IterationController
     bool converged() const { return m_res <= m_rhsn * m_resmax; }
     bool converged(double nr)
     {
-      m_res = internal::abs(nr); 
+      using std::abs;
+      m_res = abs(nr); 
       m_resminreach = (std::min)(m_resminreach, m_res);
       return converged();
     }

unsupported/Eigen/src/IterativeSolvers/MINRES.h

@@ -32,6 +32,7 @@ namespace Eigen {
                     const Preconditioner& precond, int& iters,
                     typename Dest::RealScalar& tol_error)
         {
+            using std::sqrt;
             typedef typename Dest::RealScalar RealScalar;
             typedef typename Dest::Scalar Scalar;
             typedef Matrix<Scalar,Dynamic,1> VectorType;