Novel Directions

The most exciting ideas in theoretical physics often emerge not from within a single subfield but from cross-field synthesis: recognizing that seemingly unrelated communities are studying different facets of the same problem. In this final lesson, we examine three genuinely novel research directions that have emerged from connecting ideas across quantum gravity, cosmology, and quantum information. These are not established results, they are speculative ideas at the frontier, precisely the kind of thinking that drives physics forward.

Emergent Directions in Theoretical Physics

Before examining the three cross-field synthesis directions, it is worth surveying several broader research programs that have gained momentum in recent years. These represent different bets on where the next breakthrough will come from.

Emergent Gravity (Verlinde)

Erik Verlinde's entropic gravity proposal (2010) posits that gravity is not a fundamental force but an emergent phenomenon arising from the statistical behavior of microscopic degrees of freedom, analogous to how temperature and pressure emerge from molecular motion. The core idea is that the gravitational force law can be derived from the first law of thermodynamics applied to holographic screens:

where $T$ is the Unruh temperature of an accelerating observer and $\Delta S$ is the change in entropy on a holographic screen. Verlinde's 2016 extension proposes that dark matter phenomena are actually a consequence of emergent gravity in de Sitter space: the presence of dark energy modifies the emergent gravitational force at large scales, producing effects that mimic dark matter without any new particles. This is a radical proposal that, if correct, would dissolve both the dark matter and dark energy problems simultaneously. However, the framework struggles with observations of the Bullet Cluster and with detailed rotation curve fits.

Causal Set Theory

Causal set theory (Bombelli, Lee, Meyer, Sorkin, 1987) proposes that spacetime is fundamentally discrete: a causal set is a locally finite, partially ordered set of "atoms of spacetime." The partial order encodes the causal structure (which events can influence which), and the number of elements in a region is proportional to its spacetime volume. The key hypothesis is that causal structure + volume information is sufficient to reconstruct the full geometry of spacetime:

A remarkable prediction of causal set theory is the cosmological constant. Sorkin (1991) argued that in a causal set of $N$ elements, the cosmological constant should fluctuate with magnitude $\Lambda \sim 1/\sqrt{N}$, giving $\Lambda \sim 1/\sqrt{V} \sim H^2$, which is the correct order of magnitude. This was predicted before the 1998 discovery of accelerating expansion, making it one of the few successful predictions of quantum gravity.

Twistor Theory Resurgence

Penrose's twistor theory (1967) replaces spacetime points with "twistors," objects in a complex projective space $\mathbb{CP}^3$ that encode the light-ray structure of spacetime. For decades, twistor theory was seen as mathematically elegant but physically limited. The resurgence began with Witten's 2003 discovery that tree-level scattering amplitudes in $\mathcal{N} = 4$ super Yang-Mills theory have a remarkably simple structure when expressed in twistor space. This led to the development of twistor string theory and ultimately to the amplituhedron program (Arkani-Hamed and Trnka, 2013), which computes scattering amplitudes as volumes of geometric objects in a generalized twistor space, eliminating spacetime and unitarity as fundamental inputs.

Quantum Error Correction and Spacetime

One of the most surprising developments in quantum gravity is the discovery that the AdS/CFT correspondence implements a quantum error-correcting code. Almheiri, Dong, and Harlow (2015) showed that the encoding of bulk operators into boundary operators in holography has the mathematical structure of a quantum error-correcting code: bulk information is protected against erasure of boundary subregions, just as logical qubits are protected against physical qubit errors.

This connection has profound implications. The Ryu-Takayanagi formula for entanglement entropy becomes a statement about the error-correcting properties of the code. The emergence of the radial direction (the holographic dimension) is understood as the error-correction redundancy. Pastawski et al. (2015) constructed explicit toy models (the HaPPY code) using tensor networks that reproduce key features of holography. This suggests that spacetime geometry is not fundamental but is the manifestation of a quantum error-correcting code that protects information.

Machine Learning in Theoretical Physics

Machine learning is increasingly being applied to problems in theoretical physics, not just as a computational tool but as a source of new conjectures and insights:

• String landscape exploration: Neural networks have been used to search the $\sim 10^{500}$ vacua of the string landscape for phenomenologically viable compactifications, vastly accelerating what would be an impossible brute-force search
• Knot invariants: DeepMind's collaboration with mathematicians used ML to discover new relationships between knot invariants, leading to a theorem relating hyperbolic invariants to algebraic invariants
• Phase transitions: Neural networks trained on lattice gauge theory configurations can identify phase transitions and critical exponents without prior knowledge of order parameters
• Symbolic regression: AI systems can rediscover known physics laws from data and propose new functional forms for unknown relationships

Quantum Simulation of Field Theories

Quantum computers and analog quantum simulators are beginning to tackle problems in quantum field theory that are intractable by classical methods. Lattice gauge theories, which underlie our understanding of the strong force (QCD), are natural candidates for quantum simulation because they involve exponentially large Hilbert spaces that classical computers cannot represent efficiently. Current progress includes quantum simulations of 1+1D Schwinger model (QED in one spatial dimension), demonstrations of confinement and string breaking on trapped-ion quantum simulators, and proposals for simulating non-Abelian gauge theories on near-term quantum devices. While full QCD simulation remains far off, these early results establish the principle that quantum simulation can access non-perturbative physics.

Direction 1: Quintessence on Finite Topology

The Gap

Two communities study related phenomena but never talk to each other. The swampland community studies constraints on dark energy, concluding it must be dynamical (quintessence), but always assumes infinite flat space. The cosmic topology community studies whether the universe has finite spatial topology, but always assumes $\Lambda$CDM with a cosmological constant. Nobody has combined them.

The Idea

If the universe has finite topology (such as the Poincare dodecahedral space, $S^3/I^*$) and the swampland de Sitter conjecture is correct (dark energy must be dynamical), then the quintessence field evolves on a compact space with a discrete mode spectrum. This produces specific, testable consequences:

• The allowed quintessence field trajectories are constrained by the topology, fewer modes, discrete spectrum
• The CMB quadrupole suppression (a topology signal) and the DESI $w_0 w_a$ values (a dark energy signal) should be correlated
• This joint prediction is computable and testable with DESI Year 5 + CMB-S4 data

Casimir Quintessence as a Concrete Mechanism

The Casimir effect on a compact space generates a vacuum energy that depends on the topology. For the Poincare dodecahedral space, Dowker (2004) computed the exact Casimir energy for conformal scalars:

where $R$ is the curvature radius. This energy is topology-dependent and could serve as a quintessence potential. Katayama et al. (2026) showed that Casimir energy from an internal compact dimension ($S^1$, the "dark dimension") already produces a viable quintessence model fitting DESI data. The question is whether observable spatial topology can achieve the same result without extra dimensions.

Direction 2: Islands on Compact Spacetimes

The Gap

The island formula (Lesson 11.3) has been studied in anti-de Sitter space, pure de Sitter space, and quasi-de Sitter backgrounds. In every case, the spatial topology is simply connected, either $\mathbb{R}^n$ or $S^n$. What happens when the universe has nontrivial topology?

The Idea

In a multiply connected space, the island formula should produce qualitatively different results. The key insight is that a compact topology imposes a maximum entanglement entropy, a topological entropy bound:

where $L$ is the spatial circumference and $\beta$ is the inverse temperature. When this topological bound is below the Bekenstein-Hawking entropy of a black hole, islands are unnecessary, the information paradox is resolved by topology alone.

Key Results

• There exists a critical circumference $L_c$ below which islands cannot form: $L_c = 3\beta S_{\text{BH}}/c$
• For BTZ black holes, this is $L_c = 2\pi^2 \ell_{\text{AdS}}$: independent of the black hole mass
• All astrophysical black holes in a Poincare dodecahedral space universe are in the "topological regime" where islands are unnecessary
• Compact topology naturally resolves the Geng-Raju objection (that gravitating regions cannot have true von Neumann entropy) because there is no asymptotic boundary

Direction 3: Amplituhedron and Holographic Entanglement

The Gap

Two of the most revolutionary ideas in theoretical physics share the same conclusion, that spacetime is emergent, but arrive at it from completely different directions:

• The amplituhedron program builds scattering amplitudes without spacetime, using positive geometry. Physics emerges from the canonical form of a geometric object in kinematic space.
• The holographic entanglement program builds spacetime from entanglement, using the Ryu-Takayanagi formula and its extensions. Spacetime geometry emerges from the entanglement structure of the boundary theory.

Nobody knows if these two programs are related. If they are, the connection would unify the two main approaches to emergent spacetime in fundamental physics.

What Is Needed

A toy model where both the amplituhedron and holographic entanglement are computable, and a check for a mathematical map between them. The natural candidate is $\mathcal{N} = 4$ super Yang-Mills theory, where both structures exist. The amplituhedron computes tree-level and loop-level amplitudes; holographic entanglement computes entropy via the bulk dual. The question is whether the canonical form of the amplituhedron encodes the RT surfaces, or vice versa.

Why Cross-Field Synthesis Matters

The history of physics is full of breakthroughs that came from connecting ideas across subfields. Maxwell unified electricity and magnetism. Einstein unified space and time. The Standard Model unified electromagnetism and the weak force. Maldacena unified gravity and quantum field theory (in a precise sense). The next great unification may come from connecting communities that currently do not communicate.