The parametric equations of a projective line passing through $P = \left[X_1:Y_1:Z_1\right]$ and $Q = \left[X_2:Y_2:Z_2\right]$ are :

$\left[X:Y:Z\right] = \alpha\left[X_1:Y_1:Z_1\right] + \beta\left[X_2:Y_2:Z_2\right]$, $\alpha,\beta \in \mathbb{R}$?.

Is it an abuse of notation?

  • $\begingroup$ You should take $\beta=1-\alpha.$ $\endgroup$ – Jean Marie Feb 1 '17 at 17:19
  • 1
    $\begingroup$ At present; this parameterization allows $\alpha=\beta=0$. But $[0:0:0]$ isn't in $\Bbb P^2$. $\endgroup$ – Semiclassical Feb 1 '17 at 17:21
  • $\begingroup$ Because $\beta = 1 - \alpha$? if consider $(\alpha,\beta) \in \mathbb{R^2}$ \ $\lbrace 0 \rbrace $? $\endgroup$ – Orested Feb 1 '17 at 17:34

Your equation is perfectly valid (with the caveat spotted by @Semiclassical).

The parametric equation of a line defined by 2 points in projective space is usually derived from a reasoning on the rank of matrix:

$$\left(\begin{array}{l l l} X_1 & X_2 & X \\ Y_1 & Y_2 & Y \\ Z_1 & Z_2 & Z \end{array}\right) $$

By definition of a projective line, this rank has to be 2; which is equivalent to the fact that the last column can be expressed as linear combination of the first two columns:

$$\tag{1}M:=\left(\begin{array}{l l l} X \\ Y \\ Z \end{array}\right)=\alpha_1\left(\begin{array}{l l l} X_1 \\ Y_1 \\ Z_1 \end{array}\right)+\alpha_2\left(\begin{array}{l l l} X_2 \\ Y_2 \\ Z_2 \end{array}\right)$$

Edit (following a remark by @MvG pointing an error of mine in a first draft):

Two cases can occur:

  • either $\alpha_1+\alpha_2 \neq 0$, and point $M$ is said "an ordinary point". In this case, we can further normalize this linear combination (considering that we can take taking $\alpha_1+\alpha_2=1$), due do the fact that projective coordinates are defined up to a multiplicative factor. Thus (1) implies:

$$M:=\left(\begin{array}{l l l} X \\ Y \\ Z \end{array}\right) \sim k\alpha_1\left(\begin{array}{l l l} X_1 \\ Y_1 \\ Z_1 \end{array}\right)+k\alpha_2\left(\begin{array}{l l l} X_2 \\ Y_2 \\ Z_2 \end{array}\right)$$

where $k$ is any nonzero real number, for example $1/(\alpha_1+\alpha_2)$, and symbol $\sim$ means "proportional to" (= with equivalence relationship "defined up to a non zero multiplicative factor"). We check immediately that the new coefficients $\alpha'_1:=\alpha_1/(\alpha_1+\alpha_2),\alpha'_2:=\alpha_2/(\alpha_1+\alpha_2)$ sum up to $1$.

Remark: this way of writing things is important because, in this way, as we are in affine geometry, the bridge is made with the barycentrical approach.

  • or $\alpha_1+\alpha_2=0$. In this case, (1) becomes:

$$M:=\left(\begin{array}{l l l} X \\ Y \\ Z \end{array}\right) \sim \left(\begin{array}{l l l} X_1 \\ Y_1 \\ Z_1 \end{array}\right)-\left(\begin{array}{l l l} X_2 \\ Y_2 \\ Z_2 \end{array}\right)$$

i.e., is equivalent to a vector. Point $M$ is said "at infinity".

Remark: A key point in the understanding of projective geometry is the bijective correspondence between vectors (up to a sign change) and points at infinity.

  • $\begingroup$ I consider the part about further normalization to be confusing. If I understand you correctly, you seem to imply that one could always choose $\alpha$ and $\beta$ such that $\alpha+\beta=1$, but that is not the case for $\alpha=-\beta$. Therefore if you make that assumption you loose one point, turning the projective line into an affine one. $\endgroup$ – MvG Feb 6 '17 at 10:58
  • $\begingroup$ @MvG You are right. I had my head elsewhere. I have modified all the second part of my answer. $\endgroup$ – Jean Marie Feb 6 '17 at 16:19
  • $\begingroup$ Better now. In my opinion this choice of “infinity” is pretty arbitrary. Some people would consider $[\alpha:\beta]=[Z_2:-Z_1]$ as infinity, as it leads to a zero $Z$ coordinate for the result, corresponding to the line at infinity in the standard $Z=1$ embedding. Having you introduce a different concept of infinity might be confusing, particularly since it depends on representatives. I also would consider the homogeneous coordinates to be vectors, so I can't subscribe to your statement that only points at infinity correspond to vectors. $\endgroup$ – MvG Feb 6 '17 at 17:09
  • $\begingroup$ @MvG I understand your point. As the OP has not said that he/she works with the standard "embedding" (Z≠0) I preferred this presentation which is a way one can "projectify" affine plane defined with barycentrical coordinates. I agree that it depends on representatives; but when you are a beginner in this domain... sticking to coordinates is not bad. $\endgroup$ – Jean Marie Feb 6 '17 at 18:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.