In three dimensions you have three independent directions: $x$, $y$, $z$. A fourth dimension is simply a fourth independent direction — call it $w$ — such that no combination of $x$, $y$, $z$ can reach it. It is not "time" (though time can play this role). It is not a hidden spatial direction too small to see. It is a direction that is genuinely orthogonal to all three directions you already have.
Your 3D intuition will resist this. Every visualisation you attempt will project the fourth dimension back onto three. That projection is not a failure — it is the method. The rotating tesseract below shows all sixteen corners of the 4D hypercube, but always as a projection. Learning to read what the projection distorts is the first skill of 4D geometry.
$\mathbb{R}^4 = \{(x_1, x_2, x_3, x_4) : x_i \in \mathbb{R}\}$. The unit 4-cube (tesseract) is $[0,1]^4 \subset \mathbb{R}^4$. It has $2^4 = 16$ vertices, $4 \cdot 2^3 = 32$ edges, $6 \cdot 2^2 = 24$ square faces, and $2 \cdot 4 = 8$ cubic cells. Each cubic cell is a standard 3-cube — a copy of the space from Chapters 1–3.
| Dimension | Object | Vertices | Edges | Faces | Cells |
|---|---|---|---|---|---|
| 1D — line | segment | 2 | 1 | — | — |
| 2D — plane | square | 4 | 4 | 1 | — |
| 3D — space | cube | 8 | 12 | 6 | 1 |
| 4D — hyperspace | tesseract | 16 | 32 | 24 | 8 |
The simulation below projects the tesseract from $\mathbb{R}^4$ into $\mathbb{R}^3$, then onto your screen. Two independent rotations are active: one in the $xy$-plane (ordinary 3D rotation) and one in the $xw$-plane (the 4D rotation that moves the $w$ coordinate). Drag to control the 3D rotation. Use the slider to dial in the 4D rotation. Watch the inner cube appear to pass through the outer cube — that is not a visual glitch; in 4D those cells never intersect.
In §3.9 we established that the dm³ contact manifold near its attractor is locally the 3-sphere $S^3 \subset \mathbb{R}^4$. Now we can see what that means geometrically: $S^3$ is the set of points at unit distance from the origin in $\mathbb{R}^4$. It is a hypersurface — a 3-dimensional boundary of the 4-dimensional ball $B^4$.
This is the exact analogue of how $S^2$ (the ordinary sphere) is the boundary of the 3-ball $B^3$. You can visualize $S^2$ by imagining the surface of a globe — all points at distance 1 from the centre in $\mathbb{R}^3$. $S^3$ is one dimension up: all points at distance 1 from the origin in $\mathbb{R}^4$.
When the tesseract rotates in the $xw$-plane, what you see is a 3D object (a cube or distorted polyhedron) passing through its mirror image. That passage — through and out the other side — is what $S^3$ does continuously. The 3-sphere cannot be embedded in 3D without self-intersection; in 4D it sits cleanly, unobstructed.
§ 4.4The contact structure on $S^3$ has a natural 4D companion — the symplectic structure. Given the contact form $\alpha$ on $S^3$, define the symplectisation as the 4-manifold:
The symplectisation is not just a product space with a structure bolted on — it is the canonical 4-manifold associated to any contact 3-manifold, encoding all the geometry of $\alpha$ in a 2-form $\omega$ that is easier to work with analytically. The contact form $\alpha$ on $S^3$ becomes the symplectic form $\omega = d\alpha$ in the tangent directions, plus the new $ds$ direction.
A contact structure on a 3-manifold $M$ is a maximally non-integrable 2-plane field. The symplectic form $\omega$ on $M \times \mathbb{R}$ is a 2-form — it measures areas of 2-dimensional surfaces. In 4D, holomorphic curves (complex 1-dimensional surfaces, real 2-dimensional) are the natural objects that $\omega$ integrates over. This is why contact topology problems — like Weinstein's conjecture, proved in 3D — require 4D techniques: holomorphic curves in the symplectisation.
Chapter 3's dm³ ODE had coordinates $(r, \theta, z)$. The 4D lift adds a coordinate $w$ governed by the contact structure. The extended system is:
The key question is whether the helical attractor $\{r=1\}$ persists when we allow the system to move in the $w$-direction. The answer is yes — and the reason is a theorem in symplectic topology, not just a calculation.
The closed Reeb orbit $\gamma^* = \{r=1, \theta \in [0,2\pi), z = z_0, w = 0\}$ of System (3.1) persists as an attractor of the 4D system (4.1). Its Lyapunov exponents in the $w$-direction are determined by the symplectic area $\int_D \omega$ of a disk $D$ bounded by $\gamma^*$ in the symplectisation. For the dm³ contact form, this area equals $2\pi$, and the $w$-direction exponent is bounded by $-2$.
The $2\pi$ that appears in the area integral is the same $T^* = 2\pi$ that appeared as the Reeb period in Chapter 3. The period, the area, and the Lyapunov rate are all expressions of the single number $\pi$ that anchors the dm³ system — each a different face of the same geometric object.
§ 4.6In 3D, the tools are vector fields, flows, and fixed-point theorems. In 4D, the native tool is the holomorphic curve — a map $u: \Sigma \to W$ from a Riemann surface $\Sigma$ into the symplectisation $W = S^3 \times \mathbb{R}$ satisfying the Cauchy-Riemann equation:
Gromov's 1985 paper introducing pseudo-holomorphic curves transformed symplectic topology. The key insight: holomorphic curves have a compactness property controlled by the symplectic area $\int_\Sigma u^*\omega$. If you have a sequence of holomorphic curves with bounded area, a subsequence converges — possibly to a broken curve whose pieces are asymptotic to Reeb orbits. This is Gromov compactness, and it is why the closed Reeb orbits (like the dm³ attractor) are not just locally stable but globally detected.
In 3D, Theorem 3.2 (attractor convergence) was proved by direct ODE estimates — Gronwall's inequality, Lyapunov functions, the DOP853 numerical correction. Those tools are local: they tell you what happens near $r=1$. Holomorphic curves in the 4D symplectisation are global: they detect $\gamma^*$ from anywhere in the manifold, not just from nearby initial conditions. The 4D lift turns a local convergence theorem into a global topological statement.
Return to the rotating tesseract. Its eight cubic cells correspond to eight "faces" of the 4D arena. The contact manifold $S^3$ is the diagonal cross-section — the set of points equidistant from all eight cells. When you watch the inner cube pass through the outer cube during 4D rotation, you are watching a contact transformation: the Reeb flow carries points on $S^3$ along the fibres of the Hopf fibration, and the 4D rotation is a visual proxy for that flow.
The gold edges in the simulation hold fixed in 3D. The green edges rotate in the $xw$-plane. The moment of "passing through" — when a green vertex coincides with a gold edge — corresponds to a point on $S^3$ where the $w$-coordinate passes through zero. That locus is exactly the 3D contact manifold of Chapter 3, now visible as a cross-section of the 4D object.
1D: fermion seed $\psi^2 = 0$, pure potential.
2D+t: extended phase space, contact form $\alpha = dy - f\,dx$ born.
3D: contact 3-manifold, Reeb flow, helical attractor $T^* = 2\pi$, $S^3$ identified.
4D: $\mathbb{R}^4$ entered, $S^3$ as hypersurface, symplectisation $\omega = d\alpha$, holomorphic curves appear.
Next: Chapter 5 lifts to 5D — jet space $J^1(\mathbb{R}, \mathbb{R}^2)$, where prolongation carries the ODE into a 5-dimensional contact structure.