functionregRoot2

Anti-symmetric approximation of square root with discontinuous factor so that the first derivative is finite and continuous

Extends from Modelica.Icons.Function (Icon for functions).

Information

Approximates the function

y = if x ≥ 0 then sqrt(k1*x) else -sqrt(k2*abs(x)), with k1, k2 ≥ 0

in such a way that within the region -x_small ≤ x ≤ x_small, the function is described by two polynomials of third order (one in the region -x_small .. 0 and one within the region 0 .. x_small) such that

  • The derivative at x=0 is finite.
  • The overall function is continuous with a continuous first derivative everywhere.
  • If parameter use_yd0 = false, the two polynomials are constructed such that the second derivatives at x=0 are identical. If use_yd0 = true, the derivative at x=0 is explicitly provided via the additional argument yd0. If necessary, the derivative yd0 is automatically reduced in order that the polynomials are strict monotonically increasing [Fritsch and Carlson, 1980].

Typical screenshots for two different configurations are shown below. The first one with k1=k2=1:

regRoot2_a.png

and the second one with k1=1 and k2=3:

regRoot2_b.png

The (smooth) derivative of the function with k1=1, k2=3 is shown in the next figure:

regRoot2_c.png

Literature

Fritsch F.N. and Carlson R.E. (1980):
Monotone piecewise cubic interpolation. SIAM J. Numerc. Anal., Vol. 17, No. 2, April 1980, pp. 238-246

Inputs

TypeNameDefaultDescription
RealxAbscissa value
Realx_small0.01Approximation of function for |x| <= x_small
Realk11y = if x>=0 then sqrt(k1*x) else -sqrt(k2*|x|)
Realk21y = if x>=0 then sqrt(k1*x) else -sqrt(k2*|x|)
Booleanuse_yd0false= true, if yd0 shall be used
Realyd01Desired derivative at x=0: dy/dx = yd0

Outputs

TypeNameDefaultDescription
RealyOrdinate value

Contents

NameDescription
regRoot2_utilityprotectedInterpolating with two 3-order polynomials with a prescribed derivative at x=0

Revisions

  • Sept., 2010 by Martin Otter:
    Improved so that k1=0 and/or k2=0 is also possible.
  • Nov., 2005 by Martin Otter:
    Designed and implemented.