New results for Heisenberg dynamics for non self-adjoint Hamiltonians

Here is an explanation of the paper using simple language, everyday analogies, and metaphors.

The Big Picture: A World Where "Rules" Change

Imagine you are watching a movie of a quantum system (like a tiny particle or a group of atoms). In the standard world of physics (where the rules are "Hermitian" or "self-adjoint"), the movie plays out perfectly. The total amount of "stuff" (probability) stays the same. If you multiply two things together, the order doesn't matter, and the math is clean and predictable.

But this paper explores a strange, alternative universe where the "Hamiltonian" (the rulebook that tells the system how to move) is non-Hermitian.

In this strange universe:

The movie gets blurry: The total "amount" of the particle (its wave function) might grow infinitely large or shrink to nothing as time passes. It's like a video that gets brighter and brighter until it's just white noise, or fades to black.
The math gets messy: If you take two variables, say $X$ and $Y$ , and watch them evolve over time, the result of watching them evolve together is not the same as watching them evolve separately and then multiplying them. It's like trying to bake a cake by baking the flour and the eggs separately, then mixing them, versus mixing them first and then baking. The results are different.

The Problem: How Do We Watch the Movie?

The author, Fabio Bagarello, asks a crucial question: If the movie is getting blurry (the numbers are exploding or vanishing), how do we make sense of what's happening?

In the real world, if a video gets too bright, we don't throw it away; we adjust the brightness (normalize it) so we can still see the actors.

The paper proposes a new way to look at these systems:

Old Way: Watch the raw, unadjusted wave function $\Psi(t)$ . (This leads to the "exploding" or "vanishing" problem).
New Way: Watch the normalized version, $\hat{\Psi}(t)$ . This is the raw wave function, but we constantly "turn down the volume" or "adjust the contrast" so the total probability always equals 100%.

The Twist: The Rules Become "Non-Linear"

Here is the surprising part. When you force the system to stay normalized (like keeping a balloon at a fixed size while air is being pumped in or out), the rules of the game change.

The Linear Hamiltonian: In the old world, the rulebook ( $H$ ) is a fixed machine. It treats every particle the same, regardless of how many particles are there.
The Non-Linear Hamiltonian: In this new, normalized world, the rulebook becomes self-aware. The new rulebook ( $H_{nl}$ ) looks at the current state of the system and changes its own rules based on what it sees. It's like a traffic light that changes its timing based on how many cars are currently at the intersection.

This makes the math much harder (non-linear), but it also reveals something magical.

The Discovery: Hidden Treasures (Conserved Quantities)

In the messy, non-Hermitian world, we usually expect chaos. We expect nothing to stay the same. But the author finds that when you look at the normalized system, some things actually do stay constant.

Think of it like a chaotic dance party where everyone is running around, changing partners, and the music is speeding up and slowing down.

If you look at the raw data, it's a mess.
But if you look at the average behavior of the dancers (the normalized view), you might notice that the total number of people on the dance floor never changes, even though individuals are constantly entering and leaving.

The paper proves that there are specific "observables" (things we can measure) that remain constant in time, even in this chaotic, non-Hermitian environment, provided we look at them through the lens of the normalized state.

The Example: The Decision-Making Agents

To prove this isn't just abstract math, the author uses a model from Decision Making. Imagine three agents (people) making choices.

They can be in state 0 (No) or state 1 (Yes).
They influence each other.
The math describing their choices is non-Hermitian (it's not perfectly balanced).

When the author simulates this:

The individual choices of the agents change wildly over time.
However, the sum of their choices (the total "Yes" votes) turns out to be a constant. It doesn't matter how the system evolves; the total number of "Yes" votes remains locked at a specific number.

This is a "Conserved Quantity." It's a hidden rule that survives the chaos.

Why Does This Matter?

New Physics: It suggests that in systems where energy isn't perfectly conserved (like open systems interacting with an environment), there might be other things that are conserved if we look at them the right way.
Better Models: It helps us model real-world things like biological systems, economic markets, or decision-making processes, where things aren't perfectly balanced (Hermitian) but still follow hidden patterns.
Mathematical Challenge: The paper admits that the math is tricky. The "time evolution" of these normalized systems doesn't follow the standard rules (it's not a simple "automorphism"). It's a new kind of dance that mathematicians are just starting to learn the steps to.

Summary in One Sentence

This paper shows that even in a chaotic, unbalanced quantum world where things usually explode or vanish, if you constantly "normalize" your view (keep the scale steady), you can discover hidden, unchanging laws that govern the system's behavior.

Here is a detailed technical summary of the paper "New results for Heisenberg dynamics for non self-adjoint Hamiltonians" by F. Bagarello.

1. Problem Statement

The paper addresses the mathematical and physical challenges associated with Heisenberg dynamics generated by non-Hermitian (non-self-adjoint) Hamiltonians ( $H \neq H^\dagger$ ).

In standard quantum mechanics with Hermitian Hamiltonians, the time evolution of observables preserves the algebraic structure (automorphism), and the norm of the state vector is conserved. However, when $H \neq H^\dagger$ :

The standard time evolution of a product of observables, $(XY)(t)$ , is generally not equal to the product of the time-evolved observables, $X(t)Y(t)$ .
The norm of the wave function $\Psi(t) = e^{-iHt}\Psi(0)$ is not preserved; it may explode or decay to zero over time.
Consequently, the standard definition of conserved quantities (integrals of motion) and symmetries becomes problematic.

The author investigates a specific approach to resolve these issues: analyzing dynamics not on the raw wave function $\Psi(t)$ , but on the normalized wave function $\hat{\Psi}(t) = \Psi(t) / \|\Psi(t)\|$ . This introduces a non-linear dependence into the dynamics, motivated by applications in decision-making models and the need to define meaningful physical observables (mean values) even when the system's norm is not conserved.

2. Methodology

The paper employs a rigorous operator-theoretic approach within a finite-dimensional Hilbert space $\mathcal{H}$ (dimension $N < \infty$ ).

A. Mathematical Framework

Standard $\gamma$ -dynamics: The author first reviews the standard Heisenberg-like evolution for non-Hermitian systems defined by the map $\gamma_t(X) = e^{iH^\dagger t} X e^{-iHt}$ .
- It is shown that $\gamma_t$ is not an automorphism of the algebra of bounded operators $B(\mathcal{H})$ unless $H = H^\dagger$ . Specifically, $\gamma_t(\mathbb{1}) \neq \mathbb{1}$ generally, and $\gamma_t(XY) \neq \gamma_t(X)\gamma_t(Y)$ .
- The generator of this dynamics is the $\gamma$ -derivation: $\delta_\gamma(X) = i(H^\dagger X - XH)$ .
Normalized Dynamics ( $\hat{\Psi}$ -dynamics): The core of the paper shifts to the dynamics of the normalized state $\hat{\Psi}(t)$ .
- The Schrödinger equation for $\hat{\Psi}(t)$ is derived, revealing that it obeys a non-linear Schrödinger equation:
  $i \frac{d}{dt}\hat{\Psi}(t) = H_{nl}(t)\hat{\Psi}(t)$
  where $H_{nl}(t) = H + \frac{1}{2}\langle \hat{\Psi}(t), (H^\dagger - H)\hat{\Psi}(t) \rangle$ .
- This effective Hamiltonian $H_{nl}(t)$ is explicitly time-dependent and state-dependent (non-linear).

B. Definition of Conserved Quantities

The author defines two distinct sets of "integrals of motion" based on the behavior of the mean value $x_{\hat{\Psi}}(t) = \langle \hat{\Psi}(t), X \hat{\Psi}(t) \rangle$ :

$\hat{\Psi}$ -integral of motion: An operator $X$ such that the operator-valued derivation vanishes: $\delta_{\hat{\Psi}}(X; t) = 0$ .
Weak $\hat{\Psi}$ -integral of motion: An operator $X$ such that the mean value of the derivation vanishes: $\langle \hat{\Psi}(t), \delta_{\hat{\Psi}}(X; t)\hat{\Psi}(t) \rangle = 0$ .

The derivation $\delta_{\hat{\Psi}}$ is defined as:
$\delta_{\hat{\Psi}}(X; t) = i(H_{nl}^\dagger(t)X - X H_{nl}(t)) = \delta_\gamma(X) - iX \langle \hat{\Psi}(t), (H^\dagger - H)\hat{\Psi}(t) \rangle$

3. Key Contributions and Results

A. Non-Linearity and Time Dependence

The paper establishes that replacing $\Psi(t)$ with $\hat{\Psi}(t)$ fundamentally alters the dynamics. The generator $\delta_{\hat{\Psi}}$ is:

Time-dependent: Unlike the linear case where $\delta_\gamma$ is constant.
State-dependent: It depends on the instantaneous state $\hat{\Psi}(t)$ .
Non-automorphic: Even in the weak sense, the time evolution of a product of observables does not equal the product of their evolutions.

B. Properties of Conserved Quantities

Identity Operator: The identity operator $\mathbb{1}$ is not a $\hat{\Psi}$ -integral of motion (since $\delta_{\hat{\Psi}}(\mathbb{1}; t) \neq 0$ ), but it is a weak $\hat{\Psi}$ -integral of motion because its mean value is always 1 (and its derivative is 0).
Symmetry Relations: The paper proves that if $X$ is a weak integral of motion, then $X^\dagger$ is also a weak integral of motion. Furthermore, if $X$ is a $\hat{\Psi}$ -integral of motion, then $H^\dagger X$ , $XH$ , $X^\dagger H$ , etc., are also $\hat{\Psi}$ -integrals of motion.
Special Case (Eigenstates): If the initial state $\hat{\Psi}(0)$ is an eigenstate of $H$ (with eigenvalue $E_{k_0}$ ), the derivation simplifies to $\delta_{\hat{\Psi}}(X; t) = \delta_\gamma(X) - 2E^{(i)}_{k_0}X$ (where $E^{(i)}$ is the imaginary part). In this specific case, the series expansion of the time evolution converges to a well-defined operator $\gamma_t^{\hat{\Psi}}(X)$ .

C. Concrete Example: Decision Making Model

The author analyzes a specific 3-fermion system (originally from a decision-making model) with Hamiltonian $H = b_1^\dagger(\lambda b_2 + \mu b_3)$ .

Observation: While individual occupation numbers $n_j(t) = \langle \hat{\Psi}(t), N_j \hat{\Psi}(t) \rangle$ evolve non-trivially, their sum $N = N_1 + N_2 + N_3$ remains constant.
Verification: The paper explicitly calculates $\delta_{\hat{\Psi}}(N; t)$ and shows that while the operator itself is non-zero, its expectation value on the state $\hat{\Psi}(t)$ is zero: $\langle \hat{\Psi}(t), \delta_{\hat{\Psi}}(N; t)\hat{\Psi}(t) \rangle = 0$ .
Significance: This confirms that the total number operator is a weak $\hat{\Psi}$ -integral of motion, validating the theoretical framework with a concrete physical model where $H \neq H^\dagger$ .

4. Significance and Implications

Redefining Conservation: The paper argues that for non-Hermitian systems, the concept of a "conserved quantity" must be relaxed from an operator-level constant to a weakly conserved mean value. This is crucial for physical systems where the norm is not preserved (e.g., open systems, effective non-Hermitian descriptions).
Mathematical Rigor: It provides a systematic framework for handling the non-linearity introduced by normalization, distinguishing between strong (operator) and weak (expectation value) conservation laws.
Applications: The results are directly applicable to fields using non-Hermitian dynamics, such as:
- Decision Making: Where the model in Section IV originates.
- Open Quantum Systems: Where effective non-Hermitian Hamiltonians describe dissipation.
- PT-Symmetric Systems: Though not explicitly the focus, the isospectral conditions discussed are relevant.
Open Problems: The paper highlights that the series expansion $\sum \frac{t^k}{k!} \delta_{\hat{\Psi}}^k(X; t)$ converges, but it remains an open question whether this series always represents the physical time evolution of $X$ in the general non-linear case, and how to define a generalized Heisenberg equation for these systems.

In conclusion, Bagarello demonstrates that by shifting the focus to normalized states, one can recover meaningful conservation laws (specifically weak integrals of motion) in non-Hermitian dynamics, offering a robust mathematical tool for analyzing complex, non-conservative quantum-like systems.