The target of this paper is two-fold. On the one hand, we address the question of the reliability of' many recently proposed high-order finite elements. On the other hand, new quadrilateral edge elements are proposed and the improvements of the performances they permit are stressed.
INTRODUCTIONHigh-order edge elements make it possible to improve the performances of finite element simulators. For this reason, in recent years, quite a lot of high-order edge elements have been introduced [1,2,3] after the pioneering work of Nedelec [4]. However, before analysing their performances, one should establish the reliability of the finite elements considered. This is an important issue since, for many years, finite element simulators were plagued by the so-called spurious modes. Thus, in this paper, we firstly report some recent results of the present authors assessing the reliability of many of the new elements recently introduced. Once the class of a priori reliable finite elements has sufficiently been widened, it makes sense to compare the performances in order to find the best one. However, the "best" element could also be an element not defined so far and, for this reason, it is important to keep on looking for new and more efficient elements. As a first attempt in this direction, we will firstly define a new element on rectangles and then we will compare its performances with those of other well known spurious-free elements. In the simple example considered the performances of the new element are significantly better than those of all existing elements of the same order.