MIST: Measurement-Induced State Transition

• MIST is a phenomenon where measurement backaction causes quantum state transitions, leading to leakage from the computational subspace in systems like superconducting qubits.
• The underlying mechanism involves dressed-state interactions and level crossings in models such as the Jaynes–Cummings Hamiltonian, which explains observed readout instabilities.
• Beyond circuit QED, MIST concepts extend to mesoscopic and many-body systems, where measurements induce changes in occupation, correlations, and entanglement.

Measurement-induced state transition (MIST) denotes a backaction-driven change of state caused by measurement or monitoring. In dispersively read out superconducting qubits, it refers to unwanted transitions out of the computational subspace when a microwave readout pulse populates the resonator and brings dressed qubit–photon states into resonance with higher levels, thereby degrading the quantum non-demolition character of readout (Hirasaki et al., 2024, Nesterov et al., 2024, Sank et al., 2016). In mesoscopic and many-body settings, related usage describes measurement-controlled changes of occupation, correlations, entanglement, or boundary conditions, including a many-body transition in a double quantum dot and measurement-induced phase transitions in monitored quantum systems (Ferguson et al., 2020, Buchhold et al., 2021, Feng et al., 3 Feb 2025).

1. Terminology and scope

A recurring source of confusion is terminological overlap across subfields. In the superconducting-qubit literature, MIST is a concrete readout-error mechanism: the measurement pulse itself drives the qubit into higher noncomputational states such as $|2\rangle, |3\rangle,\dots$, or into higher transmon, fluxonium, or array-mode excitations (Hirasaki et al., 2024, Zwanenburg et al., 16 Jun 2026). In monitored many-body dynamics, the standard label is usually measurement-induced phase transition (MIPT), referring to a change between entangling and disentangling phases or between long-range and short-range correlated regimes (Buchhold et al., 2021, Feng et al., 3 Feb 2025).

The overlap is not merely lexical. An exactly solvable Liouvillian analysis states that one may equivalently view the many-body phenomenon as a non-equilibrium “state transition” between phases characterized by long-range (or algebraic) correlations and short-range (exponential) correlations (Paviglianiti et al., 2024). A common misconception is therefore to treat all uses of MIST as synonymous. The literature instead supports a layered picture: in circuit QED, MIST is a non-QND leakage process during dispersive readout; in mesoscopic transport, it can denote detector-backaction-driven switching of many-body occupations; and in monitored many-body systems, closely related language is used for trajectory-level phase structure (Ferguson et al., 2020, Paviglianiti et al., 2024).

2. Microscopic mechanism in dispersive readout

In superconducting circuits, the basic mechanism is the dressing of a weakly anharmonic qubit by photons in the readout resonator. For a transmon–cavity system, a representative starting point is the Jaynes–Cummings Hamiltonian including $|g\rangle$, $|e\rangle$, and $|f\rangle$,

$$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$

which in the dispersive limit yields an effective cavity pull $\chi$ and an inherited cavity Kerr constant $K=2\chi^2/\alpha$ (Choi et al., 2024). Under readout, a coherent drive populates the resonator with $\langle n\rangle$ photons; the combined eigenstates $|i,n\rangle$ are then dressed by the qubit–resonator interaction, and MIST occurs when a dressed state such as $|1,n\rangle$ comes into near resonance with a higher dressed state $|g\rangle$0 (Hirasaki et al., 2024).

The threshold is commonly expressed as a dressed-level crossing condition. In the transmon analysis of excited-state bistability, the “usual MIST crossing condition” is

$|g\rangle$1

or equivalently $|g\rangle$2 (Choi et al., 2024). In the dispersive-readout framework of dressed coherent states, the same physics appears as the breakdown of an eigenbranch picture when the resonator ring-up approaches an avoided crossing between dressed ladders (Nesterov et al., 2024).

Two distinct microscopic descriptions coexist in the literature. One line of work attributes the transitions to level crossings within the Jaynes–Cummings ladder mediated by terms in the Hamiltonian that are typically ignored by the rotating wave approximation, and identifies an unexpected broken symmetry in the qubit potential as the most important of these terms (Sank et al., 2016). Another line shows that a semi-classical model within the rotating wave approximation already predicts the onset of state transitions in the regime where the resonator frequency is lower than the qubit frequency, and suggests that the transmon is excited to levels near the top of its cosine potential, where charge dispersion explains noisy threshold behavior (Khezri et al., 2022). The coexistence of these descriptions reflects parameter dependence rather than a single universal microscopic reduction.

3. Threshold criteria, bistability, and dynamical regimes

A static prediction strategy is provided by dressed coherent states. Starting from the drive-free Hamiltonian, one defines a qubit purity error $|g\rangle$3, with $|g\rangle$4, and a matrix-element error

$|g\rangle$5

Away from avoided-level-crossings, $|g\rangle$6 remains very small and $|g\rangle$7; near the onset of MIST, both rise sharply, allowing one to compute a robust $|g\rangle$8 solely from the static Hamiltonian spectrum (Nesterov et al., 2024).

In one fluxonium example, the anticrossing of $|g\rangle$9 and $|e\rangle$0 occurs at $|e\rangle$1, the purity error rises sharply in the region $|e\rangle$2, and the inferred safe operating range is $|e\rangle$3–$|e\rangle$4 (Nesterov et al., 2024). This supports the broader claim of that work that the same metrics can be applied to transmons and other superconducting qubits without qubit-type-specific approximations.

In transmon excited-state readout, however, the critical response is not always set directly by the dressed-level crossing. The measurements and semiclassical dynamics model of the cavity photon state lead to the following comparison (Choi et al., 2024):

Manifold	$|e\rangle$5	$|e\rangle$6 and $|e\rangle$7
$|e\rangle$8	$|e\rangle$9	$|f\rangle$0
$|f\rangle$1	$|f\rangle$2	$|f\rangle$3
$|f\rangle$4	$|f\rangle$5	$|f\rangle$6

In the ground-state manifold, $|f\rangle$7 lies within errors of $|f\rangle$8. In the $|f\rangle$9 and $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$0 manifolds, by contrast, the observed onset is dramatically below $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$1. The semiclassical dynamics picture resolves the discrepancy: bistable “dim” and “bright” cavity states appear first, and the saddle-node photon numbers $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$2 agree with experiment within experimental uncertainty, so bistability pre-empts MIST in the excited-state sectors (Choi et al., 2024).

A later driven-dissipative reduction makes the nonequilibrium structure more explicit. In a two-photon resonance between $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$3 and a higher state $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$4, the reduced master equation yields rates $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$5 and $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$6, a steady-state inversion condition $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$7, and a “super-MIST” regime characterized by steady-state qubit inversion and slow relaxation beyond semiclassical Landau–Zener predictions (Pan et al., 18 Aug 2025). The same framework identifies a transient readout condition in which the resonator becomes highly populated while the qubit remains near its original state because the resonator relaxes on the scale $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$8 while qubit populations evolve on the slower scale $$H_{JC} = \hbar\omega_r a^\dagger a + \sum_{j=g,e,f}\hbar\omega_j |j\rangle\langle j| - \hbar\frac{\alpha}{2} b^\dagger b^\dagger b b + \hbar g(a^\dagger b + a b^\dagger),$$9 (Pan et al., 18 Aug 2025).

4. Experimental phenomenology across superconducting platforms

Time-resolved data from an IBM Quantum transmon processor show that MIST can appear as temporal instability rather than a static threshold alone. On the $\chi$0-qubit device ibm_kawasaki, measurements on qubit $\chi$1 found that $\chi$2 undergoes abrupt jumps from $\chi$3 to $\chi$4, with high-error periods lasting tens to hundreds of seconds. Histograms of the high-error durations span $\chi$5 up to $\chi$6, and the interpretation given is that temporal fluctuations such as transmon offset charge move dressed energies in and out of resonance, turning MIST “on” and “off” (Hirasaki et al., 2024).

Fluxonium extends the phenomenology by adding a dense excited-state structure and superinductor array modes. Across the full external flux range, experimental characterization and numerical modeling identified eleven regions with increased MIST. Regions (1)–(6) are captured by conventional branch analysis and correspond to resonances with higher fluxonium levels; regions (7)–(11) are dominated by transitions involving the transmission-line-like array modes of the superinductor, and Floquet-branch analysis reproduces all five array-mode hot spots with excellent quantitative agreement in flux offset and relative amplitude (Zwanenburg et al., 16 Jun 2026).

The phenomenon also changes qualitatively in multi-qubit chips. In a two-transmon setting, the presence of a spectator qubit lowers the MIST threshold of the readout qubit and the spectator can itself be impacted by the measurement-induced transition of the readout qubit. Adding a coupler mode further modifies these effects by introducing additional resonance manifolds and, in some regimes, destructive interference of charge-matrix elements (Hoyau et al., 3 Jun 2026). This suggests that MIST in processors cannot be reduced to a single isolated qubit–resonator pair.

Several hardware directions attempt to stabilize or suppress the effect. Inductively-shunted transmons eliminate offset-charge dependence and experimentally exhibit MIST in agreement with quantum and semiclassical models, with the central advantage that the MIST locations in readout parameter space no longer drift with offset charge (Zobrist et al., 12 Mar 2026). A different approach, the cos$\chi$7-coupling readout scheme, uses symmetry to suppress nonparity-conserving MIST; at zero flux it is reported to be free of MIST up to high powers, with more than $\chi$8 photons in the readout mode, and the protected regime can be controllably broken by flux-tuning, which activates specific leakage pathways such as $\chi$9 and $K=2\chi^2/\alpha$0 (Mori et al., 5 Sep 2025).

5. Mesoscopic and many-body uses of the concept

Outside circuit QED, detector backaction can itself induce a state transition of a many-body device. In a mesoscopic double quantum dot in the Coulomb-blockade regime, a nearby charge-sensor dot switches the electron population through measurement. The rate-equation description uses broadened transition rates

$K=2\chi^2/\alpha$1

and the hallmark is an “S-shape” in the charge-degeneracy line. The transition appears when the backaction broadening $K=2\chi^2/\alpha$2 becomes comparable to or exceeds thermal broadening, roughly $K=2\chi^2/\alpha$3 (Ferguson et al., 2020).

In monitored many-body systems, the standard theoretical object is the measurement-induced phase transition. An $K=2\chi^2/\alpha$4-replica Keldysh field theory for measured Dirac fermions shows an exact decoupling into one mode that heats to infinite temperature and $K=2\chi^2/\alpha$5 “cold” modes governed by an effective non-Hermitian Hamiltonian; after bosonization this becomes a non-Hermitian sine-Gordon model with a Berezinskii–Kosterlitz–Thouless transition between a gapless phase with logarithmic entanglement scaling and a gapped area-law phase (Buchhold et al., 2021). In a noisy and disordered Heisenberg chain, continuous measurements of $K=2\chi^2/\alpha$6 induce a steady-state transition from volume-law to area-law entanglement, with extracted correlation-length exponent $K=2\chi^2/\alpha$7 independent, within errors, of noise and disorder strength (Boorman et al., 2021).

The many-body literature also shows that “measurement-induced state transition” is not a single universal object. In the finite-$K=2\chi^2/\alpha$8 Sachdev–Ye–Kitaev model, entanglement and purification transitions are found to be distinct phenomena because entanglement can revive after a completely projective measurement but impurity cannot (Haldar et al., 2023). An exactly solvable quasi-free Liouvillian model demonstrates that any imperfect postselection, $K=2\chi^2/\alpha$9, produces a finite correlation length and rounds the transition into a crossover (Paviglianiti et al., 2024). By contrast, a tree-shaped Haar-random circuit with weak measurements admits a postselection-free experimental observation on a trapped-ion processor and an exact critical point $\langle n\rangle$0, equivalent to $\langle n\rangle$1 (Feng et al., 3 Feb 2025). A separate line of work shows that even a single round of measurements on a gapless state can induce measurement-induced boundary transitions governed by distinct boundary conformal field theories (Liu et al., 2024).

6. Mitigation, design principles, and extensions

The practical mitigation literature is dominated by threshold management. One prescription is to operate at photon numbers well below the lowest $\langle n\rangle$2 set by any multiphoton crossing, in practice $\langle n\rangle$3, to optimize the ring-up rate around $\langle n\rangle$4, to use an adiabatic drive envelope, and to incorporate a Purcell or band-pass filter (Nesterov et al., 2024). Closely related strategies are to increase detuning $\langle n\rangle$5, reduce $\langle n\rangle$6, slow the resonator ring-up and ring-down, and operate exactly at the transmon sweet spot to suppress the cubic term that opens important non-RWA channels (Sank et al., 2016).

Device-level redesign can shift or remove the dominant leakage channels. In fluxonium, balancing the capacitances of the upper and lower islands or engineering the array layout to restore inversion symmetry is proposed to suppress coupling to the first array mode at sweet spot; other suggested mitigations are fewer photons, dynamically detuning the readout tone, and larger anharmonicity (Zwanenburg et al., 16 Jun 2026). In multi-qubit processors, a plausible implication is that MIST-aware design must include spectators and couplers explicitly, because they alter the resonance landscape rather than merely perturbing it (Hoyau et al., 3 Jun 2026). Inductive shunting stabilizes the MIST landscape by eliminating offset-charge dependence (Zobrist et al., 12 Mar 2026), while cos$\langle n\rangle$7-coupling uses parity selection rules to suppress one-photon exchange processes and thereby pushes MIST to much higher readout powers (Mori et al., 5 Sep 2025).

Beyond hardware, the concept continues to expand. A conjectural extension interprets self-organized MIPT as a learnability transition in cognitive networks and associates a measurement-induced state transition with the formation of stable concepts in semantic memory (Gorsky, 5 Jun 2025). That proposal remains distinct from the experimentally established superconducting-qubit and monitored-matter usages, but it underscores the central theme shared across the literature: measurement is not merely a diagnostic operation, but a dynamical ingredient capable of reorganizing state structure, occupation, or entanglement in a quantitatively predictable way (Gorsky, 5 Jun 2025).