Good question (thanks for making me stretch a bit). The short answer: yes, Danielle S. Bassett (Dani Bassett) and her team do use models that are very much in the energy-landscape / maximum‐entropy / statistical‐physics family — but I don’t find evidence that they literally adopt the same “EBM + block Gibbs / TSU hardware” framework as THRML. In other words: overlap exists, but it’s not a perfect match. Here’s the breakdown.

What Bassett

does

in her research

• Bassett’s work uses network neuroscience: brain regions as nodes, structural/functional connectivity as edges.
• She also uses algebraic‐topology / simplicial complex representations of neural data (i.e., beyond pairwise edges) — e.g., “Two’s company, three (or more) is a simplex” by Giusti, Ghrist & Bassett.
• There is work by her team directly using a maximum entropy model of brain states: e.g., the article “The Energy Landscape of Neurophysiological Activity Implicit in Brain Network Structure” with Gu, Cieslak et al (Bassett as author) uses a binary model of brain region activation, and defines an energy landscape.

Important extracts from that article:

• They model each brain region as binary “active/inactive” and use a pairwise maximum entropy model informed by structural connectivity.
• They interpret the energy landscape of brain states (local minima, basins) and link those to functional observations.
• They note: “the maximum entropy model is appropriate for systems at equilibrium” and they mention future work should incorporate higher‐order interactions (simplices) because some co‐activations cannot be captured by pairwise alone.

So yes: she is using “energy‐based” style modeling (i.e., energies over states, probabilities ∝ exp(–E), looking at minima, basins) which is very much in the EBM conceptual family.

What Bassett

doesn’t (apparently) do

, as far as I found

• I did not find evidence that she uses block‐Gibbs sampling or explicitly uses hardware architecture like TSUs to realize sampling of EBMs. The “Energy Landscape” paper used a “random walk” to sample minima, not a full Gibbs sampler.
• I did not find evidence that her simplicial‐complex work explicitly formulates energies or sampling over higher‐order simplices in the way that a full EBM sampler library might. The simplicial complex work is more about structural/ topological characterization, not necessarily about a full probabilistic generative model with sampling. For example “Two’s company, three (or more) is a simplex” focuses on topology of neural data rather than explicit energy-based generative modeling.