int2d
definite 2D integral by quadrature method
Syntax
[I, err] = int2d(X, Y, f) [I, err] = int2d(X, Y, f, params) [I, err] = int2d(xmin, xmax, ymin, ymax, f) [I, err] = int2d(xmin, xmax, ymin, ymax, f, params)
Arguments
- X
a 3 by
N
array containing the abscissae of the vertices of the N triangles- Y
a 3 by
N
array containing the ordinates of the vertices of the N triangles- xmin, xmax, ymin, ymax
real scalars defining a rectangle in the plane
- f
external (function or list or string) defining the integrand
f(u,v)
- params
a real vector
[tol, iclose, maxtri, mevals, iflag]
. The default value is[1.d-10, 1, 50, 4000, 1]
.- tol
the desired bound on the error. If
iflag=0
,tol
is interpreted as a bound on the relative error; ififlag=1
, the bound is on the absolute error.- iclose
an integer parameter that determines the selection of LQM0 or LQM1 methods. If
iclose=1
then LQM1 is used. Any other value oficlose
causes LQM0 to be used. LQM0 uses function values only at interior points of the triangle. LQM1 is usually more accurate than LQM0 but involves evaluating the integrand at more points including some on the boundary of the triangle. It will usually be better to use LQM1 unless the integrand has singularities on the boundary of the triangle.- maxtri
the maximum number of triangles in the final triangulation of the region
- mevals
the maximum number of function evaluations to be allowed. This number will be effective in limiting the computation only if it is less than 94*
maxtri
when LQM1 is specified or 56*maxtri
when LQM0 is specified.- iflag
if
iflag=0
,tol
is interpreted as a bound on the relative error; ififlag=1
, the bound is on the absolute error.
- I
the integral value
- err
the estimated error
Description
int2d
computes the two-dimensional integral of a
function f
over a region consisting of
N
triangles or over a single rectangle [xmin,xmax]x[ymin,ymax] (internally
divided into two triangles). A total error estimate is obtained and
compared with a tolerance - tol
- that is provided as
input to the subroutine. The error tolerance is treated as either relative
or absolute depending on the input value of iflag
. A
'Local Quadrature Module' is applied to each input triangle and estimates
of the total integral and the total error are computed. The local
quadrature module is either subroutine LQM0 or subroutine LQM1 and the
choice between them is determined by the value of the input variable
iclose
.
If the total error estimate exceeds the tolerance, the triangle with
the largest absolute error is divided into two triangles by a median to
its longest side. The local quadrature module is then applied to each of
the subtriangles to obtain new estimates of the integral and the error.
This process is repeated until either (1) the error tolerance is
satisfied, (2) the number of triangles generated exceeds the input
parameter maxtri
, (3) the number of integrand
evaluations exceeds the input parameter mevals
, or (4)
the function senses that roundoff error is beginning to contaminate the
result.
Examples
deff('z=f(x,y)','z=cos(x+y)') // computes the integral over the triangle (0,0),(1 0),(0,1) X = [0 1 0]'; Y = [0 0 1]'; [I,e] = int2d(X,Y,f) // computes the integral over the square [0,1]x[0,1] [I,e] = int2d(0,1,0,1,f)
See also
Report an issue | ||
<< impl | Equations différentielles | int3d >> |