Lattice field theory for superconducting circuits

The Big Picture

Imagine predicting the behavior of a symphony orchestra by modeling every musician, every instrument, every vibrating string and resonating chamber, all at once, from first principles. That’s roughly the challenge facing engineers who want to understand large superconducting quantum circuits. These devices form the backbone of quantum computers built by Google, IBM, and others, and they can contain hundreds of interconnected quantum components. Simulating them all at once is a combinatorial nightmare.

The standard workaround has been approximation. Fluxonium qubits, a promising design known for holding quantum states for unusually long periods, are built from hundreds of Josephson junctions, the tiny quantum switches inside these circuits. Physicists typically collapse all those junctions into a single simplified element called a superinductance. It works in many cases, but it misses physics that matters, especially when the circuit operates at high impedance.

At high impedance, collective quantum effects called coherent quantum phase slips begin to dominate. These are coordinated quantum tunneling events that can propagate through the circuit and degrade performance.

Researchers at MIT have now borrowed a computational framework from nuclear and particle physics, lattice field theory, and retooled it to simulate superconducting circuits from scratch, without approximation. Their approach does what no other method does cleanly: thousands of complete qubit simulations, all at the microscopic level, with none of the quantum complexity discarded.

Key Insight: By treating a superconducting circuit as a lattice field theory and using Monte Carlo sampling over the full quantum phase space, this method computes qubit properties without the systematic errors that plague existing approaches and captures many-body quantum effects in real hardware that other approaches miss.

How It Works

The idea: take the mathematical machinery physicists use to study quarks and gluons inside protons and apply it to a circuit etched onto a silicon chip. The connection runs deeper than analogy.

In circuit QED, the detailed microscopic theory of superconducting devices, each superconducting island carries a phase variable: a quantum degree of freedom that lives on a circle (technically, a U(1) variable). An array of 100 Josephson junctions means 100 such variables, all quantum-mechanically entangled. The governing equation is, in general, unsolvable by brute force.

Lattice field theory gets around this. The method proceeds in four steps:

1. Formulate dynamics as a path integral. Instead of solving for quantum wavefunctions directly, quantum amplitudes are expressed as sums over all possible histories of the system. This is a Feynman path integral reformulated in imaginary (Euclidean) time. Standard in particle physics; new in this context.
2. Discretize the time direction. Continuous Euclidean time is replaced by a finite lattice of time slices. The circuit’s spatial structure (its network of capacitors, inductors, and junctions) is already discrete, so this is a natural fit.
3. Sample with Monte Carlo. The path integral is evaluated by randomly sampling configurations of the phase variables, directly over the full circular U(1) space without truncating it. This sidesteps the systematic errors other methods incur.
4. Extract the spectrum from correlators. Qubit energies and matrix elements are recovered from how quantum correlations decay across Euclidean time slices. This is exactly how particle physicists extract hadron masses from lattice QCD.

The advantage over tensor network (TN) methods, currently the state of the art for many-body circuit simulation, lies in step three. Tensor networks must choose a finite basis to represent each junction’s quantum state space, which is technically infinite-dimensional. That introduces truncation errors requiring careful justification. The lattice method keeps variables on their natural domain and avoids the problem entirely.

There’s a practical bonus, too. After computing one device, standard lattice techniques like reweighting can simulate nearby devices in fabrication space at almost no additional cost. When you’re exploring many design parameters, that adds up fast.

Why It Matters

The team applies their method to fluxonium, a qubit built from a single small Josephson junction shunted by an array of 100 or more larger junctions. The array functions as a superinductor, an element with unusually high inductance that gives fluxonium its resistance to environmental noise. But the array also hosts collective quantum excitations called array modes, and at high impedance, coherent phase slips can propagate through it and disrupt the qubit’s quantum coherence.

Thousands of individual fluxonium qubits are simulated, each with a different random realization of charge disorder across the junction array. This kind of statistical study at the microscopic level, explicitly averaging over disorder across the full circuit, is difficult to achieve with any other existing method. From these simulations, they extract the charge noise dephasing rate as a function of impedance and find deviations from existing analytic formulas. At very high impedances, they observe signatures of multiple simultaneous phase slip events that current analytic theories cannot capture.

Then there’s the ground capacitance result. Every circuit element has some small capacitance to the chip’s ground plane, and these stray capacitances are usually ignored. The lattice calculations show that ground capacitances actually reduce charge noise. That’s counterintuitive and directly useful for hardware design. It also goes beyond what tensor network methods have so far achieved for fluxonium.

Next-generation qubit designs like the 0-π qubit, blochnium, bifluxon, and harmonium all rely on long Josephson junction arrays where many-body effects matter. As quantum processors scale toward fault-tolerant operation, a method that predicts device behavior without approximation becomes essential rather than optional.

Bottom Line: Borrowing tools from nuclear physics for quantum hardware design, this paper introduces the first lattice field theory approach to superconducting circuits, simulating thousands of realistic fluxonium qubits at the microscopic level and exposing many-body physics that no other technique can currently resolve.