Quadratic functions in two variables

We originally might have the following set of data on the triangle with vertices (1,1),(3,1),(1.5,3):
x1 x2 f( x1 , x2 )
1.01.01.0
2.01.0-2.0
3.01.00.0
1.252.0-1.0
1.53.02.0
2.252.01.0
Taking the vertex (1.0,1.0) as reference point we might consider this triangle as the image of the standard triangle with vertices (0,0),(1,0),(0,1) under the linear transformation
( x1 x2 )=( 1 1 )+( 20.5 02 )( ξ1 ξ2 )

and hence the inverse transformation is
( ξ1 ξ2 )=( 0.5-0.125 00.5 )( x1 -1 x2 -1 )

Now, on the standard triangle we have the data
ξ1 ξ2 f( ξ1 , ξ2 )
0.00.01.0
0.50.0-2.0
1.00.00.0
0.00.5-1.0
0.01.02.0
0.50.51.0
and expressing the quadratic here in ( ξ1 , ξ2 ) is quite simple: we get the linear system
( 100000 10.500.2500 110100 100.5000.25 101001 10.50.50.250.250.25 )( f0 g1 g2 h11 /2 h12 /2 h22 /2 )=( 1 -2 0 -1 2 1 )

with the solution
q( ξ1 , ξ2 )=1-11 ξ1 -9 ξ2 +\tfrac12(20 ξ2 2 +40 ξ1 ξ2 +20 ξ2 2 )

Expressing the variables ξi in the variables xi we get q expressed in these. This technique works in any dimension on any full simplex.


File translated from TEX by TTM Unregistered, version 4.03.
On 16 Jun 2016, 17:26.