§1The Gap
Two bodies of literature exist and do not speak to each other.
The first is special relativistic causality, centered on the light cone and the frame-dependence of simultaneity. Its most striking formulation for cosmological distances is the Andromeda paradox (Penrose 1960, later developed in Penrose 2004): two people walking past each other on Earth in opposite directions assign "now in Andromeda" to events in M31 separated by approximately 80,000 years. Neither is wrong. Simultaneity at spacelike separation is a choice of reference frame, not a physical fact. The standard response in the literature is that this is simply the correct behavior of Minkowski spacetime and requires no further resolution.
The second is galaxy merger astrophysics. The Milky Way and Andromeda are on a collision course; first pericentre passage is predicted at approximately 3.9 Gyr from now, with full merger at approximately 6 Gyr, producing a single elliptical galaxy (van der Marel et al. 2012). This literature is rich in N-body dynamics, stellar population synthesis, chemical evolution, and black hole co-evolution. It does not contain the word "simultaneity" or engage with SR causal structure.
The Gap · Formally Stated
No paper in either literature identifies the Milky Way–Andromeda merger as a causal integration event in the SR sense: the earliest physical event whose causal past encompasses the complete history of both systems. No paper defines a "causal integration time" for gravitationally bound systems. No paper distinguishes between the geometric moment when future light cones first intersect and the physical moment when systems merge into a single causal unit. The Andromeda paradox is treated as requiring philosophical or conventional resolution; its physical resolution — the merger — is not identified as such.
This paper names the gap, defines the concepts, states the theorems, applies them to Milkomeda, and notes the connection to the contact geometry framework of the Principia Orthogona series. [ORIGINAL — no prior source found as of August 2026]
§2Definitions
We work in Minkowski spacetime (special relativity; GR corrections are noted where relevant). Signature $(-,+,+,+)$, $c = 1$ (natural units).
Definition 2.1 · Causal Future and Causal Past
The causal future $J^+(E)$ of an event $E$ is the set of all events reachable from $E$ by a signal at speed $\leq c$ (i.e., all events $M$ with $ds^2(E \to M) \leq 0$ and $M$ in the future of $E$). The causal past $J^-(M)$ of $M$ is the set of all events that can send a signal to $M$. Both $J^+(E)$ and $J^-(M)$ are Lorentz-invariant: all inertial observers agree on their membership.
Definition 2.2 · System History
The history $\mathcal{H}(A)$ of a physical system $A$ (e.g., a galaxy) is the set of all events that occur within $A$ from its formation to the present. For the Milky Way, $\mathcal{H}(\text{MW})$ includes all events in the galaxy from $\sim 13.6$ Gyr ago to today. For Andromeda, $\mathcal{H}(\text{And})$ includes all events in M31 from its formation to today.
Definition 2.3 · Causal Integration Event
An event $M$ is a causal integration event for systems $A$ and $B$ if:
(i) $\mathcal{H}(A) \subset J^-(M)$ — the entire history of $A$ is in the causal past of $M$, and
(ii) $\mathcal{H}(B) \subset J^-(M)$ — the entire history of $B$ is in the causal past of $M$.
That is: $M$ is an event from which, in principle, complete information about both $A$ and $B$ could have been received.
Definition 2.4 · Causal Integration Time
The causal integration time $T_{\text{CI}}(A, B)$ is the infimum of proper times $\tau(M)$ over all causal integration events $M$ for systems $A$ and $B$, where $\tau(M)$ is measured from the earlier of the two systems' formation events along a reference worldline (e.g., the cosmic time of a comoving observer).
If no causal integration event exists, $T_{\text{CI}}(A,B) = \infty$.
§3Two Regimes of Causal Integration
Theorem 3.1 · Geometric vs. Physical Causal Integration · [Original]
For two systems $A$ and $B$ separated today by distance $d$ (in some comoving frame), there are two distinct causal integration regimes:
(a) Geometric causal integration at $T_{\text{geo}} \approx d/c$: the first moment at which the future light cones $J^+(A_0)$ and $J^+(B_0)$ (where $A_0, B_0$ are present events in each system) have a non-empty intersection. Any event $M$ in $J^+(A_0) \cap J^+(B_0)$ is a causal integration event for the specific events $A_0$ and $B_0$, but NOT for the complete histories $\mathcal{H}(A)$ and $\mathcal{H}(B)$: events deep in the history of $A$ and $B$ (far from the boundary visible at time $T_{\text{geo}}$) are still outside $J^-(M)$.
(b) Physical causal integration at $T_{\text{phys}}$: if systems $A$ and $B$ are gravitationally bound and will merge at time $T_{\text{merge}}$, then for any event $M$ in the merged system at time $T \gg T_{\text{merge}}$, we have $\mathcal{H}(A) \subset J^-(M)$ and $\mathcal{H}(B) \subset J^-(M)$. The merger creates a single physical object whose every future event has complete access (in principle) to both systems' histories.
$T_{\text{geo}} \ll T_{\text{phys}}$ in general. For the Milky Way–Andromeda pair: $T_{\text{geo}} \approx 2.537 \text{ Myr}$, $T_{\text{phys}} \approx 4.5\text{–}6 \text{ Gyr}$.
Theorem 3.2 · Permanent Separation · [VERIFIED — standard cosmology; Davis & Lineweaver 2004]
For two systems $A$ and $B$ separated by a distance exceeding the cosmological event horizon ($d \gtrsim 16$ Gly in the current $\Lambda$CDM model), the accelerating expansion of the universe ensures that no future event $M$ lies in $J^+(A_0) \cap J^+(B_0)$. The future light cones never intersect. $T_{\text{CI}}(A,B) = \infty$. No causal integration event exists — not geometric, not physical. These systems are permanently and irrevocably causally separated.
Consequence: the observable universe divides sharply into two classes — systems that will causally integrate (gravitationally bound structures: Local Group, galaxy clusters) and systems that are permanently causally separated (galaxies beyond the event horizon, receding at effective $v > c$). The Milky Way and Andromeda belong to the first class. [VERIFIED]
§4The Andromeda Paradox · What It Resolves and What It Does Not
Penrose's Andromeda paradox states: two observers walking past each other on Earth at $v \approx 5$ km/h have simultaneity surfaces that differ by $\sim 40{,}000$ years in Andromeda ($\Delta t = v \Delta x / c^2 \approx (5\text{ km/h}/c) \times 2.537 \times 10^6 \text{ yr}$). Neither is wrong. Simultaneity is frame-dependent.
The paradox is typically presented as requiring philosophical acceptance — there is no "true" now in Andromeda, only frame-relative nows. The standard resolution is: accept frame-dependence, move on.
Observation 4.1 · The Physical Resolution · [Original]
The Andromeda paradox admits a frame-independent complement that has not been identified in the literature. While the question "what is happening in Andromeda right now" has no frame-independent answer (simultaneity is conventional), the question "when can a single physical observer have complete information about both galaxies' entire histories" has a frame-independent answer: after $T_{\text{CI}}(\text{MW}, \text{And})$.
This does not resolve the simultaneity question — simultaneity remains frame-dependent even after the merger. What it resolves is the causal encompassment question: the moment when the distinction "this happened in the Milky Way" vs. "this happened in Andromeda" ceases to be a distinction in causal access. After Milkomeda, a star born in the merged galaxy has both galaxies' complete histories equally in its causal past — not by convention, not by choice of frame, but by the Lorentz-invariant structure of $J^-(M)$.
Concretely: a Milkomeda star at cosmic time $t = 20$ Gyr is in the causal future of every event in both galaxies' histories (since both histories lie within 13.8 Gyr of formation, and the merger at $\sim 6$ Gyr has already occurred). The star's past light cone encompasses both. This is frame-independent — all observers agree that $\mathcal{H}(\text{MW}) \subset J^-(\text{star})$ and $\mathcal{H}(\text{And}) \subset J^-(\text{star})$.
§5The Milkomeda Case
The Milky Way–Andromeda pair is the first known concrete astrophysical case of a system with a finite, measurable physical causal integration time.
Now
(cosmic time ~13.8 Gyr)
$d(\text{MW}, \text{And}) \approx 2.537$ Mly. Approach velocity $\approx 110$ km/s (van der Marel et al. 2012). MW and Andromeda are spacelike separated: events "now" in each are outside the other's light cone. "Now in Andromeda" is frame-dependent by $\sim 40{,}000$ years per 5 km/h of relative motion.
+2.537 Myr
Geometric CI
First geometric causal integration: a signal from Earth today reaches Andromeda. Future light cones of today's Earth and today's Andromeda first intersect. Any event in this intersection is a causal integration event for the specific events $A_0$ and $B_0$ — but not for the full histories $\mathcal{H}(\text{MW})$ and $\mathcal{H}(\text{And})$. Events from billions of years deep in each galaxy's history are still outside any intersection.
+3.9 Gyr
First pericentre
First close passage. Tidal distortions, starburst ignition, galactic bars destabilize. Not yet the causal integration event — the galaxies separate after first passage and require multiple orbits before final merger.
+4.5–6 Gyr
Milkomeda
Physical CI
Physical causal integration. The merger is complete; Milkomeda is a single elliptical galaxy. Every event in Milkomeda from this point forward has $\mathcal{H}(\text{MW}) \subset J^-(M)$ and $\mathcal{H}(\text{And}) \subset J^-(M)$ — in the limit of sufficient elapsed time for light from the farthest reaches of both galaxies to have arrived. This is the causal integration event: the first moment at which a single physical system has both histories in its causal past. $T_{\text{CI}}(\text{MW}, \text{And}) \approx 4.5\text{–}6$ Gyr.
≫ Milkomeda
Far future
All Milkomeda stars have complete access (in principle) to both galaxies' histories. The question "did this happen in the Milky Way or Andromeda" is a historical question, not a causal access question. Both are equally in the past. The distinction has dissolved — not by convention, but by causal geometry.
Right now, "what is happening in Andromeda" has no single answer. In 5 billion years, that question has one answer — because by then Andromeda does not exist as a separate causal system. Its history and ours share the same past. Not because we chose the same reference frame. Because the light cones closed.
Spacelike today. Timelike forever. Milkomeda is the "when" that includes both.
§6Connection to the Contact Geometry Framework
Within the Principia Orthogona series, each galaxy is identified as a helical attractor on the contact 3-manifold $(\mathbb{R}^3, \alpha = dz - r^2 d\theta)$ at the g⁹⁶ scale (chGravity-scales.html): the spiral arm is the Reeb flow, the galactic plane is the contact surface $dz = r^2 d\theta$, the bulge is the inner basin ($r < r^*$), and the disc is the outer basin ($r > r^*$). This identification is marked [CONJECTURE] in the series.
Observation 6.1 · Milkomeda as fix(G_LocalGroup) · [Original — builds on CONJECTURE in chGravity-scales]
Under the g⁹⁶ attractor identification, the Milky Way and Andromeda are two separate helical attractors $\Gamma_{\text{MW}}$ and $\Gamma_{\text{And}}$ on the contact 3-manifold. The merger is the event at which the outer basins of both attractors overlap and the two limit cycles fold onto a single attractor $\Gamma_{\text{Milkomeda}}$.
In the language of the dm³ G-chain: $G = U \circ F \circ K \circ C$. The Local Group's gravitational dynamics is the G-chain applied to the two-attractor system. The fixed point is:
$\text{fix}(G_{\text{Local Group}}) = \Gamma_{\text{Milkomeda}}$
The fold event (the merger) is an $A_1$ Whitney singularity at galactic scale — the same geometric object as the levitation gap at $r^* = 0.77594058$, magnified by $\sim 10^{25}$ in physical scale (identified in chGravity-scales.html; status: CONJECTURE).
The causal integration time $T_{\text{CI}}$ is the proper time at which $G_{\text{Local Group}}$ converges to its fixed point. Causal integration and dynamical convergence coincide: the merger IS the fixed point, and the fixed point IS the causal integration event. [ORIGINAL — builds on CONJECTURE; status: CONJECTURE]
§7Open Problems
Open Problems · [OPEN]
O1 · GR corrections to $T_{\text{CI}}$. The definition of $T_{\text{CI}}$ above uses flat Minkowski spacetime. The actual merger occurs in curved spacetime ($g_{\mu\nu}$ from the Local Group mass distribution). How do GR corrections alter $T_{\text{CI}}$? Does the merger of two Schwarzschild-like potentials introduce corrections at the percent level or are they negligible for the causal integration question?
O2 · $T_{\text{CI}}$ for the full Local Group. The Local Group contains $\sim 54$ known galaxies, including M33 (Triangulum), the Large and Small Magellanic Clouds, and numerous dwarf galaxies. Each pair has its own $T_{\text{CI}}$. What is the causal integration time for the full Local Group — the moment at which all $\sim 54$ members' histories are in the causal past of a single event? Is it Milkomeda, or does the Local Group never fully merge?
O3 · The causal integration boundary in cosmological simulations. Large-scale structure simulations (IllustrisTNG, EAGLE, etc.) track merger trees of galaxies. $T_{\text{CI}}$ could be computed from these merger trees as a post-processing step. Does $T_{\text{CI}}$ correlate with any existing astrophysical measure (merger timescale, dynamical friction time, etc.)?
O4 · The g⁹⁶ fold radius for galactic attractors. The Whitney $A_1$ fold radius $r^*$ is known precisely in the LAW3M framework ($r^* = 0.77594058$). The analogous fold radius for the galactic-scale attractor — the Milkomeda $r^*_{\text{galactic}}$ — has no closed-form expression. Deriving it from the contact geometry would make the CONJECTURE in chGravity-scales.html a theorem.
First derivation attempt: WP58 (August 2026). Result: $r^*(\varepsilon_{\rm gal})$ where $\varepsilon_{\rm gal} = \nu/\kappa$; for the MW, $r^*_{\rm gal} \approx 0.713$, $R^*_{\rm phys} \approx 2.5\,\rm kpc$; exact universality $r^*=0.77594058$ iff $\nu = 2\kappa$ at $R_{\rm disc}$ (2:1 resonance). Identification unproved — see O.4.2 in WP58.
O5 · Observational signature of $T_{\text{CI}}$. Is there an observable difference between "before geometric causal integration" and "after geometric causal integration" that a hypothetical observer in the future (or, via light from the past, today) could detect? What is the observational signature of a causal integration event?
[OPEN — all five problems above are original; none appears in the literature as of August 2026]
References
Penrose, R. (1960).
The apparent shape of a relativistically moving sphere. Proc. Camb. Phil. Soc. 55, 137. · Penrose, R. (2004).
The Road to Reality. Jonathan Cape. [Andromeda paradox, ch. 17]
van der Marel, R.P. et al. (2012).
The M31 Velocity Vector. II. Radial Orbit toward the Milky Way. ApJ 753, 8. DOI:
10.1088/0004-637X/753/1/8 [merger timeline: first pericentre 3.87±0.16 Gyr, coalescence 5.86±0.72 Gyr]
Davis, T.M. & Lineweaver, C.H. (2004).
Expanding Confusion: Common Misconceptions of Cosmological Horizons. PASA 21, 97. DOI:
10.1071/AS03040 [cosmological event horizon ~16 Gly]
Grossi, P.N. (2026). chGravity-scales.html — Milkomeda as g⁹⁶ attractor merger, Whitney fold at galactic scale [CONJECTURE]. totogt.github.io/geometry.
Grossi, P.N. (2026). ch7-crystalline.html — Milkomeda as fix(G) of the Local Group. totogt.github.io/geometry.
Grossi, P.N. (2026). WP56 — Special Relativity as Contact Geometry, §9 (causality and the light cone), §10 (Milkomeda, initial statement). totogt.github.io/geometry/book7/wp56-special-relativity.html.