Deriving the Lindblad Master Equation

Introduction

In our previous lessons, we've used the Kraus operator-sum representation to describe the impact of discrete noise events on a qubit. This framework is powerful but describes the evolution in discrete steps. However, many physical processes, like the spontaneous emission that causes amplitude damping, happen continuously over time. To model such phenomena, we need a differential equation that governs the evolution of the density matrix $\rho (t)$.

This lesson focuses on deriving the most important equation for continuous-time open quantum system dynamics: the Lindblad master equation. Our learning outcome is to derive this equation for systems undergoing Markovian noise.

Your background in physics, particularly with differential equations for modeling physical systems, will be a great asset here. We will approach the derivation from three complementary perspectives to build a robust understanding:

1. From Infinitesimal Kraus Maps: A direct mathematical path that shows how the Lindblad equation arises as the continuous-time limit of the Kraus representation we've already studied.
2. From Microscopic Physics: A fundamental "first-principles" derivation starting with a system-environment Hamiltonian, which reveals the physical approximations required to arrive at the Lindblad form.
3. From Quantum Jumps: An intuitive, physical picture that "unravels" the density matrix evolution into individual quantum trajectories, connecting the abstract equation to observable phenomena.

By the end of this lesson, you will understand not just the form of the Lindblad equation, but also the physical assumptions that underpin it and the meaning of its different terms.

The Markovian Approximation: A Memoryless Environment

The ability to write a first-order differential equation for the density matrix, of the form $\dot{\rho }(t)=\mathcal{L}[\rho (t)]$, hinges on a crucial assumption: the evolution is Markovian. This means that the state of the system at time $t+dt$ depends only on its state at time $t$, and not on its history at earlier times. In essence, the environment has no "memory" of its past interactions with the system.

As the notes explain, the Markovian approximation is valid when there's a clear separation of timescales. If the environment's internal dynamics are much faster than the system's dynamics, any information the environment gains from the system is quickly dissipated within the environment itself. The environment essentially "forgets" the interaction before it can influence the system's future evolution. This allows us to model the environment as a memoryless bath that continuously induces decoherence.

Derivation 1: From Infinitesimal Kraus Maps

The most direct way to derive the Lindblad equation is to start from the Kraus representation and consider an infinitesimal time evolution from $t$ to $t+\delta t$.

Let's summarize the logic presented in the notes.
The evolution over a small time step $\delta t$ is given by a Kraus map:
$\rho (t+\delta t)={\sum }_k{M}_k(\delta t)\rho (t)M_k^{†}(\delta t)$

For this to be a continuous evolution, one of the Kraus operators, say $M_0$, must be close to the identity matrix $I$, representing the high probability of "nothing happening". The other operators, $M_k$ for $k>0$, must be small, representing the low probability of a "quantum jump". We can parameterize this as:

• $M_0(\delta t)=I+(-iH+K)\delta t$
• $M_k(\delta t)=\sqrt{\delta t}{L}_k$ for $k>0$

Here, $H$ is a Hermitian operator that will become the system Hamiltonian. The operators $L_k$ are the Lindblad operators or jump operators. The operator $K$ is determined by the trace-preserving condition ${\sum }_k{M}_k^{†}{M}_k=I$. Expanding this to first order in $\delta t$ gives:

For this to hold, the $\delta t$ term must be zero. Assuming $H$ is Hermitian ($H^{†}=H$), this simplifies to $K+K^{†}={\sum }_{k>0}{L}_k^{†}{L}_k$. A standard choice that makes $M_0$ a well-behaved operator is to choose $K$ to be Hermitian, which gives:

(Note: The resource LINK uses a slightly different convention where $K$ is anti-hermitian, but the final result is identical. We use the more common convention here).

Now, substitute these forms for $M_k$ back into the evolution equation for $\rho (t+\delta t)$, keep terms up to first order in $\delta t$, and rearrange to find $\dot{\rho }={\lim }_{\delta t\to 0}\frac{\rho (t+\delta t)-\rho (t)}{\delta t}$. This yields the Lindblad master equation:

where $\{A,B\}=AB+BA$ is the anticommutator. This equation is the cornerstone of modeling Markovian open quantum systems.

Derivation 2: From Microscopic Physics

The previous derivation was mathematically clean but didn't explain where the operators $H$ and $L_k$ come from. A more fundamental approach starts from the total Hamiltonian of the system plus its environment, $H_T=H_S\otimes I_E+I_S\otimes H_E+H_I$, and systematically derives an effective equation for the system alone.

This derivation is quite involved, relying on a series of well-controlled approximations. Given your preference for original sources, we will use a peer-reviewed paper that walks through this derivation in detail.

Let me summarize the key logical steps and the physical meaning of the approximations you just read about:

1. Starting Point: The von Neumann equation for the total system+environment, ${\dot{\rho }}_T=-i[H_T,{\rho }_T]$.
2. Interaction Picture: The dynamics are moved into the interaction picture, where evolution is driven only by the interaction Hamiltonian $H_I$.
3. Born Approximation: Assumes weak coupling between the system and environment. This allows us to assume the total density matrix remains approximately separable at all times: ${\rho }_T(t)\approx {\rho }_S(t)\otimes {\rho }_E$. It justifies treating the environment as a static reservoir, unaffected by the system.
4. Markov Approximation: This is the crucial step that introduces memorylessness. The derivation shows that the change in ${\rho }_S(t)$ depends on an integral over its past history. The Markov approximation assumes the environment's correlation functions decay much faster than the system's dynamics, allowing the integral over time from $0$ to $t$ to be extended to $\infty$. This makes the evolution of ${\rho }_S$ depend only on its state at the current time $t$, yielding a true differential equation (the Redfield equation).
5. Rotating Wave Approximation (RWA): This final step neglects rapidly oscillating terms in the equation. Physically, it corresponds to ignoring processes that do not conserve energy. Mathematically, it is required to ensure that the resulting dynamical map is completely positive, a necessary condition for any physical evolution. This simplification transforms the Redfield equation into the Lindblad master equation.

This derivation is powerful because it shows how the phenomenological Lindblad equation is connected to an underlying microscopic reality, and it makes explicit the physical conditions under which it is valid: weak coupling and a fast-relaxing environment.

Derivation 3: The Quantum Jump Perspective

The third approach provides a compelling physical intuition for the two parts of the dissipative term in the Lindblad equation. It "unravels" the ensemble-averaged evolution of the density matrix into stochastic trajectories of single quantum wave functions.

This perspective was pioneered by experimentalists who were beginning to observe individual quantum systems, like single trapped ions, exhibiting sudden "quantum jumps" (e.g., turning on and off of fluorescence).

Let's distill the key insights from the video:

• Evolution is a choice: At every infinitesimal time step $\delta t$, the system faces a probabilistic choice. For a spontaneous emission process with rate $\gamma$ and an excited state population ${\rho }_{11}$, two things can happen:

  1. A "jump" occurs: With a small probability $\delta p=\gamma {\rho }_{11}\delta t$, the environment effectively measures the system. A photon is emitted, and the state collapses. For spontaneous emission, the jump is described by the operator $L=\sqrt{\gamma }{\sigma }^{-}$, so the state becomes $\frac{L\psi }{\Vert L\psi \Vert }$.
  2. No jump occurs: With high probability $1-\delta p$, no photon is emitted. This absence of a measurement is also information! It tells us the system is now less likely to have been in the excited state. This knowledge is incorporated by evolving the state with a non-Hermitian effective Hamiltonian, $H_{\text{eff}}=H-\frac{i}{2}{L}^{†}L$. The imaginary part causes the amplitude of the excited state to decay.

• Rebuilding the Master Equation: The evolution of the ensemble density matrix $\rho$ is the weighted average of these two outcomes:

  $$\rho (t+\delta t)=(1-\delta p)\left(\frac{e^{-i{H}_{\text{eff}}\delta t}\rho (t)e^{i{H}_{\text{eff}}^{†}\delta t}}{\text{Tr}(...)}\right)+\delta p\left(\frac{L\rho (t)L^{†}}{\text{Tr}(...)}\right)$$

  Expanding this expression to first order in $\delta t$ and using the forms of $\delta p$ and $H_{\text{eff}}$ leads exactly back to the Lindblad master equation!

This unraveling provides a beautiful interpretation of the Lindblad equation's terms:

• ${\sum }_k{L}_k\rho L_k^{†}$: This corresponds to the incoherent "jumps" that can occur.
• $-i[H,\rho ]$: This is the standard coherent evolution.
• $-\frac{1}{2}{\sum }_k\{L_k^{†}{L}_k,\rho \}$: This term arises from the non-Hermitian evolution for the cases where no jump occurs. It ensures that the total probability is conserved.

Conclusion

In this lesson, we have derived the Lindblad master equation, the primary tool for describing the continuous-time evolution of Markovian open quantum systems. By approaching it from three different angles, we have built a multi-faceted understanding.

Key Takeaways:

• The Lindblad master equation $\dot{\rho }=-i[H,\rho ]+{\sum }_k(L_k\rho L_k^{†}-\frac{1}{2}\{L_k^{†}{L}_k,\rho \})$ governs the time evolution of a density matrix under Markovian noise.
• The derivation is contingent on the Markovian approximation, which is physically justified when the environment's correlation time is much shorter than the system's characteristic evolution time.
• The equation can be derived rigorously from microscopic physics (system-bath Hamiltonian with Born-Markov-RWA approximations) or more directly from the properties of infinitesimal Kraus maps.
• The quantum jump picture provides a powerful intuition: the Lindblad equation represents an ensemble average over stochastic quantum trajectories, each consisting of periods of coherent evolution punctuated by random "jumps".

Preview of the next lesson:

We now have the formal tools to describe continuous noise processes. A crucial task in experimental quantum computing is to quantify the impact of this noise on our quantum gates. In the next lesson, we will learn about single-qubit randomized benchmarking, a powerful and widely used experimental protocol that allows us to extract the average fidelity of our gates, effectively measuring the strength of the noise processes we've been studying.