NUMERICAL METHODS
27.25 Laplace’s equation,
∂^2 V
∂x^2
+
∂^2 V
∂y^2
=0,
is to be solved for the region and boundary conditions shown in figure 27.7.
40 40 40 40 40 40 40
20 20 20
V=80
V=0
−∞ ∞
Figure 27.7 Region, boundary values and initial guessed solution values.
Starting from the given initial guess for the potential valuesV, and using the
simplest possible form of relaxation, obtain a better approximation to the actual
solution. Do not aim to be more accurate than±0.5 units, and so terminate the
process when subsequent changes would be no greater than this.
27.26 Consider the solution,φ(x, y), of Laplace’s equation in two dimensions using a
relaxation method on a square grid with common spacingh.Asinthemaintext,
denoteφ(x 0 +ih, y 0 +jh)byφi,j. Further, defineφm,ni,j by
φm,ni,j≡
∂m+nφ
∂xm∂yn
evaluated at (x 0 +ih, y 0 +jh).
(a) Show that
φ^4 i,j,^0 +2φ^2 i,j,^2 +φ^0 i,j,^4 =0.
(b) Working up to terms of orderh^5 , find Taylor series expansions, expressed in
terms of theφm,ni,j,for
S±, 0 =φi+1,j+φi− 1 ,j,
S 0 ,±=φi,j+1+φi,j− 1.
(c) Find a corresponding expansion, to the same order of accuracy, forφi± 1 ,j+1+
φi± 1 ,j− 1 and hence show that
S±,±=φi+1,j+1+φi+1,j− 1 +φi− 1 ,j+1+φi− 1 ,j− 1
has the form
4 φ^0 i,j,^0 +2h^2 (φ^2 i,j,^0 +φ^0 i,j,^2 )+
h^4
6
(φi,j^4 ,^0 +6φ^2 i,j,^2 +φ^0 i,j,^4 ).
(d) Evaluate the expression 4(S±, 0 +S 0 ,±)+S±,±and hence deduce that a possible
relaxation scheme, good to the fifth order inh, is to recalculate eachφi,jas
the weighted mean of the current values of its four nearest neighbours (each
with weight^15 ) and its four next-nearest neighbours (each with weight 201 ).