A Direct Finite - Di erence Scheme for SolvingPDEs over
نویسنده
چکیده
Classical nite-diierence methods apply only over regions that are rectangular or decomposed in several rectangular subdomains. We call a nite-diierence scheme direct if it can be applied over any region decomposed by means of a logically-rectangular grid. We introduce a second-order and easy-to-use direct scheme to solve boundary-value problems. Deduction of the method is elementary and its results compare well with other methods. The design of numerical methods to solve partial diierential equations (PDEs) must take in account the geometrical complexity of the physical region where the problem is deened. The classical nite-diierence schemes are well-suited for simple regions and, in that case, their implementation is straightforward. Over diicult regions, the nite-element method is commonly used. It has a strong mathematical-background supporting it and tools for subdividing the region into simpler cells are used; however, the complexity of data structure and the amount of memory required to its implementation has suggested the search for simpler methods. The use of the Mapping Method ((8]) and nite-diierence schemes is an alternate way, it requires the knowledge of a diieomorphism {a change of coordinates{ between the unit cell (logical space) and the region. The problem is then transformed and solved, not on the original region, but on the unit cell. Changes of coordinates are known in an analytical form only for few regions; in general, they have to be calculated in a discrete way, this is done using numerical grid generation methods. A disadvantage of the mapping method is that calculations involve the use of the jacobian; if the mapping is nearly singular in a point of the region, the jacobian is close to zero with the corresponding lack of precision. Another approach is to design schemes for discretizing partial derivatives directly on the physical region. As with the mapping method, it requires to have a grid over the region. Some such schemes have been reported by Steinberg, Shashkov, Hyman and Castillo ((10], 6]). In the present paper we introduce a simple bidimensional nite-diierence scheme applied directly on a grid over the physical region. We present some examples over irregular regions in order to show its performance. In section 1, we introduce the kind of problems to be solved. Section 2 contains a brief review of grid generations. Section 3 is devoted to explain our scheme. Finally, in section 4, some test problems are solved and we discuss the results.
منابع مشابه
Orr Sommerfeld Solver Using Mapped Finite Di?erence Scheme for Plane Wake Flow
Linear stability analysis of the three dimensional plane wake flow is performed using a mapped finite di?erence scheme in a domain which is doubly infinite in the cross–stream direction of wake flow. The physical domain in cross–stream direction is mapped to the computational domain using a cotangent mapping of the form y = ?cot(??). The Squire transformation [2], proposed by Squire, is also us...
متن کاملA Non-dissipative Staggered Fourth-order Accurate Explicit Finite Difference Scheme for the Time-domain Maxwell’s Equations
We consider a divergence-free non-dissipative fourth-order explicit staggered nite di erence scheme for the hyperbolic Maxwell's equations. Special one-sided di erence operators are derived in order to implement the scheme near metal boundaries and dielectric interfaces. Numerical results show the scheme is long-time stable, and is fourth-order convergent over complex domains that include diele...
متن کاملPhysicsUpwind Compact and Explicit High - Order Finite Di erenceSchemes for Direct Numerical Simulation of High - Speed
Direct numerical simulation of transitional and turbulent hypersonic boundary layers using the Navier-Stokes equations requires high-order accurate numerical methods to resolve a wide range of time and length scales. Compact or explicit nite di erence methods used for such simulation have mainly been central di erence schemes containing only phase errors without any numerical dissipation. Centr...
متن کاملTheoretical Analysis of Finite Difference Algorithms for Linear Shape from Shading
This paper presents and analyzes four explicit, two implicit and four semi-implicit finite difference algorithms for the linear shape from shading problem. Comparisons of accuracy, solvability, stability and convergence of these schemes indicate that the weighted semi-implicit scheme and the box scheme are better than the other ones because they can be calculated more easily, they are more accu...
متن کاملNumerical Simulations of the Complex Modi ed Korteweg de Vries Equation
In this paper implementations of three numerical schemes for the numerical simulation of the complex modi ed Korteweg de Vries CMKdV equation are reported The rst is an integrable scheme derived by methods related to the Inverse Scattering Transform IST The second is derived from the rst and is called the local IST scheme The third is a standard nite di erence scheme for the CMKdV equation Trav...
متن کامل