The Isotropic Cone of the Bilinear Form on $\mathbb{C}^2$

Introduction

The complex bilinear form $$ B(Q_1,Q_2)=Q_{1\blacktriangle}Q_{2\blacktriangle}+Q_{1\blacktriangledown}Q_{2\blacktriangledown} $$ is symmetric and nondegenerate, and it is indefinite: there are nonzero vectors $Q$ with $B(Q,Q)=0$. The set of those vectors is the isotropic cone $$ \mathcal{N}=\{Q\in\mathbb{C}^2:B(Q,Q)=0\}=\{Q:Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=0\}. $$ The equation factors over $\mathbb{C}$, $$ Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=(Q_{\blacktriangle}+iQ_{\blacktriangledown})(Q_{\blacktriangle}-iQ_{\blacktriangledown}), $$ so the cone is the union of two complex lines. This is the first place where the complex bilinear form differs sharply from the Hermitian form: the Hermitian form is definite and has no isotropic vectors, the symplectic form is nondegenerate and has many isotropic subspaces (every complex line is isotropic for it), while $B$ has exactly two isotropic lines.

This article treats the cone, its two lines, its real picture, its image in $\mathbb{CP}^1$, its Hopf link on $S^3$, and its comparison with the null quadric of the split-complex numbers. The forms are from The Three Canonical Forms on $\mathbb{C}^2$; the subspaces and the Lagrangian lines are from The Subspaces of $\mathbb{C}^2$; the sphere and the Hopf fibration are from The Unit Sphere and the Hopf Fibration; the split-complex numbers are from Split-Complex Numbers.

The cone is a reducible quadric: two complex lines meeting at the origin. Its real locus is only the origin, but its real picture is the union of two real two-planes in $\mathbb{R}^4$. Its projective image is two points of $\mathbb{CP}^1$, and its intersection with $S^3$ is a pair of great circles forming the Hopf link.

The Bilinear Form and the Cone

The form $B$ has inertia $(2,2)$ over the reals, and its complex isotropic cone is the zero set of a reducible quadratic form.

Definition. The complex bilinear form is $$ B(Q_1,Q_2)=Q_{1\blacktriangle}Q_{2\blacktriangle}+Q_{1\blacktriangledown}Q_{2\blacktriangledown}, $$ and the isotropic cone is $\mathcal{N}=\{Q:B(Q,Q)=0\}$.

Proposition (the inertia). As a real bilinear form on $\mathbb{C}^2\cong\mathbb{R}^4$, the real part $$ \operatorname{Re}B(Q_1,Q_2)=q_{\blacktriangle}u_{\blacktriangle}-q'_{\blacktriangle}v_{\blacktriangle}+q_{\blacktriangledown}u_{\blacktriangledown}-q'_{\blacktriangledown}v_{\blacktriangledown} $$ has inertia $(2,2)$, and the imaginary part $$ \operatorname{Im}B(Q_1,Q_2)=q_{\blacktriangle}v_{\blacktriangle}+q'_{\blacktriangle}u_{\blacktriangle}+q_{\blacktriangledown}v_{\blacktriangledown}+q'_{\blacktriangledown}u_{\blacktriangledown} $$ also has inertia $(2,2)$. The complex form $B$ is nondegenerate, symmetric and $\mathbb{C}$-bilinear.

Proof. In the real coordinates $(q_{\blacktriangle},q'_{\blacktriangle},q_{\blacktriangledown},q'_{\blacktriangledown})$ the matrix of $\operatorname{Re}B$ is $\operatorname{diag}(1,-1,1,-1)$, of inertia $(2,2)$; the matrix of $\operatorname{Im}B$ is the symmetric matrix with blocks $\begin{pmatrix}0&1\\1&0\end{pmatrix}$, whose eigenvalues are $\pm1$. The complex form is nondegenerate because its matrix is the identity.

Proposition (the factorization). The cone is the union of the two complex lines $$ L_+=\{Q:Q_{\blacktriangle}=iQ_{\blacktriangledown}\},\qquad L_-=\{Q:Q_{\blacktriangle}=-iQ_{\blacktriangledown}\}. $$ The two lines are distinct, they meet only at the origin, and $\mathcal{N}=L_+\cup L_-$.

Proof. $Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=(Q_{\blacktriangle}+iQ_{\blacktriangledown})(Q_{\blacktriangle}-iQ_{\blacktriangledown})$, and $\mathbb{C}[Q_{\blacktriangle},Q_{\blacktriangledown}]$ is an integral domain, so the product vanishes exactly when one factor does. The lines $Q_{\blacktriangle}=iQ_{\blacktriangledown}$ and $Q_{\blacktriangle}=-iQ_{\blacktriangledown}$ are distinct and intersect at the origin.

Remark (the cone is singular). The cone is a complex algebraic hypersurface of degree $2$, and it is singular at the origin: the two lines meet there, and the Zariski tangent space at the origin is the whole of $\mathbb{C}^2$. The cone is not a complex manifold at the origin, and the two lines are its two irreducible components.

The Two Isotropic Lines

The two isotropic lines are complex lines, hence real two-planes; each is totally isotropic for $B$ and Lagrangian for the holomorphic symplectic form $\omega$.

Proposition (the lines are isotropic). For $Q_1\in L_\pm$ and $Q_2\in L_\pm$, $B(Q_1,Q_2)=0$. Hence each line is totally isotropic for $B$.

Proof. On $L_+$, $Q_{1\blacktriangle}=iQ_{1\blacktriangledown}$ and $Q_{2\blacktriangle}=iQ_{2\blacktriangledown}$, so $B(Q_1,Q_2)=Q_{1\blacktriangle}Q_{2\blacktriangle}+Q_{1\blacktriangledown}Q_{2\blacktriangledown}=(i^2+1)Q_{1\blacktriangledown}Q_{2\blacktriangledown}=0$. The same computation with $Q_{1\blacktriangle}=-iQ_{1\blacktriangledown}$ gives $L_-$.

Proposition (the lines are Lagrangian for $\omega$). Each complex line is Lagrangian for the holomorphic symplectic form $\omega$ of The Three Canonical Forms on $\mathbb{C}^2$: $\omega$ vanishes on it, and its real dimension is half the real dimension of $\mathbb{C}^2$.

Proof. The form $\omega$ is a $(2,0)$-form, and a $(2,0)$-form vanishes on every complex line because the line has only one holomorphic direction. The real dimension of a complex line is $2$, half of the real dimension $4$ of $\mathbb{C}^2$.

Remark (the contrast with the Hermitian form and with $\omega$). The Hermitian form $H$ is positive definite, so it has no isotropic vectors; the holomorphic symplectic form $\omega$ has every complex line as an isotropic subspace, and its isotropic vectors are exactly the points of those lines; the bilinear form $B$ has exactly two isotropic lines. The three forms therefore have three different isotropic geometries, and the cone of $B$ is the only one that is a proper algebraic subvariety. The Kähler form $\Omega$ of The Flat Kähler Structure of $\mathbb{C}^2$ is a fourth object: it has no isotropic vectors, and its Lagrangian planes are the totally real planes, not the complex lines.

Remark (the stabilizer of the cone). The complex orthogonal group $O(2,\mathbb{C})$ of The Classical Groups of $\mathbb{C}^2$ preserves $B$, hence preserves the cone and permutes its two lines. The subgroup preserving each line is the identity component $\mathbb{C}^{\times}$ of $O(2,\mathbb{C})$, and the full group exchanges the two lines through the swap of the two isotropic coordinates, an element of determinant $-1$.

The Real Picture

The real locus of the cone is only the origin, but the underlying real algebraic set is the union of two real two-planes in $\mathbb{R}^4$, and the cone is the intersection of the null cones of $\operatorname{Re}B$ and $\operatorname{Im}B$.

Proposition (the real locus). The real points of $\mathcal{N}$ are $$ \mathcal{N}\cap\mathbb{R}^2=\{0\}. $$

Proof. For real $Q_{\blacktriangle},Q_{\blacktriangledown}$, the equation $Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=0$ forces $Q_{\blacktriangle}=Q_{\blacktriangledown}=0$.

Proposition (the real picture). The complex lines $L_\pm$ are real two-planes in $\mathbb{R}^4$, and $$ \mathcal{N}=L_+\cup L_- $$ as a real algebraic set. The two planes meet only at the origin, and together they form a cone of real dimension $2$ inside $\mathbb{R}^4$.

Proof. A complex line is a real two-plane, because it is closed under multiplication by $i$. The union is the cone, and the intersection is the origin because the lines are distinct.

Proposition (the intersection of two null cones). The cone $\mathcal{N}$ is the intersection of the real null cone of $\operatorname{Re}B$ and the real null cone of $\operatorname{Im}B$: $$ \mathcal{N}=\{Q:\operatorname{Re}B(Q,Q)=0\}\cap\{Q:\operatorname{Im}B(Q,Q)=0\}. $$

Proof. $B(Q,Q)=0$ exactly when its real and imaginary parts vanish.

Remark (the real picture of the split-complex cone). The split-complex numbers have the norm $x^2-y^2$, whose null quadric is the union of the two real lines $x=\pm y$. The complex cone of $B$ is the complexification of that real quadric: after the complex change of variable $Q_{\blacktriangledown}\mapsto iQ_{\blacktriangledown}$, the equation $Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=0$ becomes $Q_{\blacktriangle}^2-Q_{\blacktriangledown}^2=0$, the complexification of $x^2-y^2=0$. The real picture gains one dimension: the two real lines become two real planes.

Remark (the cone and the complex structure). The cone is invariant under $J$, because $B(JQ_1,JQ_2)=-B(Q_1,Q_2)$ and the zero set is unchanged; each of the two lines is $J$-invariant, since it is a complex line. The cone is therefore a complex subvariety, not merely a real one, and its real picture is the union of the two $J$-invariant planes.

The Projective Image

The image of the cone in $\mathbb{CP}^1$ is the two-point set of the isotropic directions; the projective quadric is reducible and degenerate.

Proposition (the image). The projection $\mathbb{C}^2\setminus\{0\}\to\mathbb{CP}^1$ sends the cone to the two points $$ [i:1],\qquad [-i:1]. $$ The projective quadric $\{Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=0\}\subset\mathbb{CP}^1$ is exactly this two-point set.

Proof. A point of $L_+$ is $(iQ_{\blacktriangledown},Q_{\blacktriangledown})$ with $Q_{\blacktriangledown}\neq0$, whose class is $[i:1]$; a point of $L_-$ gives $[-i:1]$. The homogeneous equation $Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=0$ has the two solutions $[Q_{\blacktriangle}:Q_{\blacktriangledown}]=[\pm i:1]$.

Remark (the degeneracy). A nondegenerate conic in $\mathbb{CP}^2$ is a smooth curve of genus zero; the quadric of $B$ in $\mathbb{CP}^1$ is reducible, two points, because a binary quadratic form over $\mathbb{C}$ factors into two linear forms. The two points are the two isotropic directions, and in $\mathbb{CP}^1$ they are the whole projective image of the cone.

Remark (the two points and the two lines). The correspondence is the projectivization of the two isotropic lines: the cone is the union of the two lines through the origin over the two points, and its projectivization is the two-point set. The cone is the simplest example of a reducible projective quadric.

The intersection of the cone with the unit sphere is a pair of great circles, and the two circles are linked with linking number one: they form the Hopf link.

Proposition (the intersection with $S^3$). The intersection $\mathcal{N}\cap S^3$ is the union of the two circles $$ C_+=\{(iQ_{\blacktriangledown},Q_{\blacktriangledown}):\lvert Q_{\blacktriangledown}\rvert=1/\sqrt2\},\qquad C_-=\{(-iQ_{\blacktriangledown},Q_{\blacktriangledown}):\lvert Q_{\blacktriangledown}\rvert=1/\sqrt2\}. $$ Each circle is a fibre of the Hopf fibration $S^3\to\mathbb{CP}^1=S^2$, over the points $[i:1]$ and $[-i:1]$.

Proof. On $L_+$, a point is $(iQ_{\blacktriangledown},Q_{\blacktriangledown})$; its norm is $\sqrt{2}\lvert Q_{\blacktriangledown}\rvert$, so it lies on $S^3$ exactly when $\lvert Q_{\blacktriangledown}\rvert=1/\sqrt2$. The fibres of the Hopf fibration are the intersections of the complex lines through the origin with $S^3$, and the two lines are the fibres over the two projective points.

Theorem (the linking number). The two circles $C_+$ and $C_-$ have linking number $1$.

Proof. The linking number of two fibres of the Hopf fibration over distinct points of $S^2$ is $1$; this is proved in The Unit Sphere and the Hopf Fibration by projecting one fibre to the other and computing the intersection number, or by the Gauss integral.

Remark (the cone as a real surface). The intersection $\mathcal{N}\cap S^3$ is a pair of circles, and the cone itself is the union of the two planes. The Hopf link is the boundary of the cone at infinity, and it is the simplest nontrivial link: two circles that cannot be separated without crossing.

Remark (the isotropic lines and the Hopf fibration). Every complex line through the origin meets $S^3$ in a Hopf fibre, and the Hopf fibration is the projectivization of $\mathbb{C}^2\setminus\{0\}$. The isotropic lines are two special fibres, the fibres over the two points where the projective quadric meets the projective line; the other fibres are not isotropic for $B$.

The Split-Complex Comparison

The complex isotropic cone of $B$ is the complexification of the split-complex null quadric; the dual numbers give the degenerate case of a double line, and the three cases are the three real forms of the same quadratic equation.

Definition. The split-complex numbers are $\mathbb{R}[j]$ with $j^2=1$; the null quadric is the set of $x+yj$ with $x^2-y^2=0$.

Proposition (the four readings). The equation $u^2+v^2=0$ has four readings: $$ \begin{aligned} \text{real: } & u^2+v^2=0 \quad\text{(only the origin)},\\ \text{split-complex: } & u^2-y^2=0 \quad\text{(two real lines)},\\ \text{complex: } & Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=0 \quad\text{(two complex lines)},\\ \text{dual: } & u^2=0 \quad\text{(the axis, doubled)}. \end{aligned} $$ The split form is a real form of the complex form under the real structure $Q_{\blacktriangledown}=iy$; the definite form is the real form under $Q_{\blacktriangledown}=v$ real; the dual case is a degeneration, not a real form.

Proof. The dual numbers have $v^2=0$ and the equation reduces to $u^2=0$; the split-complex numbers have $j^2=1$ and the equation factors as $(u-y)(u+y)=0$; the complex numbers have $i^2=-1$ and the equation factors as $(Q_{\blacktriangle}+iQ_{\blacktriangledown})(Q_{\blacktriangle}-iQ_{\blacktriangledown})=0$. The change $Q_{\blacktriangledown}\mapsto iy$ carries the complex equation to $u^2-y^2=0$.

Remark (the real forms). The complex quadratic form $Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2$ has two real forms: the definite form $x_1^2+x_2^2$, which has no nonzero isotropic vectors, and the split form $x_1^2-y^2$, which has two isotropic lines. The complex cone is the complexification of the split cone, and the definite real form sees only the origin. The dual case is the degeneration in which the two lines coincide.

Remark (the comparison with the dual numbers). The dual numbers have $\epsilon^2=0$ and the null cone is the axis, counted twice. The complex cone is the split case, not the dual case: the equation is nondegenerate, and the cone has two distinct irreducible components. The dual case is the degeneration in which the two lines coincide; it is the subject of Dual Numbers and of The Dual-Numbers Null Quadric and Projective Geometry.

Summary

The complex bilinear form $B(Q_1,Q_2)=Q_{1\blacktriangle}Q_{2\blacktriangle}+Q_{1\blacktriangledown}Q_{2\blacktriangledown}$ has inertia $(2,2)$ over the reals, and its isotropic cone $\mathcal{N}=\{Q_{\blacktriangle}^2+Q_{\blacktriangledown}^2=0\}$ is the union of the two complex lines $Q_{\blacktriangle}=\pm iQ_{\blacktriangledown}$. The lines are totally isotropic for $B$ and Lagrangian for the holomorphic symplectic form $\omega$; the Hermitian form $H$ is definite and has no isotropic vectors, so the three forms have three different isotropic geometries.

The real locus of the cone is only the origin, but the cone as a real algebraic set is the union of two real two-planes in $\mathbb{R}^4$, and it is the intersection of the null cones of the real and imaginary parts of $B$. Its projective image is the two-point set $\{[i:1],[-i:1]\}\subset\mathbb{CP}^1$. Its intersection with $S^3$ is a pair of great circles, the Hopf fibres over the two points, and they form the Hopf link with linking number one. The cone is the complexification of the split-complex null quadric $x^2-y^2=0$; the dual numbers give the degenerate case of a double line.

Summary of Notation

Symbol Meaning
$B(Q_1,Q_2)=Q_{1\blacktriangle}Q_{2\blacktriangle}+Q_{1\blacktriangledown}Q_{2\blacktriangledown}$ the complex bilinear form
$H(Q_1,Q_2)=Q_{1\blacktriangle}\overline{Q_{2\blacktriangle}}+Q_{1\blacktriangledown}\overline{Q_{2\blacktriangledown}}$ the Hermitian form
$\omega$ the holomorphic symplectic form
$\mathcal{N}=\{B(Q,Q)=0\}$ the isotropic cone
$L_\pm=\{Q_{\blacktriangle}=\pm iQ_{\blacktriangledown}\}$ the two isotropic lines
$[i:1],[-i:1]$ the two isotropic directions in $\mathbb{CP}^1$
$C_\pm=\mathcal{N}\cap S^3\cap L_\pm$ the two Hopf fibres of the cone
$x^2-y^2=0$ the split-complex null quadric
$u^2=0$ the dual-number null quadric

Further Reading