⚛️ quantum physics

Typicality of Steering for Two-qubit States

This paper investigates the typicality of quantum steering in two-qubit states by deriving analytical and numerical results showing that the probability of observing steering increases systematically with the number of measurements, substantially exceeding Bell nonlocality probabilities and approaching certainty for generic states as the number of settings grows.

Original authors: Gerard Anglès Munné, Paweł Cieśliński, Tamás Vértesi, Wiesław Laskowski

Published 2026-07-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Gerard Anglès Munné, Paweł Cieśliński, Tamás Vértesi, Wiesław Laskowski

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the quantum world as a giant, invisible dance floor where two particles, Alice and Bob, are holding hands. Sometimes, they are so perfectly linked that if Alice does a little spin, Bob spins instantly, no matter how far apart they are. Scientists call this "entanglement." But there's a specific, tricky kind of connection called quantum steering. It's like Alice has a magic remote control: if she presses a button (makes a measurement), she can actually force Bob's particle into a specific state. If Bob can't explain what's happening using a simple "cheat sheet" of pre-agreed instructions (a local hidden state model), then Alice has successfully "steered" him.

The big question this paper asks is: How often does this magic actually happen?

The Great Quantum Dice Roll

To find out, the authors didn't just look at one special pair of particles. Instead, they imagined a massive factory churning out millions of random two-particle states. They treated these states like lottery tickets. Then, they simulated Alice and Bob performing random "moves" (measurements) on these particles, like rolling dice to decide which direction to look.

They wanted to know: If you pick a random pair of particles and roll the dice for measurements, what are the odds that Alice can actually steer Bob?

The Surprising Results: Steering is Everywhere!

Here is the cool part: Steering is way more common than you might think.

The paper shows that for many types of random particle pairs, steering happens surprisingly often. In fact, it's much more likely to happen than another famous quantum phenomenon called "Bell nonlocality" (which is like the ultimate proof that the universe isn't just following a script). The authors found that the chance of seeing steering is at least ten times higher than the chance of seeing Bell nonlocality.

Think of it like this: If Bell nonlocality is a rare, golden ticket found in one out of a hundred chocolate bars, quantum steering is a silver ticket you might find in ten of them.

The "Environment" Factor

The paper also looked at how "messy" the particles are. In the real world, particles often get dirty or mixed up with their surroundings (like a clean sock getting lost in a pile of laundry). The authors simulated different levels of this "messiness."

The Cleanest Pairs (Minimal Coupling): When the particles are kept very clean and isolated from the environment, steering becomes almost guaranteed if you try enough different measurements. If you keep rolling the dice (increasing the number of measurement settings), the probability of steering climbs up and up, eventually reaching 100%. It becomes a rule rather than an exception.
The Messy Pairs: Even for particles that are quite mixed up (highly mixed states), steering still happens a decent amount of the time, though it's not as guaranteed as the clean ones.

The "Werner" Special Case

The authors also did some heavy math to solve a specific, famous type of particle pair called a "Werner state." They figured out exact formulas for how likely steering is when you use 2 or 3 measurement settings.

With 2 settings, steering only happens if the particles are "clean" enough (specifically, if a value called visibility $v$ is greater than $1/\sqrt{2}$ ).
With 3 settings (restricted to a flat plane), the math gets a bit more complex, involving integrals and angles, but the result is the same: more settings mean a better chance of steering.

What They Didn't Find (The "No" List)

It's important to know what this paper doesn't say.

It does not say that every random pair of particles will steer. For very messy states or very few measurements, steering often doesn't happen.
It does not claim that steering is easy to spot with just two measurements for most messy states. In fact, for the "messiest" random states (the Hilbert-Schmidt ensemble), the chance of steering with just 2 measurements is almost zero. You need to try more measurements (like going up to 10) to see the steering kick in.
The paper explicitly rules out the idea that steering is as rare as Bell nonlocality. Their simulations show the gap is huge.

How Sure Are They?

The authors are very confident in their numbers, but with a specific caveat:

For the general random states, they didn't prove it with a single mathematical formula for every case. Instead, they ran computer simulations on one million ( $10^6$ ) random samples. They say the results are statistically solid, with a tiny margin of error (about 0.1%).
For the specific "Werner states," they proved the math analytically (using pure equations), so those numbers are exact.
They found that as you add more measurement settings (going from 2 up to 10), the probability of steering goes up systematically.

The Bottom Line

If you grab a random pair of quantum particles and start playing with them using random measurements, you have a very good chance of discovering that one can steer the other. It's a robust, common feature of the quantum world, especially if you keep trying different angles. While it's not always there (especially if the particles are very messy and you only try a couple of moves), it is far more typical and reliable than the "super-rare" phenomenon of Bell nonlocality. The more you look, the more you find that the universe is indeed steering itself.

Technical Summary: Typicality of Steering for Two-qubit States

Problem Statement
Quantum steering is a form of nonclassical correlation where measurements performed by one party (Alice) influence the conditional states of another (Bob) in a way that cannot be explained by a Local Hidden State (LHS) model. While the typicality of Bell nonlocality has been extensively studied for generic states and random measurements, the typicality of quantum steering remains less characterized. Specifically, it is unclear how likely it is to observe steering in generic two-qubit mixed states when subjected to a finite number of Haar-random projective measurements. Unlike pure states, where entanglement and steering are nearly certain under random measurements, mixed states depend heavily on the chosen ensemble and the degree of environmental coupling. This work investigates the probability ( $P_S$ ) of observing steering across various mixed-state ensembles and measurement settings ( $m$ ).

Methodology
The authors employ a framework of randomized measurements to assess steering typicality.

State Ensembles: The study focuses on mixed two-qubit states generated via the partial trace of a pure Haar-random state in a higher-dimensional space ( $H_d \otimes H_k$ $H_{d} \otimes H_{k}$ ). This defines ensembles $E_k$ $E_{k}$ , where $k$ $k$ represents the dimension of the ancilla.
- $k=2$ : Minimal environmental coupling (rank-2 states).
- $k=3, 4$ : Increasingly mixed states, with $k=4$ corresponding to the Hilbert-Schmidt ensemble.
- Specialized families: Bell-diagonal states and Werner states ( $\varrho_W(v)$ ) are also analyzed.
Measurements: Alice and Bob perform $m$ local two-outcome projective measurements, sampled independently from the Haar measure on $SU(2)$.
Steering Detection: The existence of an LHS model is tested using Semidefinite Programming (SDP). For a given assemblage, the SDP maximizes a parameter $\mu$ subject to constraints derived from the LHS decomposition. If the optimal $\mu < 0$ , the state is deemed steerable.
Scenarios: The authors distinguish between one-way steering ( $A \to B$ or $B \to A$ ) and two-way steering ( $S_{AB}$ ), as well as the unsteerable case ( $S_\emptyset$ ).

Key Contributions and Results

Analytical Expressions for Werner States:
The paper derives closed-form analytical expressions for the steering probability of Werner states:
- Two-setting scenario ( $m=2$ ): $P_S(v) = \sqrt{2}v^2 - 1/v^2$ for $v > 1/\sqrt{2}$ , and 0 otherwise. This recovers the known critical visibility $v_{crit} = 1/\sqrt{2}$ .
- Three-setting scenario ( $m=3$ ): Restricted to coplanar measurements on the Bloch sphere, the authors provide integral expressions for $P_S(v)$ , identifying a critical visibility of $v_{crit} = 2/3$ for this specific configuration.
Numerical Analysis of Generic Ensembles:
Using $10^6$ random samples per data point, the authors map $P_S$ as a function of $m$ (up to $m=10$ ) for ensembles $E_2, E_3,$ and $E_4$ :
- Dependence on $m$ : $P_S$ increases systematically with the number of measurement settings.
- Dependence on $k$ : Ensembles with minimal environmental coupling ( $k=2$ ) exhibit the highest steering probabilities. For $k=2$ and $m=10$ , the probability of two-way steering ( $S_{AB}$ ) reaches $\approx 63\%$ , while the unsteerable fraction ( $S_\emptyset$ ) drops to $\approx 15\%$ .
- Asymptotic Behavior: For $k=2$ , the authors argue that since almost all states are rank-2, they are steerable in the limit $m \to \infty$ , implying $P_S \to 100\%$ . For $k=4$ (Hilbert-Schmidt), steering is negligible at $m=2$ but rises to $\approx 10\%$ for $S_{AB}$ at $m=10$ .
Comparison with Bell Nonlocality:
A critical finding is the quantitative gap between steering and Bell nonlocality. For the Hilbert-Schmidt ensemble, the probability of observing steering is at least one order of magnitude larger than the probability of violating a Bell inequality across all tested $m$ . For instance, at $m=10$ , steering probability is $\approx 14.9\%$ (total steerable), whereas Bell nonlocality is only $\approx 0.2\%$ .
Specialized Families:
- Bell-diagonal states: The asymptotic steering probability ( $m \to \infty$ ) is estimated at $24.15\%$ .
- Noisy Pure States: The authors analyze Haar-random pure states subjected to depolarizing noise and separable noise. They demonstrate that for states with separable noise defined as $\varrho_{AB}(v) = v |\psi_H\rangle\langle\psi_H| + \frac{1-v}{2}(I \otimes \text{tr}_A|\psi_H\rangle\langle\psi_H|)$ , the steering probability from Alice to Bob matches that of the Werner state exactly, due to local filtering equivalence.

Significance and Claims
The paper claims to establish that quantum steering is a "typical" phenomenon for generic two-qubit states, provided a sufficient number of random measurement settings are available.

Prevalence: Steering is shown to be significantly more prevalent than Bell nonlocality, suggesting that steering is a more robust resource for quantum information protocols involving mixed states.
Resource Identification: The work provides a detailed characterization of steering probabilities across different ensembles, identifying those with minimal environmental coupling ( $k=2$ ) as having the highest potential for steering.
Methodological Extension: By moving beyond fixed steering inequalities to a general SDP-based approach with random measurements, the study offers a more comprehensive view of steering typicality, extending previous results on Bell nonlocality.

The authors conclude that while highly mixed states can exhibit steering, it becomes a dominant feature only in ensembles with minimal environmental coupling and a sufficient number of settings. In the asymptotic limit of infinite settings, steering approaches 100% typicality for minimally coupled systems.