- Optimization of TransformHelperPoly::multiply(Vect): no realloc() calls, directly allocate the final vector avoiding the whole buffer copy.
- Implementation of TransformHelperPoly::multiply(Matdense)
- Optimization of TransformHelperPoly::basisChebyshev: building factors in parallel with OpenMP.
- Multiple impls of TransformHelperPoly::poly to compute linear combination of polynomials.