sandbox/easystab/stab2014/Lspacing.m

LSPACING function

Coded by Paul Valcke and Luis Bernardos. This function is used for controlling the discretization of a mesh in a given direction. It is especially suited for shrinking the mesh near the walls. It accepts a lambda \lambda\in[-3,3] parameter that interpolates the linearly-spaced x vector with a sinus or cosinus spacing.

function [ x, x_linear ] = Lspacing( Nx, lambda, Lx )

INPUTS:

• Nx: Number of divisions.
• lambda: Kind of division.
• Lx: Total Length of the spaced item.

OUTPUTS:

• x: Distance positioning of the control point. Vector.
• x_linear: Regularly (linear) spaced control points. Vector.

Xtr = (0:(1/Nx):1)'; % Transition vector
x_linear = zeros((Nx-1),1);
for ii = 1:Nx
    x_linear(ii,1) = (Xtr(ii+1)+Xtr(ii))/2;
end

Interpolation between linear and sinus(\lambda\in[3,2])

The output spacing ranges from linear to a sinus spacing:

\displaystyle x = 1 + sin(\frac{\pi}{2}(-1+x_l))

if lambda <= 3 && lambda >= 2 % interpolation between
   L3 = x_linear;
   L2 = 1 + sin((pi/2).*(-1+x_linear));
   A = 3 - lambda;
   B = lambda - 2;
   x = B.*L3 + A.*L2;

Interpolation between sinus and cosinus(\lambda\in[2,1])

The output spacing ranges from sinus spacing:

\displaystyle x = 1 + sin(\frac{\pi}{2}(-1+x_l))

To a cosinus spacing:

\displaystyle x = \frac{1}{2}(1+cos(\pi(1+x_l)))

elseif lambda <= 2 && lambda >= 1
   L2 = 1 + sin((pi/2).*(-1+x_linear));
   L1 = 0.5.*(1+cos(pi.*(1+x_linear)));
   A = 2 - lambda;
   B = lambda - 1;
   x = B.*L2 + A.*L1;

Interpolation between cosinus and linear(\lambda\in[1,0])

The output spacing ranges from cosinus x = \frac{1}{2}(1+cos(\pi(1+x_l))) to linear sinus spacing.

elseif lambda <= 1 && lambda >= 0
   L1 = 0.5.*(1+cos(pi.*(1+x_linear)));
   L0 = x_linear;
   A = 1 - lambda;
   B = lambda;
   x = B.*L1 + A.*L0;

Interpolation between linear and cosinus(\lambda\in[0,-1])

The output spacing ranges from linear to cosinus x = \frac{1}{2}(1+cos(\pi(1+x_l))).

elseif lambda <= 0 && lambda >= -1
   L0 = x_linear;
   L1 = 0.5.*(1+cos(pi.*(1+x_linear)));
   A = - lambda;
   B = lambda - (-1);
   x = B.*L0 + A.*L1;

Interpolation between cosinus and sinus (\lambda\in[-1,-2])

The output spacing ranges from cosinus x = \frac{1}{2}(1+cos(\pi(1+x_l))) to sinus x=sin(\frac{\pi}{2}x_l) .

elseif lambda <= -1 && lambda >= -2
   L1 = 0.5.*(1+cos(pi.*(1+x_linear)));
   L2 = sin((pi/2).*x_linear);
   A = (-1) - lambda;
   B = lambda - (-2);
   x = B.*L1 + A.*L2;

Interpolation between sinus and linear (\lambda\in[-2,-3])

The output spacing ranges from sinus x=sin(\frac{\pi}{2}x_l) to linear .

elseif lambda <= -2 && lambda >= -3
   L2 = sin((pi/2).*x_linear);
   L3 = x_linear;
   A = (-2) - lambda;
   B = lambda - (-3);
   x = B.*L2 + A.*L3;
else
    error('lambda = %0.2f out of limits. Lambda should be between [3,-3].',lambda);
end

x = x.*Lx;
x_linear = x_linear.*Lx;
end