--- title: FeFET-based QUBO Solvers for Combinatorial Optimization url: https://www.emergentmind.com/topics/fefet-based-qubo-solvers type: topic --- # FeFET-based QUBO Solvers for Combinatorial Optimization FeFET-based QUBO solvers are specialized hardware accelerators leveraging ferroelectric field-effect transistors (FeFETs) within compute-in-memory (CiM) or oscillator-based architectures to solve quadratic unconstrained binary optimization (QUBO) problems. These solvers exploit the non-volatility, analog programmability, and high endurance of FeFET devices for both encoding problem coefficients and in-memory computation, providing an energy-efficient, fast, and scalable alternative for tackling combinatorial optimization problems (COPs) with or without constraints. Several architectures, including the HyCiM system and FeFET-coupled oscillator Ising machines, have demonstrated substantial improvements in search-space reduction, area efficiency, and computation speed, with practical performance validated on non-trivial problem instances such as quadratic knapsack, graph coloring, Max-Cut, and MIMO detection [2410.14111][2309.13853][2511.00479]. ## 1. QUBO Formulation and Mapping to FeFET Hardware QUBO problems are expressed as $$ E(\mathbf{x}) = \mathbf{x}^T Q\,\mathbf{x} $$ with binary variables $\mathbf{x} \in \{0,1\}^n$ and coefficient matrix $Q \in \mathbb{R}^{n \times n}$. The binary optimization can encode COPs, including those represented by Ising Hamiltonians via the mapping $\sigma_i = 1-2x_i$: $$ H_P(\boldsymbol{\sigma}) = \sum_{i10³), multilevel storage via polarization programming, endurance (>10¹⁰ cycles), and ≥10-year data retention. In a 28 nm process, FeFETs can be structured as 1FeFET–1R cells to further suppress read variation [2309.13853][2511.00479]. Crossbar arrays organize FeFET cells to implement parallel vector-matrix-vector (VMV) multiplication ($E(\mathbf{x}) = \mathbf{x}^T Q \mathbf{x}$) in a single computational step: - **Data Encoding:** Input variables $x_i$ are supplied on wordlines (WL), and coefficients $Q_{ij}$ are represented by programmable $V_{\text{th}}$ in FeFETs or by conductance in crossbar patterns. - **Parallel Computation:** For each SA iteration, candidate $\mathbf{x}$ is broadcast, row and column currents representing energy are summed and digitized; at M-bit precision, multiple bit-sliced crossbars or per-cell quantization can be employed. - **Scalability:** Arrays are implemented at 32×32 size in silicon with tiling to support $\sim 1000$ variables; lossless matrix compression further improves density [2309.13853]. ## 3. FeFET-based Solvers for Constrained QUBO: The Inequality-Filter Paradigm Traditional D-QUBO methods for combinatorial problems with constraints require embedding $$ \sum_{i=1}^n w_i x_i \leq C $$ as a penalty, leading to an augmented space ($2^{n+C}$). The HyCiM architecture introduces a novel transformation that enforces constraints via a dedicated inequality-checking circuit (the "inequality filter") prior to QUBO energy evaluation [2410.14111]: - **Device-circuit co-design:** Each inequality-filter cell is a 1FeFET–1R stack, storing multi-level weights and supporting current-mode, phase-based comparison of partial sums. - **Array operation:** An $m \times n$ array stores weights across $n$ columns, with match-line voltages encoding $\sum w_i x_i$. A replica array encodes the constraint $C$; a two-stage comparator passes only feasible $\mathbf{x}$ for VMV evaluation. - **Impact:** This approach removes the need for auxiliary bits, reducing the search space from up to $2^{2536}$ ($n=100$, $C \approx 2536$) to only $2^{100}$, and delivering hardware area savings of 88–99.96%. ## 4. Annealing Algorithms, Compression, and Computational Workflow FeFET-based QUBO solvers commonly employ simulated annealing (SA) or enhanced variants such as MESA (multi-epoch SA): - **Annealing Process:** At each iteration, candidate configurations $\mathbf{x}_{\rm new}$ are tested against feasibility constraints (if present), and energies $E_{\rm new}$ are computed in-place in the crossbar. Acceptance follows the Metropolis rule: $$ P_{\rm accept} = \begin{cases} 1, & \Delta E \leq 0 \\ \exp(-\Delta E/T(t)), & \Delta E > 0 \end{cases} $$ with exponential or linear temperature decay [2410.14111]. - **Sparse QUBO Compression:** For sparse problems, compression maps the Q matrix to reduced-dimension blocks without energy loss; for example, producing dimension $p\times q$ with $p,q \ll N$, reducing area by 60–80% [2309.13853]. - **Oscillator Ising Machines:** Alternative architectures map QUBO couplings onto FeFET-controlled conductances in coupled oscillator arrays ("OIMs"). Oscillator phase synchronization, modeled by discrete Kuramoto dynamics, enables all-to-all QUBO minimization with time-to-solution scaling as $O(\log N)$ [2511.00479]. ## 5. Performance Metrics and Experimental Results FeFET-based QUBO solvers exhibit significant improvements in benchmarked metrics: | Metric | HyCiM (FeFET CiM) | D-QUBO Baseline | OIM (FeFET+CMOS) | |-----------------------|---------------------------|------------------------|--------------------------| | Search Space | $2^{100}$ (w/ filter) | up to $2^{2636}$ | $2^{M}$ (M spins) | | Area Savings | 88–99.96% vs. D-QUBO | — | N/A | | SA Time-to-solution | $\sim 1$–$10\,\mu$s @ $n=100$ | $>100\times$ higher | $O(\log N)$ scaling | | Solution Quality | 98.54% success (QKP) | 10.75% (QKP) | Near-ML BER, $M\leq 100$ | Additional results include: - Energy per QUBO evaluation: $\sim$2 pJ (crossbar), 0.5 pJ (filter) [2410.14111]. - MESA achieves target energies in 10–100× fewer iterations than conventional SA [2309.13853]. - Ring-Oscillator FeFET OIMs achieve logarithmic computation time scaling for dense QUBOs such as MIMO detection, limited by conductance tuning windows (1–60 μS per device) [2511.00479]. ## 6. Device- and System-Level Considerations FeFET-based designs are subject to various device-level effects and practical system constraints: - **Threshold Variation:** Device-to-device $V_{\rm th}$ variation $\sim 100$ mV; can be mitigated via calibration [2410.14111]. - **Conductance Tuning:** OIM performance is optimal for $G_{\min} \sim 1~\mu$S, $G_{\max} \sim 60~\mu$S; outside this window, coupling weakens or stalls. - **Endurance and Retention:** HfO₂ FeFETs documented to $>10^{10}$ write/erase cycles and $>10$ years retention [2410.14111]. - **Scalability:** Demonstrated filter/annealer arrays range from 16×100 up to 256×256; crossbar chaining and compression allow scaling to $N\sim 1000$–$4096$ [2309.13853]. - **Noise and Robustness:** Simulated annealing and Kuramoto OIMs robust against analog noise; periodic refreshing mitigates drift. ## 7. Outlook and Application Domains FeFET-based QUBO solvers offer a scalable platform for hardware acceleration of binary combinatorial optimizations: - **Versatility:** Efficient support for both constraint-free and constrained COPs (e.g., quadratic knapsack, Max-Cut, graph coloring, ML MIMO detection). - **Integration:** Three-terminal FeFETs compatible with advanced CMOS, enabling in-memory or near-memory compute paradigms. - **Future Prospects:** Anticipated advances include denser 3D FeFET×CMOS integration, automated annealing protocols, expanded conductance range, and improved device uniformity. Applications extend to edge-AI, logistics, communication systems, and real-time optimization [2511.00479][2410.14111][2309.13853]. FeFET-based QUBO solvers thus represent a leading architecture in the evolution of hardware-accelerated combinatorial optimization, providing order-of-magnitude benefits in energy, area, and solution speed, while maintaining competitive or superior solution quality on established benchmarks. Source: https://www.emergentmind.com/topics/fefet-based-qubo-solvers