Multi-valued logic research brief

Multi-State Computing: Beyond Binary

A concise overview of the current state of non-binary computing, from ternary logic and memristors to qudits and probabilistic hardware.

Most modern computers are built on binary logic: 0 and 1. But computing does not have to stop there. Researchers have long explored systems that use three or more stable states, often called multi-state, multi-valued, or non-binary computing. Today, the field is active again, driven by new materials, new device physics, logic-in-memory designs, and renewed interest in architectures that can represent more information per element than a conventional bit.

Research direction, not a mainstream replacement Strongest momentum in memory-centric and specialized hardware Serious technical hurdles remain

Current state of the field

Promising, active, and still far from standard practice

Multi-state computing is best understood as a promising research direction rather than an established replacement for mainstream binary processors. The most mature and widely discussed branch is ternary logic, especially radix-3 systems, because ternary is often treated as a practical compromise between increased information density and manageable circuit complexity.

Recent work includes device-level demonstrations of binary and ternary logic-in-memory, reconfigurable binary/ternary conversion-in-memory, transistor designs that expose an intermediate logic state, memristor-based multi-valued logic, and higher-dimensional quantum computing using qudits. These developments suggest that non-binary logic may be most useful first in specialized or memory-centric systems rather than as a direct drop-in replacement for general-purpose CPUs.

At the same time, there are still major obstacles: fabrication complexity, signal integrity, noise margins, device variability, software tooling, and the lack of a mature ecosystem comparable to binary CMOS computing.

Why it matters

Why researchers keep returning to non-binary architectures

More information per symbol.
Potential reductions in interconnect and wiring overhead.
Possible efficiency gains in some arithmetic and memory operations.
A good fit with emerging devices that naturally support intermediate or multiple stable states.
Relevance to AI accelerators, in-memory computing, optimization hardware, and quantum-adjacent architectures.

Main research areas

Five fronts where multi-state computing is advancing

Ternary and multi-valued logic circuits

CMOS, CNTFET, organic, and transistor-level designs for arithmetic, memory, and control logic.

Logic-in-memory and memristors

Architectures that combine storage and computation, often making multi-state representations more natural than in standard CMOS pipelines.

Spintronics and emerging devices

Research using magnetic tunnel junctions and other non-CMOS technologies for ternary logic and logic-in-memory.

Qudit quantum computing

Higher-dimensional quantum information processing, along with compilers, simulators, and mixed-dimensional toolchains.

Probabilistic and physics-based computing

Multi-level probabilistic hardware aimed especially at difficult optimization problems.

Selected recent pointers

Selected pointers, not an exhaustive bibliography.

Design implementations of ternary logic systems: A critical review (2024)
A broad review of ternary logic implementations, comparing design approaches, device technologies, and open practical constraints.
Binary and ternary logic-in-memory using nanosheet feedback field-effect transistors (2024)
Device-level work showing binary and ternary logic-in-memory using nanosheet feedback FET structures.
A reconfigurable binary/ternary logic conversion-in-memory (2023)
A reconfigurable conversion-in-memory design that moves between binary and ternary logic within the memory fabric.
Reconfigurable binary and ternary logic devices enabling logic applications (2025)
Reports transistor-level logic devices with controllable intermediate states, aimed at flexible binary and ternary operation.
Ternary computing using a novel spintronic multi-operator logic-in-memory design (2025)
Explores a spintronic logic-in-memory architecture for ternary computing and its implications for compact logic operations.
MQT Qudits: A Software Framework for Mixed-Dimensional Quantum Computing (2024)
A software-oriented framework for simulating, compiling, and experimenting with qudit and mixed-dimensional quantum workflows.
Universal quantum computing with qubits embedded in trapped-ion qudits (2024)
Demonstrates how higher-dimensional trapped-ion systems can support universal quantum computing with embedded qubit structure.
Multi-level probabilistic computing: application to the multiway number partitioning problems (2025)
Connects multi-level probabilistic hardware to optimization problems, extending binary probabilistic ideas toward multiway partitioning.

Bottom line

A serious field with near-term value in specialized systems

Multi-state computing is no longer just a historical curiosity, but it is not yet a mainstream computing paradigm either. The field is advancing through specialized hardware research, especially where device physics makes multiple states natural and useful.

The likely near-term future is not a universal replacement for binary laptops and servers, but a growing set of niche and high-performance applications in memory-centric systems, AI accelerators, optimization hardware, and quantum-adjacent computing.