Quantum Optimal Control without Arbitrary Waveform Generators
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.
Kai: We're looking at the paper: "Quantum Optimal Control without Arbitrary Waveform Generators".
Paper summary: Kai: So looking at the whole paper, "Quantum Optimal Control without Arbitrary Waveform Generators," the authors are proposing a way to achieve arbitrary control using just a sequence of simple on and off pulses.
Mira: The main thing they claim is that this method offers simple, precise, robust control for systems that are really hard to manage with arbitrary waveform generators.
Lev: They show how you can design these switching instances using standard quantum optimal control algorithms while drastically reducing the computational resources needed for matrix exponentials by using this pulsed time propagation technique.
Kai: It means they’re proposing a scalable protocol, and they apply it to superconducting quantum circuits, showing fidelities like nine point nine nine nine with their optimized pulse trains.
Mira: The implication is that this method could be a practical way to realize universal control for superconducting qubits without needing the heavy machinery of arbitrary waveform generators.
Lev: What I see is that the performance evaluation is fast because they use matrix decomposition instead of slow matrix exponentiation, which makes it viable for real hardware implementation.
Kai: So, in simple terms, they’re replacing complex waveform design with a sequence of simple pulses that works well and is computationally cheap.
Mira: The paper suggests this protocol could open new ways for seeking time-optimal control pulses in general quantum systems by making the control process much simpler to implement.
Conclusion: Kai: So, this paper is about getting arbitrary control over quantum systems without needing those complicated arbitrary waveform generators we usually have to build.
Mira: Right, it’s titled "Quantum Optimal Control without Arbitrary Waveform Generators." The authors are showing that you can get what you want just by turning fields on and off in a specific order.
Lev: That’s the core idea—using a simple pulse train instead of trying to design a perfect continuous waveform. It bypasses those huge calculations for propagation operators.
Kai: So, they found an equivalence between an arbitrary waveform and these PWM pulse trains, which is pretty neat for making things simpler on the hardware side.
Mira: What's interesting is how they do it—they show that the Fourier components are almost identical below a certain frequency threshold, which makes the pulse train look like the original signal in practice.
Lev: And from an engineering standpoint, they avoid those numerically expensive matrix exponentials by using a finite set of Hamiltonians instead. That’s a real win for running this on actual processors.
Kai: So, they conclude that this method is simple, precise, robust to things like time jitter and switching delays, and scalable for building bigger quantum systems.
Mira: It means we don't have to be so tied down by perfect continuous control signals anymore if we just use these sequences of on and off pulses.
Lev: It’s a protocol that could really speed up how we think about finding time-optimal control pulses in general quantum systems.
Qi-Ming Chen, *Herschel Rabitz, †Re-Bing Wu
QCD Labs · Princeton University · Tsinghua University
quant-ph
Submitted: 2022-09-20
Updated: 2022-09-20
DOI: 10.1103/PhysRevApplied.20.064016
Code: https://github.com/chenqmion/QPWM
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 91/100
The gist: Simple, precise, and robust control is demanded for operating a large quantum information processor Here…
Key concepts
- Waveform-Pulse Train Correspondence
- This concept establishes that any complex quantum control signal (waveform) can be perfectly replicated by a sequence of simple on/off pulses. The authors show that below a certain frequency threshold, the Fourier components of the arbitrary waveform and the PWM pulse train are essentially identical, meaning they produce nearly the same effect.
- Pulsed Time Propagation
- Instead of calculating complex propagation operators, this method treats pulse widths as control variables. By choosing pulses to be zero or $\pm\xi_k$, the system Hamiltonian at any moment is drawn from a small, finite set. This allows for efficient calculation using pre-diagonalized matrices instead of computationally expensive matrix exponentials.
- Numerical Efficiency
- The PWM method offers a computational advantage in simulations because the resulting time propagation operator belongs to a finite set. This means calculations can use diagonal matrix exponents instead of full matrix exponentials, significantly speeding up the simulation compared to conventional design efforts, especially for larger systems.
Terminology
Summary
Simple, precise, and robust control is demanded for operating a large quantum information processor Here, arbitrary control of a quantum system can be achieved by simply turning on and off the control fields in a proper sequence This method is motivated by the pulse width modulation (PWM) technique in classical control systems The authors reveal the equivalence between an arbitrary waveform and a PWM pulse train These results indicate a simple, precise, robust, and scalable quantum control protocol which may greatly facilitate the building of large controlled quantum systems
Waveform-Pulse Train Correspondence
The paper establishes that the control of a general quantum system can be realized by simply turning on and off the driving fields at prescribed time instances This method is motivated by the pulse width modulation (PWM) technique in classical control systems The authors generalize this idea to quantum-mechanical systems and reveal the equivalence between an arbitrary waveform and a PWM pulse train Specifically, they define a PWM sequence as s1(t) = X M m=1 ξk θ t − τ on k,m − θ t − τ off k,m (1) The Fourier components of the two fields are almost identical below a prescribed threshold, omegak ≈ 2π/τ, in frequency This equivalence is demonstrated by showing that the PWM sequence results in a 2nd-order approximation to the ideal unitary propagation, i.e., δU[mτ,(m − 1)τ] ∝ O τ3
Pulsed Time Propagation
To design a PWM sequence for high-fidelity quantum control, one may simply treat the pulse widths, τk,m, as control variables and resort to numerical QOC algorithms The PWM approach circumvents the technical challenge of calculating propagation operators by using exact matrix decomposition Because sk(t) can be either zero or ±ξk, the system Hamiltonian at any time instance is chosen from a finite set H(t) ∈ 3 (3) Thus, one may diagonalize these Hamiltonians in advance and convert the matrix exponentials into scalar exponentials and matrix products This diagonalization process describes the change of basis for representing the matrices, H0 and H0 ± ξ1H1, such that they are always written in a diagonal form The pulsed time propagation can be efficiently calculated for a relatively small number of control degrees of freedom, i.e. – 10
Numerical Efficiency
In numerical simulation, the replacement, uk(t) → sk(t), has an advantage that the resulting generator of time propagation belongs to a finite set H(t) ∈ 3 (S13) This observation indicates that the numerically expensive calculation of matrix exponents in conventional design efforts can be circumvented with diagonal matrix exponents instead The computational time is measured by using the perf counter command in the time package, and the matrix product is calculated by the @ operator The PWM method outperforms the default method for N ≤ 4, K. 10 and N ≥ 5, K. 5
Generalizations
The PWM control protocol can be generalized in several ways Examples of generalizations include:
-
The resulting binary sequence is called the 2-level PWM pulse sequence in the literature, which keeps the signal power unchanged and alters only the polarization of the switch
-
One can also define a n-level PWM sequence with n discrete polarizations of sk(t)
-
The pulse magnitude ξk can be stretched by a factor of κ and shrunk by 1/κ at the same time, while keeping the control performance almost unchanged
The authors also investigate Gaussian pulses, which can be used in real implementations instead of perfect rectangular pulses
Application to SQC
For illustration, the authors consider a chain of superconducting artificial atoms with XYcontrol, un,x/y(t) for n = 1, · · ·, N They evaluate the control performance by the average fidelity over all the relevant qubit states J(T) = tr U(T, 0)P U† (T, 0)P + tr U† gU(T, 0)P / 2N (2N + 1), The optimized pulse trains and the converted staircase waveforms are shown in Fig. 2(a) top and middle, respectively, of which the fidelities are 0. p. 39999 and 0. p. 39848 These examples demonstrate the potential of using a PWM control protocol for realizing the universal control of superconducting qubits
Conclusions and Outlook
The authors propose the PWM method that enables arbitrary control of a general quantum system with a sequence of simple pulses The performance of the pulse train can be efficiently evaluated by a matrix decomposition method, which avoids numerically expensive matrix exponentials and accelerates the TDSE solver for a relatively small number of controls The control protocol is robust to time jitter, switching delay, and leakage These results indicate a simple, precise, robust, and scalable control protocol for a general quantum system They anticipate that the PWM method may open up new ways for seeking time-optimal control pulses in a general quantum system
SI. WAVEFORM-PULSE TRAIN CORRESPONDANCE
To demonstrate the equivalence between a waveform and a pulse train, they start from the simplest case where the former is a sinusoidal function of time, u1(t) = sin(ωt+φ) By utilizing a Fourier series, this equation can be equivalently written as s1(t) = X n6=0 h X M m=1 ξ1 τ1,m / Mτ e − inω (m− 1/2) τ i e inωt The frequency threshold is omega1 ≈ Mωmin, below which the Fourier components of the pulse sequence, s1(t), and the waveform, u1(t), are indistinguishable from each other
SII. ERROR ANALYSIS
The propagation error is estimated as hδU[mτ,(m − 1)τ]i ≈ PK k=1 σ2 k ξ2 kH2 k / 2, where h·i is the ensemble average The time jitter leads to a 0.01% relative error that may be neglected for T ' 103 τ The Fourier components of uk(t) and sk(t) are exactly the same below the threshold omegak ≈ Mωmin, as long as τk,m = 1/ξk Z mτ (m−1)τ dtuk(t)
**SIV.
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Switching instances can be designed by conventional quantum optimal control algorithms while reducing computational resources for matrix exponential calculations to a substantially reduced amount. This allows for simpler and more efficient implementation of control protocols compared to relying solely on arbitrary waveform generators (AWGs).
-
The PWM approach circumvents the technical challenge of calculating propagation operators by allowing the system Hamiltonian at any time instance to be chosen from a finite set, such as H(t) in equation 3. This enables diagonalization of these Hamiltonians in advance, converting matrix exponentials into scalar exponentials and matrix products, which is significantly faster than scaling and squaring methods.
-
The control protocol is robust to time jitter and finite switching speed errors; the time jitter leads to a 0.01% relative error that may be neglected for T over 10 cubed tau, while finite switching speed distortions in Gaussian pulses are identical to the ideal case below a prescribed frequency threshold Omega k.
-
High-order accuracy can be achieved by inserting more pulses in each time interval, where the exact short-time propagation of uk(t) is written as Uu[m tau, (m - 1) tau] = S(m)2(tau) + C(m)3(tau) cubed + O(tau 4). This allows for the derivation of a 3rd-order accurate PWM sequence when the condition 2s cubed + (1 - 2s) cubed equals zero.
-
The system can implement arbitrary control of a general quantum system with a sequence of simple pulses, which may greatly facilitate the building of large controlled quantum systems. This protocol is simple, precise, robust, and scalable.
Related papers
- Computational regimes in matrix-product-state-based quantum trajectory simulations
- Toward chemically accurate quantum simulations on quantum hardware: A quantum-HPC hybrid approach
- Transversal surface-code game with reconfigurable qubits
- Fulqrum: a generalized framework for quantum subspace diagonalization
- Flux-Activated Resonant Control of a Bosonic Quantum Memory
- Machine learning the effects of many quantum measurements