EP-4742112-A1 - METHOD AND APPARATUS FOR GENERATING LOW-POWER QUBIT CONTROL SIGNAL FOR SUPERCONDUCTING QUANTUM COMPUTER CONTROL

EP4742112A1EP 4742112 A1EP4742112 A1EP 4742112A1EP-4742112-A1

Abstract

Disclosed are a method and an apparatus for generating a qubit control signal, wherein a main pulse and a DRAG pulse used to control a superconducting quantum computer are generated with low power and without a memory. The method comprises the steps of: generating a first sine curve having a first phase obtained by multiplying an angular velocity by a predetermined time of each period, the angular velocity being obtained by adding a first angular velocity of a center frequency and a second angular velocity at a predetermined time of each period; and generating a second sine curve having a second phase obtained by multiplying an angular velocity by a predetermined time, the angular velocity being obtained by subtracting the second angular velocity from the first angular velocity. The DRAG pulse is generated through an operation of adding the first sine curve and the second sine curve.

Inventors

SIM, JAE YOON
KANG, KI SEO

Assignees

POSTECH Research and Business Development Foundation

Dates

Publication Date: 20260513
Application Date: 20231201

Claims (20)

A method for generating a qubit control signal for controlling a quantum computer, comprising: generating a first sine wave having a first phase obtained by multiplying an angular frequency, which is a sum of a first angular frequency corresponding to a center frequency and a second angular frequency corresponding to a frequency located to the left of the center frequency, by an arbitrary time of each cycle; and generating a second sine wave having a second phase obtained by multiplying an angular frequency, which is a difference between the first angular frequency and the second angular frequency, by the arbitrary time; wherein a DRAG (derivative removal by adiabatic gate) pulse for the qubit control signal is generated through an addition operation of the first and second sine waves.
The method of claim 1, further comprising: generating a third sine wave having a third phase obtained by multiplying an angular frequency, which is a sum of the first angular frequency and the second angular frequency, by the arbitrary time; and generating a fourth sine wave having a fourth phase obtained by multiplying an angular frequency, which is a difference between the first angular frequency and the second angular frequency, by the arbitrary time; wherein a main pulse of the qubit control signal is generated through a subtraction operation between the third and fourth sine waves.
The method of claim 2, further comprising: setting a current bias of the first and second sine waves such that a scale factor is multiplied to the DRAG pulse.
The method of claim 3, wherein the scale factor is any real value selected from a range of 0 or more and less than 1.
The method of claim 3, further comprising: wave-shaping the first to fourth sine waves together so as to synthesize the main pulse and the DRAG pulse to generate a qubit control pulse.
The method of claim 5, further comprising: converting the qubit control pulse into a radio frequency output signal by adding a main frequency of a phase-locked loop according to a plurality of phase signals applied to a mixer.
The method of claim 2, further comprising: specifying an initial phase for generating each of the first to fourth sine waves.
The method of claim 2, wherein each of the generating of the first to fourth sine waves comprises: repetitively accumulating a frequency codeword (FCW) within a range from zero to a preset upper limit from a predetermined initial phase; inverting even-numbered rising cycles of accumulated values obtained in the accumulating of the FCW to convert the accumulated values into a sine waveform; converting preset upper bits of sine waveform-converted accumulated values into a first analog signal; converting preset lower bits of the sine waveform-converted accumulated values into a second analog signal; and interpolating the first analog signal by adding the second analog signal to the first analog signal.
The method of claim 2, further comprising: synchronizing processing timing of data or signals input to four controllers, each of supplies with a phase signal and a frequency codeword (FCW) for determining an initial phase of each of the first to fourth phases, by four clock calculators coupled to the controllers, respectively.
The method of claim 9, further comprising: delivering, to each of the four clock calculators, some of the sub-clocks obtained by dividing a reference clock supplied from outside the qubit control signal generator into n sub-clocks, where n is 8 or 16.
The method of claim 10, further comprising: determining a first main frequency according to the reference clock to generate a first phase signal and a first clock signal based on the first main frequency; determining a second main frequency, different from the first main frequency, according to the reference clock to generate a second phase signal and a second clock signal based on the second main frequency; and selecting one of the first phase signal and the second phase signal to supply the selected one to first to fourth frequency generators to respectively generate the first to fourth sine waves.
The method of claim 9, further comprising: distributing data and control signals input from outside to the four controllers by a main controller connected to the four controllers.
A qubit control signal generation apparatus for quantum computer control, comprising: a first frequency generator configured to generate a first sine wave having a first phase obtained by multiplying an angular frequency, which is a sum of a first angular frequency corresponding to a center frequency and a second angular frequency corresponding to a frequency located to the left of the center frequency, by an arbitrary time of each cycle; and a second frequency generator configured to generate a second sine wave having a second phase obtained by multiplying an angular frequency, which is a difference between the first angular frequency and the second angular frequency, by the arbitrary time; wherein a DRAG (derivative removal by adiabatic gate) pulse for a qubit control signal is generated through an addition process of the first and second sine waves.
The apparatus of claim 13, further comprising: a third frequency generator configured to generate a third sine wave having a third phase obtained by multiplying an angular frequency, which is a sum of the first angular frequency and the second angular frequency, by the arbitrary time; and a fourth frequency generator configured to generate a fourth sine wave having a fourth phase obtained by multiplying an angular frequency, which is a difference between the first angular frequency and the second angular frequency, by the arbitrary time; wherein a main pulse for the qubit control signal is generated through a subtraction process of the fourth sine wave from the third sine wave.
The apparatus of claim 14, wherein, to generate a qubit control pulse in which a main pulse corresponding to a first term on the right-hand side of following equation and a DRAG pulse corresponding to a second term on the right-hand side are synthesized from the first to fourth sine waves, the device performs a subtraction operation of the fourth sine wave from the third sine wave, and an addition operation of the first and second sine waves for synthesis, O t = O on t 1 2 sin ω 01 + ω w t − sin ω 01 − ω w t + q scale sin ω 01 + ω w t + sin ω 01 − ω w t wherein, O ( t ) denotes the qubit control pulse, O on ( t ) denotes an on-time pulse of a frequency generation circuit, q scale denotes a scale factor, ω 01 denotes a phase angle at an arbitrary time t of a qubit ground state, and ω w denotes a phase angle of the qubit at arbitrary time t.
The apparatus of claim 15, further comprising: a mixer configured to convert the qubit control pulse into a radio frequency output signal by adding the main frequency of a phase-locked loop to the qubit control pulse in accordance with a plurality of phase signals of the phase-locked loop.
The apparatus of claim 15, wherein the third and fourth frequency generators respectively comprise third and fourth current biases for supplying reference-level power for operation of the third and fourth frequency generators; and the first and second frequency generators respectively comprise first and second current biases for supplying a power level obtained by multiplying the reference level by a scale factor for the operation of the first and second frequency generators.
The apparatus of claim 14, wherein each of the first to fourth frequency generators comprises: a controller configured to output a frequency codeword (FCW), a phase signal specifying an initial phase, and an enable signal; a repetitive phase accumulator configured to repetitively accumulate the frequency codeword from an initial phase determined by the phase signal within a range from zero to a preset upper limit; a waveform converter configured to convert the repetitively accumulated values into a triangular or sine waveform by inverting even-numbered rising cycles; a nonlinear differential digital-to-analog converter (DAC) configured to convert upper bits of the sine waveform-converted values into a first analog signal; an interpolation DAC configured to convert lower bits of the sine waveform-converted values into a second analog signal; and an adder configured to interpolate the first analog signal by adding the second analog signal to the first analog signal.
The apparatus of claim 18, wherein each of the first to fourth frequency generators further comprises a clock calculator configured to align the processing timing of data or control signals input to the controller.
The apparatus of claim 18, further comprising: a first phase-locked loop (PLL) configured to determine a first main frequency based on a reference clock and generate a first phase signal and a first clock signal based on the first main frequency; a second phase-locked loop configured to determine a second main frequency different from the first main frequency based on the reference clock and generate a second phase signal and a second clock signal based on the second main frequency; a clock selector configured to select one of the first and second phase signals and provide the selected phase signal to the first to fourth frequency generators; and a main controller connected to four controllers of the first to fourth frequency generators, configured to distribute data and control signals input from outside a dilution refrigerator to the four controllers.

Description

[Technical Field] The present disclosure relates to a qubit control signal generation technology for a superconducting quantum computer, and more particularly, to a qubit control signal generation method and apparatus for generating a main pulse and a DRAG (derivative removal by adiabatic gate) pulse used in superconducting quantum computer control with low power consumption and without memory. [Background Art] A superconducting quantum computer is currently controlled using radio frequency (RF) signals and phase. A transmon qubit, which is the basic data processing unit of a superconducting quantum computer, has a shorter coherence time-i.e., the duration during which information is maintained-compared to other types of qubits. Google and IBM, which are currently developing superconducting quantum computers, are attempting to reduce the length of control signals in order to perform more operations within the limited coherence time. A qubit, utilizing quantum characteristics, has a unique frequency corresponding to each energy state. The frequency used for basic quantum computing operations corresponds to the energy level of |0> ↔|1> state ω01, and the frequency used to create quantum entanglement corresponds to the energy level of |1> ↔|2> state ω12. Generally, the frequency f01 corresponding to the ground state energy ω01 is between 5-7 GHz, and the frequency f12 corresponding to the first excited state energy ω12 refers to a frequency approximately 150-350 MHz lower than the above frequency f01. When the length of the RF signal for qubit control is shortened, the frequency spectrum spreads more widely. This causes noise in the frequency region surrounding the center frequency of the qubit. That is, in consideration of the limited coherence time, for example, when a short pulse width of 30 ns or less is used, the frequency power spreads around the center frequency. This generates unwanted changes in energy state-i.e., noise-at a second frequency f12 located approximately 150-350 MHz to the left of the center frequency f01 of the qubit. This becomes a major source of error in quantum operations. To address such sources of error and to minimize frequency spreading, techniques have been proposed that use pulse shapes such as a Gaussian shape, sine shape, or raised cosine shape instead of a rectangular pulse shape. However, although the aforementioned conventional techniques may reduce frequency spreading compared to the rectangular shape, as the length of the control pulse becomes shorter, it becomes impossible to achieve sufficiently low noise at the second frequency using only the pulse shape. Another conventional technique proposes a method to reduce noise at the second frequency by generating a DRAG (derivative removal by adiabatic gate) pulse with a 90° phase difference from the frequency of the main pulse, using the derivative shape of the original signal-i.e., the main pulse. This method is currently used in most superconducting quantum computer operations. Although the method of generating a qubit control pulse using a DRAG pulse plays a significant role in reducing noise at the second frequency, the process of generating the DRAG pulse, which is a high-frequency signal with a 90° phase difference and a derivative pulse shape, significantly increases the complexity of the circuit. In addition, the DRAG pulse is originally used in conjunction with the main pulse and functions to reduce the frequency power on the left side of the center frequency while increasing it on the right side. One of the other conventional techniques that utilizes this characteristic while addressing the drawbacks of the DRAG pulse involves a single sideband (SSB) synthesis method to control frequency and phase, and in the process, generates a pulse shape that reduces noise at the second frequency located to the left of the center frequency using memory. However, the aforementioned conventional method requires a high digital system frequency, for example, 1 GHz or more, due to the characteristics of the qubit. Since high resolution is also required, a large number of bits is needed. In particular, to generate a Gaussian pulse shape and differentiated DRAG pulse shape thereof, memory is essential, and since each qubit has a unique frequency property, memory cannot be shared. Thus, a large amount of memory capable of storing values for each individual qubit is required. Specifically, the problem with the above-mentioned conventional technique is that the DRAG pulse differs from the main control signal in both phase and shape. Therefore, in order to control frequency and phase using the single sideband (SSB) method that utilizes a local oscillator (LO) frequency and an intermediate frequency (IF), the shape of the DRAG pulse is generated using data obtained by multiplying the amplitude with the main control signal stored in memory when creating the intermediate frequency. In other words, in the case of using the above method, sinc