numeric_limits<double>::min() is positive!

Contrary to what one might expect, it is the smallest representable
*positive* number.

Using -max() works because by IEEE 754, floats are symmetric!

src/mynonlinearity.hh

@@ -58,7 +58,7 @@ template <int dimension> class MyNonlinearity {
 
   void directionalDomain(const VectorType &, const VectorType &,
                          Interval<double> &dom) const {
-    dom[0] = std::numeric_limits<double>::min();
+    dom[0] = -std::numeric_limits<double>::max();
     dom[1] = std::numeric_limits<double>::max();
   }