🔬 materials science

Memory of topologically constrained disorder in Shakti artificial spin ice

This study demonstrates that topologically constrained disorder in Shakti artificial spin ice enables robust, sequence-dependent memory effects absent in conventional square spin ice, revealing how correlated disorder and cyclic driving can generate deterministic and stochastic memory in frustrated artificial materials.

Original authors: Priyanka Priyanka, Cristiano Nisoli, Yair Shokef

Published 2026-08-03

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Priyanka Priyanka, Cristiano Nisoli, Yair Shokef

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world made of tiny, invisible magnets, each one a stubborn little compass needle that wants to point in a specific direction. In the realm of physics, these are called "spins," and when you pack them together in a grid, they don't just sit quietly; they argue, push, and pull on each other. Sometimes, they find a perfect, peaceful arrangement where everyone is happy. Other times, they get stuck in a messy, chaotic tangle where no single arrangement satisfies everyone. This messy state is called "disorder." But here's the twist: not all mess is created equal. Some mess is just random noise, like static on a radio. But other mess is "constrained," meaning the chaos follows strict, hidden rules, almost like a puzzle where the pieces can only fit together in certain ways. Scientists call this "topological disorder."

Why should we care? Because this kind of constrained mess has a superpower: memory. Just like you might remember the specific path you took to get to school because of the traffic lights you passed, these magnetic materials can "remember" the exact sequence of magnetic fields they were exposed to. If you push them one way, then another, they might end up in a different spot than if you pushed them in the reverse order. This isn't just a cool party trick; it could be the key to building new kinds of computers that store data in the shape of their disorder rather than in neat, ordered rows. The big question is: does this memory work best when the material is perfectly ordered, or when it's beautifully, topologically messy?

This paper dives into that question by playing with two different types of "artificial spin ice"—which are basically man-made grids of tiny magnets designed to mimic the behavior of water ice, but with a magnetic twist. The researchers compared two very different playgrounds. The first is the "Square Spin Ice," a neat, orderly grid where the magnets are arranged in perfect squares. The second is the "Shakti Spin Ice," a more complex, jagged grid that looks like a Shakti (a Hindu deity) symbol, created by removing some magnets from the square grid. This Shakti grid is special because it forces the magnets into a state of "topological disorder," where they are stuck in a degenerate, or equally happy, mess.

The team, led by Priyanka Priyanka, Cristiano Nisoli, and Yair Shokef, didn't just sit back and watch; they actively probed these magnetic grids with external magnetic fields, acting like a conductor directing an orchestra. They used different "protocols," or musical scores, to see how the magnets responded. Some protocols were simple, pushing the magnets in just one direction (like a one-way street). Others were more complex, pushing them in two directions at once, or sending them on a round trip where they went out and came back along the same path, or even a loop where the return path was different from the outgoing path.

What they found was a tale of two very different personalities. The orderly Square Spin Ice was boringly predictable. No matter how they pushed it, or in what order, the magnets always ended up in the same spot. It had no memory of the path it took; it only cared about where it ended up. It was like a robot that always takes the shortest route home, ignoring the scenic detours.

The Shakti Spin Ice, however, was a different story entirely. It was full of surprises and had a rich, complex memory. When the researchers pushed the Shakti magnets in a specific sequence, the final arrangement of the magnets depended entirely on the order of the pushes. If they pushed hard in the X direction first, then the Y direction, the magnets settled into one pattern. If they reversed the order, they settled into a completely different pattern. This "sequence-dependent response" meant the material was remembering its history.

The study revealed that this memory wasn't just a simple on/off switch. It depended on how strong the magnets were talking to each other (a ratio the authors call $\alpha$ ). When the magnets were in a "degenerate" state (the topologically messy one), the Shakti ice showed a wild mix of behaviors. Sometimes the outcome was deterministic (predictable but different based on the path), and sometimes it was stochastic (random, like flipping a coin), where the system couldn't decide on a single final state and would fluctuate between possibilities. This happened even in the smallest repeating block of the material, a 4x4 grid.

Perhaps most fascinating was what happened when they tried to make the magnets "forget" and return to their starting point. In the "Two-Way" protocol, where they pushed the magnets out and pulled them back along the exact same path, the Shakti ice often remembered its initial state, but only if the push wasn't too strong. However, in the "Loop" protocol, where the return path was different from the outgoing path, the memory was often destroyed. The tangled flipping of the magnets during the loop scrambled the history, and the system ended up in a new, excited state, unable to return to where it started.

The paper suggests that this memory effect is a direct result of the "topological constraints" of the Shakti lattice. Because the magnets are forced into a specific kind of disorder, they have multiple paths to get from point A to point B, and the path they take changes the destination. The researchers ran these experiments using computer simulations (since they are working with theoretical models of these nano-magnets), and the results were clear: topological disorder creates a robust, programmable memory that simple, ordered materials lack.

In short, the paper shows that while a perfectly ordered system is predictable and forgetful, a system with topologically constrained disorder is a master of memory. It can remember the sequence of events that led to its current state, offering a new way to think about how materials store information. The authors propose that this isn't just a quirk of these specific magnetic grids, but a fundamental property of topological systems that could be harnessed for future technologies, turning the "mess" of disorder into a powerful tool for memory.

Technical Summary: Memory of Topologically Constrained Disorder in Shakti Artificial Spin Ice

Problem Statement
Complex systems often exhibit memory effects where their response depends on the history of external manipulations. While memory is frequently associated with the retention of metastable order in systems like ferromagnets, the specific role of "constrained disorder"—particularly disorder that is topological in nature—remains underexplored. Artificial spin ices (ASIs), arrays of frustrated nanomagnets, provide a tunable platform to study these phenomena. This paper investigates the sequence-dependent responses (memory) of two distinct ASIs: the square spin ice, which exhibits conventional antiferromagnetic Landau order, and the Shakti spin ice, which can exhibit a disordered phase characterized by algebraically correlated classical topological order. The central question is whether topological constraints in a disordered manifold generate robust memory effects distinct from those in ordered systems.

Methodology
The authors employ a combination of analytical tracking and zero-temperature Monte Carlo simulations to study the dynamics of spin flips under quasistatic external magnetic fields.

Models: The study compares the square lattice (periodic boundary conditions) with the Shakti lattice, obtained by ordered decimation of the square lattice, resulting in vertices of coordination 2, 3, and 4. The system is governed by a Hamiltonian with parallel ( $J_\parallel$ ) and perpendicular ( $J_\perp$ ) couplings, characterized by the dimensionless ratio $\alpha = J_\perp/J_\parallel$ .
Protocols: The authors apply various in-plane magnetic field protocols to probe sequence dependence:
- Unidirectional (1D): One-way (monotonic increase), Two-way (increase then return to zero), and Pulse (two successive pulses of different strengths).
- Bidirectional (2D): One-way (increasing $h_x$ then $h_y$ , or vice versa), Two-way (retracing the path to zero), and Loop (reaching a target field via one path and returning via the reverse path).
Analysis: The response is characterized by the final magnetization and the specific spin configuration reached after the protocol. The authors analyze small systems ( $4 \times 4$ supercells) analytically to track exact spin-flip dynamics and larger systems ( $L=32$ ) via simulation to observe statistical trends. They specifically examine the dependence on the interaction ratio $\alpha$ , distinguishing between regimes where the ground state is antiferromagnetic ( $\alpha > 2$ ) and where it is a degenerate, topologically constrained state ( $\alpha < 2$ ).

Key Contributions and Results

Contrast Between Square and Shakti Ice:
- Square Ice: Exhibits predictable, deterministic behavior. The response to external fields is independent of the sequence of field application. Once the critical field is exceeded, spins align with the field, and no sequence-dependent memory is observed.
- Shakti Ice: Exhibits pronounced sequence-dependent responses. The system's final state depends on the specific path taken by the external field, demonstrating that topological constraints in the disordered phase harbor memory.
Role of Interaction Ratio ( $\alpha$ ):
- $\alpha > 2$ (Ordered Antiferromagnetic Ground State): Even in this ordered phase, Shakti ice shows sequence dependence in bidirectional protocols, unlike square ice. This is attributed to the fact that the topological state represents a degenerate manifold of excitations unique to the Shakti geometry.
- $\alpha < 2$ (Degenerate Topological Ground State): The system exhibits rich, often stochastic behavior. The response is highly sensitive to the protocol, with bifurcations leading to multiple possible final states.
Protocol-Specific Findings:
- Unidirectional Protocols (1D1W, 1D2W, 1DPulse):
  - In the degenerate state ( $\alpha < 2$ ), the system displays bifurcations and non-deterministic outcomes for fields above a certain threshold.
  - In the pulse protocol, sequence dependence emerges when pulse amplitudes differ, particularly for $\alpha < 2$ where all configurations show sequence dependence, whereas for $\alpha > 2$ , only a subset of configurations do.
- Bidirectional Protocols (2D1W, 2D2W, 2DLoop):
  - 2D1W (One-way): For $\alpha > 2$ , the response is deterministic but path-dependent (XY vs. YX paths yield different magnetization). For $\alpha \le 2$ , the response becomes stochastic with multiple possible outcomes due to bifurcations.
  - 2D2W (Two-way/Return): Retracing the same path to zero field allows for memory retention. For $\alpha > 2$ , the system often returns to a degenerate state (though not necessarily the initial one). For $\alpha < 2$ , memory retention is fragile; the system often gets stuck in excited states or fluctuates between them.
  - 2DLoop (Loop): Changing the return path (e.g., XY to YX) destroys memory retention more effectively than retracing the path. For $\alpha < 2$ , the system fails to return to the initial degenerate state at relatively low field thresholds.
System Size Effects:
- While small supercells show specific bifurcation behaviors, larger systems ( $L=32$ ) exhibit a finite fraction of defects (Type II2 vertices) that remains independent of system size for $\alpha < 2$ but increases with size for $\alpha > 2$ . This suggests that the deviation from the ground state scales differently depending on the interaction regime.

Significance and Claims
The paper claims that topologically constrained disorder acts as a powerful carrier of memory, distinct from the memory associated with metastable order. The key findings are:

Topological Constraints Generate Memory: Unlike the ordered square ice, the Shakti lattice exhibits history-dependent responses even in its ordered phase ( $\alpha > 2$ ) due to its underlying topological degeneracy.
Stochasticity in Disordered Phases: In the topologically disordered phase ( $\alpha < 2$ ), the system exhibits stochastic responses and bifurcations, leading to a hierarchy of memory sensitivity governed by small changes in interaction parameters.
Kinetic vs. Thermodynamic: The results suggest that memory in these frustrated materials is not solely a thermodynamic property of the ground state but is deeply rooted in the kinetics of spin flips and the topological constraints of the phase space.
Path Dependence of Memory: The study demonstrates that retracing an input path (2D2W) allows the system to retrieve initial states for large maximal fields, whereas loop protocols (2DLoop), where the return path differs from the forward path, often destroy memory retention due to the intertwined flipping of orthogonal spins.

The authors conclude that these findings open new perspectives on how topological constraints and correlated disorder can be utilized to generate robust, programmable memory effects in frustrated artificial materials, a phenomenon previously examined mainly through the lens of ground-state kinetics and thermodynamics.

More like this