How many different types of conics exist in hyperbolic plane?
Euclidean geometry has three, of course. But when I was trying to find out results for the hyperbolic plane, the best thing I found freely accessible was this: http://www.mathnet.ru/links/0116e85b4ef8e4fdf19cbe340a1eb771/ivm1975.pdf
My Russian is a bit rusty, though. I can see that this guy found 9 distinct types, but I have no clue how they look or what properties they have.
Can anyone help me, please?
On
If you think about the three non-degenerate types ellipse, parabola and hyperbola, then these are classified by their points of intersection with the line at infinity. You either have two distinct points of intersection for a hyperbola, or a single point of tangency with algebraic multiplicity two for the parabola, or a complex conjugate pair of points for the ellipse.
Now in the world of Cayley-Klein geometries, the line of infinity is just a special conic, namely one which degenerates to a line, with the ideal circle points as distingusihed points through which all the tangents pass. Anyway, you'd have to multiply all these counts by two, and that would describe the situation in a vocabuilary you can translate.
In Hyperbolic geometry, the fundamental conic, the thing which corresponds to the line at infinity, is a real and non-degenerate conic, e.g. the unit circle of the Beltrami-Klein model. Any other conic can intersect that conic in up to four real points. Algebraically you always have exactly four points. That is unless the conics are equal, of course. So which combinations can you have? To keep things systematic, go by increasing number of real intersections.
I hope I didn't forget a situation. You could distinguish some of these even further: does the conic touch the fundamental conic from the inside or from the outside? For the first case, with no real intersection, you could consider a single conjugate pair of contact points, corresponding to hyperbolic circles which contain no ideal points. You could construct circles which touch the fundamental conic from the outside in two points. And probably a bunch of other characteristics in addition to these.
Since I have no Russian at all, I've got problems matching this to the text you referenced. But the fact that the rightmost column seems to be categorizing by four objects, $F_1,F_2,\bar F_1$ and $\bar F_2$ suggests that the central idea is the same. If you can decipher keywords there, chances are they include “real”, “conjugate”, “double” or things like this.
Doing a bit of copy and paste to Google Translate, it becomes apparent that the righternmost column doesn't discuss the points of intersection, and whether or not they are real, complex or coinciding. Instead they speak about the foci (Фокусы) which are either on the inside, the outside or the boundary of the fundamental conic (абсолюта). Apparently they pair up foci to make these distinctions, since e.g. in Euclidean geometry, an ellipse and a hyperbola have two real and two complex conjugate foci each. (Those that are real for the ellipse are complex for the hyperbola and vice versa.)
But I'm not sure what foci exactly they are referring to. If they were referring to the normal Euclidean foci, then the whole setup would have a strong Euclidean dependency. All four foci in Euclidean geometry can be obtained a focus by constructing tangents to the ideal circle points $(1,i,0)$ and $(1,-i,0)$ and intersecting these with one another. The ideal circle points, as I mentioned before, are part of the fundamental conic for Euclidean geometry. So to do this correctly, one should draw tangents to the fundamental conic of hyperbolic geometry. And for two non-degenerate conics, this would mean four common tangents, and a total of six points which could be called foci.
As the OP correctly pointed out in a comment, Figure 4 of that paper shows four collinear points labeled $F_1,F_2,\bar F_1$ and $\bar F_2$. This doesn't agree with either definition of foci I just mentioned. At the moment, the connection is unclear.
Defining Conic Sections
The first question, of course, is how to even define conic sections on the hyperbolic plane. The usual method is to use the hyperboloid model, which identifies the hyperbolic plane with the hyperboloid $$ x^2 + y^2 - z^2 = -1,\qquad z\geq 1. $$ In this model, isometries of the hyperbolic plane correspond to linear transformations of $\mathbb{R}^3$ that map this hyperboloid to itself.
Using this model, define a conic section in $\mathbb{H}^2$ to be any intersection of the hyperboloid with a set of the form $$ \{\textbf{x}\in\mathbb{R}^3 \mid \textbf{x}^{T}S\textbf{x} = 0\}\tag*{(*)} $$ where $S$ is a nondegenerate $3\times 3$ symmetric matrix. That is, a conic section is any subset of the hyperboloid $x^2+y^2-z^2 = -1$ defined by an equation of the form $$ Ax^2 + Bxy + Cy^2 + Dxz + Eyz + Fz^2 = 0 $$ where $A,B,C,D,E,F\in\mathbb{R}$.
Note that the sets of the form $(*)$ given above are precisely the elliptic doubles cones with a vertex at the origin. (This assumes $S$ is not positive or negative definite, in which case the given set consists only of the origin.) This justifies the name “conic section”. Note also that these conic sections are invariant under isometries of $\mathbb{H}^2$ (since isometries correspond to linear maps), which seems like a good property for conic sections to have.
The Klein Model
We can understand these conic sections better by switching to the Klein model of the hyperbolic plane. In this model, $\mathbb{H}^2$ is the interior of a unit disk, and hyperbolic lines are straight chords of the disk. (This is different from the more common Poincaré disk model, in which hyperbolic lines are circular arcs.) The Klein model can be thought of as the disk $x^2+y^2=1$ in the $z=1$ plane of $\mathbb{R}^3$, which projects onto the hyperboloid from the origin.
Now, here's a few things to know about the Klein model:
It's not a conformal model. That is, the angle at which two chords appear to intersect is different from the actual angle of intersection in the hyperbolic plane.
The isometries of the Klein model are precisely the projective transformations of the plane that map the unit disk to itself.
The reason that the Klein model is helpful is that hyperbolic conic sections in the Klein model are precisely Euclidean conic sections that intersect the unit disk. This has to do with the fact that the Klein disk really lies on the plane $z=1$ in $\mathbb{R}^3$, and the intersection of this plane with the elliptic cones defined above gives conic sections on the plane.
Therefore, understanding conic sections on the hyperbolic plane is precisely the same thing as understanding them on the Euclidean plane up to projective transformations that preserve the unit disk.
The Classification
It appears that the classification of hyperbolic conic sections goes back to an 1882 paper by William E. Story:
Story, William E. “On non-Euclidean properties of conics”. American Journal of Mathematics 5, no. 1 (1882): 358–381.
Story classifies conic sections according to the number and multiplicities of intersections between the conic section and the boundary circle. This results in eight types of conic sections
An ellipse is an ellipse contained entirely within the interior of the unit disk.
A hyperbola is a conic section that intersects the unit circle at four different points. (Such a conic section may be a portion of an ellipse, parabola, or hyperbola in the Euclidean plane, though which of these three types it is may change under hyperbolic isometries.)
A semi-hyperbola is a conic section that intersects the unit circle transversely at two different points. (Again, this may be an ellipse, parabola, or hyperbola in the Euclidean plane.)
An elliptic parabola is an ellipse or circle in the disk that intersects the unit circle at one point of tangency.
A hyperbolic parabola is a conic section that intersects the unit circle three times, with one being a point of tangency.
A semi-circular parabola is a conic section that has the unit circle as one of its osculating circles (i.e. they have a point of third-order contact) and also intersects the unit circle at one additional point.
A horocycle (called a “circular parabola” by Story) is an ellipse in the unit disk that has a fourth-order contact with the unit circle. That is, it is an ellipse that has the unit circle as an osculating circle at one of the endpoints of its minor axis.
A circle is a circle contained entirely in the interior of the unit disk, and an equidistance conic is an ellipse that is tangent to the unit disk at two points. (Story refers to both of these cases simply as “circles”.)
Story mentions that some of these eight types can be further subdivided, and later authors often increased the number of types to 11 or 12. For example, a classification into 12 types can be found on pg. 142 in the following book:
Liebmann, Heinrich. Nichteuklidische geometrie. Vol. 49. GJ Göschen, 1905.
I don't read German, so I can't confirm this, but this classification is reproduced in English on pg. 257 of the following book, which also has some nice figures on the following page:
Rosenfeld, Boris, and Bill Wiebe. Geometry of Lie groups. Vol. 393. Springer Science & Business Media, 2013.
In particular, Rosenfeld and Wiebe (presumably following Liebmann) make the following distinctions:
They further classify Story's “circles” into true circles and equidistant conics.
They call a hyperbola “concave” if it has four common tangent lines with the unit circle in the Euclidean plane, and “convex”
They classify hyperbolic parabolas into three types depending on the number of branches and the number of common tangent lines in the Euclidean plane.
This results in a total of 12 types. This classification has the advantage that it is symmetric with respect to projective duality, since it uses both the intersection points and the common tangent lines.