Limitations of Self-Assembly at Temperature 1

We prove that if a set $X \subseteq \Z^2$ weakly self-assembles at temperature 1 in a deterministic tile assembly system satisfying a natural condition known as \emph{pumpability}, then $X$ is a finite union of semi-doubly periodic sets. This shows that only the most simple of infinite shapes and patterns can be constructed using pumpable temperature 1 tile assembly systems, and gives evidence for the thesis that temperature 2 or higher is required to carry out general-purpose computation in a tile assembly system. Finally, we show that general-purpose computation \emph{is} possible at temperature 1 if negative glue strengths are allowed in the tile assembly model.

💡 Research Summary

The paper investigates the expressive power of deterministic tile assembly systems (TAS) operating at temperature 1, a regime in which a tile can attach to an existing assembly whenever a single glue match supplies enough binding strength. The authors introduce a natural structural constraint called “pumpability,” which requires that any infinite growth path in an assembly contains a repeatable segment that can be “pumped” (repeated) arbitrarily many times without violating the assembly rules. Pumpability captures the intuition that, in a physically realizable system, tiles can be replicated indefinitely along a growth direction.

Under the pumpability assumption, the authors prove a striking limitation: any set (X \subseteq \mathbb{Z}^2) that weakly self‑assembles at temperature 1 must be a finite union of semi‑doubly periodic sets. A semi‑doubly periodic set is defined by two independent period vectors (v_1, v_2) and a base point (p), taking the form ({p + i v_1 + j v_2 \mid i, j \in \mathbb{N}}). In other words, the only infinite shapes that can be produced are essentially rectangular lattice patterns (or finite unions thereof). Complex aperiodic structures, fractal patterns, or shapes that require branching and signal propagation cannot arise under these conditions.

The proof proceeds by analyzing the combinatorial constraints imposed by temperature 1 binding. Because a tile needs only a single matching glue, there is no way to enforce conditional attachment based on multiple glues, a mechanism that is crucial for implementing logical gates and information flow at higher temperatures. The authors show that any infinite assembly must eventually settle into a periodic “strip” that repeats indefinitely, and that any deviation from this strip would either violate pumpability or create a dead‑end that prevents further growth. By systematically covering all possible growth configurations, they demonstrate that the resulting assembled set must be a finite union of the aforementioned periodic strips.

To contrast this limitation, the paper explores an extension of the standard abstract Tile Assembly Model (aTAM) that allows negative glue strengths. A negative glue reduces the total binding strength when two tiles meet, effectively enabling a tile to detach or to prevent the attachment of a neighboring tile. By carefully designing tiles with both positive and negative glues, the authors construct a temperature‑1 system that simulates a Turing machine. The construction uses negative glues to implement conditional logic: a tile can only attach if a specific configuration of neighboring tiles is present, mimicking the cooperative binding that normally requires temperature 2. This demonstrates that the expressive weakness of temperature 1 is not inherent to the temperature itself but rather to the restriction that only non‑negative glues are permitted.

The implications of these results are twofold. First, they provide rigorous evidence supporting the prevailing belief that temperature 2 (or higher) is necessary for universal computation in self‑assembly, because only at higher temperatures can one enforce the multi‑glue cooperation needed for arbitrary logical operations and signal routing. Second, they open a new line of inquiry into how augmenting the glue model—by allowing negative interactions, reversible binding, or other non‑standard physical effects—might restore computational universality even at the lowest temperature. This is particularly relevant for experimental implementations using DNA tiles or other molecular components, where engineering negative interactions (e.g., via strand displacement or competitive binding) may be feasible.

In summary, the paper establishes a clear boundary: pumpable temperature‑1 TAS can only generate simple, periodic infinite patterns, ruling out general‑purpose computation in the standard model. However, by extending the model to include negative glue strengths, the authors show that temperature‑1 systems can regain full computational power, suggesting that the key to unlocking complexity lies not in the temperature per se but in the richness of the interaction rules governing tile binding. Future work will likely focus on the practical realization of negative glues, error‑correction strategies in such systems, and the exploration of other model extensions that could bridge the gap between theoretical universality and experimental feasibility.