2

Identity

$$(p+q+r)^2=p^2+q^2+r^2+2qr+2pr+2pq\tag{1}$$

has a classical visual representation (Fig. 1) which constitutes a "graphical proof" for positive $p,q,r$ :

enter image description here

$\textit{Fig. 1 : Classical visualization of identity (1)}.$

I have found a different graphical representation of (1) using barycentric coordinates as given on Fig. 2 :

enter image description here

$$\textit{Fig. 2 : Another visualization of identity (1) in terms of barycentric coordinates.}$$

where $(p,q,r)$ (normalized with $p+q+r=1$) are the barycentric coordinates wrt "base triangle" $ABC$ of the common point to all triangles and parallelograms.

Indeed, the different terms of the RHS of (1) are interpretable as relative areas of triangles or parallelograms ("relative" meaning in this context "divided by the area of triangle $ABC$").

The proofs are straightforward : for example, the relative area of the small triangle on the left, (with one of its sides on $AB$), is equal to the determinant :

$$\left|\matrix{(1-q)&p&p\\q&(1-p)&q\\0&0&r}\right|=r((1-p)(1-q)-pq)=r(1-p-q)=r^2$$

where the barycentric coordinates of the vertices are placed in columns.

Same type of computations for the other areas.

My question is :

I would like to present this nice elementary result to a certain audience. I am 100% sure that this result has already been found (and maybe applied to interesting problems), but haven't been able to find any anterior reference. Could somebody find one ?

Remarks :

  1. There is more to say about Fig. 2 : it provides a visualization of the barycentric coordinates of $M$ as the following ratios of lengths :

$$p=\frac{A_1A_2}{BC},\ \ q=\frac{B_1B_2}{AC},\ \ r=\frac{C_1C_2}{AB}$$

(easy proof).

  1. A same type of result exists with 3D barycentric coordinates (sometimes called tetraedric coordinates) $(p,q,r,s)$ with respect to a tetrahedron $ABCD$, this time with relative volumes in correspondence with identity :

$$(p+q+r+s)^3=p^3+q^3+...+3q^2r+3q^2s+...$$

Jean Marie
  • 88,997
  • Can you define the "certain audience"? EG Esp if it's middle school students, you might want to use similar triangles instead of matrix determinant to show the result. $\quad$ Also, why do you need a reference, as opposed to the lazy approach of "proof by obviousness"? – Calvin Lin Jan 21 '25 at 00:16
  • 1
    @Calvin Lin Thanks for answering so quickly. 1) A certain audience : undergraduate students who have some knowledge of geometry, in particular who already know barycentric coordiantes 2) I do not "need it", it's true, but, I would like to have references if it is possible, for my own sake, not necessarily for people to which I will present it... – Jean Marie Jan 21 '25 at 00:20
  • 1
    This reminds me of some of the work on splines and blossoming from about 25 or 30 years ago. Here's a reference that might be of some use (and convince you that the blossoming folks knew about this, or did NOT know...): https://www.researchgate.net/profile/Ronald-Goldman/publication/281510225_A_Tutorial_Introduction_to_Blossoming/links/5d4b13bba6fdcc370a811db3/A-Tutorial-Introduction-to-Blossoming.pdf – John Hughes Jan 21 '25 at 01:33
  • See a somewhat similar question here and its treatment without barycentric coordinates, with a lot of square roots. – Jean Marie Jan 21 '25 at 17:19
  • Questions like this one are easily solved using the property displayed on Fig. 2 – Jean Marie Jan 21 '25 at 17:20
  • Another question related to this configuration here. – Jean Marie Jan 21 '25 at 18:22
  • A 8 pages nice gathering of (mainly ill-known) formulas/methods for barycentric coordinates here. – Jean Marie Jan 21 '25 at 22:22
  • The natural uprising of squares or (in a reciprocal way) square roots in this kind of questions is illustrated in an answer of mine here. – Jean Marie Jan 22 '25 at 09:57
  • See also this answer. – Jean Marie Feb 03 '25 at 22:16

0 Answers0