# US 2021/0081779 A1 · Quantum state classifier using reservoir computing

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0081779 A1
- Document ID: US20210081779A1
- Status: Published application
- Kind: A1
- Application number: 16/570,592
- Filed: 2019-09-13
- Published: 2021-03-18
- Earliest claimed: 2019-09-13
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Naoki Kanazawa
- Web page: https://patentgrep.com/patent/US20210081779A1

## CPC

- G06F 18/2415: based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/065: Analogue means
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 10/40: Physical realisations or architectures of quantum processors or components for manipulating qubits, e.g. qubit coupling or qubit control
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A quantum state classifier includes a reservoir computing circuit for post-processing a quantum bit to obtain a readout signal, and a readout circuit, coupled to the reservoir computing circuit, for discriminating a quantum state of the quantum bit from the readout signal from among multiple possible quantum states. The readout circuit is trained in a calibration process respectively activated by a specific one of each of the multiple quantum states such that weights within the linear readout circuit are updated by minibatch learning for each of multiple measurement sequences of the calibration process. The readout circuit generates a binary output after the multiple measurement sequences in a post-calibration classification process for a test quantum bit. The quantum state classifier further includes a controller, coupled to the readout circuit, selectively triggerable to output a control pulse responsive to the quantum state of the test quantum bit indicated by the binary output.

## Claims

### Claim 1

A quantum state classifier, comprising:
a reservoir computing circuit for post-processing a quantum bit to obtain a readout signal;
a linear readout circuit, operatively coupled to the reservoir computing circuit, for discriminating a quantum state of the quantum bit from the readout signal from among multiple possible quantum states, the linear readout circuit being trained in a calibration process respectively activated by a specific one of each of the multiple quantum states such that weights within the linear readout circuit are updated by minibatch learning for each of multiple measurement sequences of the calibration process, wherein the linear readout circuit generates a binary output after the multiple measurement sequences in a post-calibration classification process for a test quantum bit; and
a controller, operatively coupled to the linear readout circuit, selectively triggerable to output a control pulse responsive to the quantum state of the test quantum bit indicated by the binary output.

### Claim 2 (depends on 1)

The quantum state classifier of claim 1, wherein the reservoir computing circuit and the linear readout circuit comprise at least one analog hardware circuit.

### Claim 3 (depends on 2)

The quantum state classifier of claim 2, wherein the at least one hardware circuit comprises at least one microwave circuit.

### Claim 4 (depends on 1)

The quantum state classifier of claim 1, wherein the reservoir computing circuit comprises a non-linear amplifier using delayed feedback to amplify the readout signal.

### Claim 5 (depends on 1)

The quantum state classifier of claim 1, wherein each of the weights is determined as a function of a gain of at least one readout amplifier comprised in the linear readout circuit.

### Claim 6 (depends on 1)

The quantum state classifier of claim 1, wherein the linear readout circuit comprises at least some different circuit paths for the calibration process versus the post-calibration classification process.

### Claim 7 (depends on 1)

The quantum state classifier of claim 1, wherein a corresponding one of the different circuit paths for the calibration process outputs a cost function of an adaptive filter.

### Claim 8 (depends on 1)

The quantum state classifier of claim 1, wherein the reservoir computing circuit comprises a comparator for binarizing a pre-processed signal representative of the quantum bit prior to the post-processing.

### Claim 9 (depends on 8)

The quantum state classifier of claim 8, further comprising a microwave switch triggerable by an output of the comparator.

### Claim 10 (depends on 1)

The quantum state classifier of claim 1, wherein the linear readout circuit is comprised in a cryostat.

### Claim 11 (depends on 10)

The quantum state classifier of claim 10, wherein the reservoir computing circuit is comprised in the cryostat.

### Claim 12 (depends on 1)

The quantum state classifier of claim 1, wherein the linear readout circuit is tuned by adjustment of the weights.

### Claim 13 (depends on 1)

The quantum state classifier of claim 1, wherein the linear readout circuit comprises an analog integrator circuit for outputting a mean activation of each of a plurality of neurons.

### Claim 14 (depends on 1)

The quantum state classifier of claim 1, wherein the reservoir computing circuit comprises a Leaky Integrator Echo State Network (LI-ESN).

### Claim 15 (depends on 1)

The quantum state classifier of claim 1, wherein each of the multiple measurement sequences correspond to a respective one of the multiple possible quantum states.

### Claim 16

A method for quantum state classification, comprising:
post-processing, by a reservoir computing circuit, a quantum bit to obtain a readout signal;
discriminating, by a linear readout circuit operatively coupled to the reservoir computing circuit, a quantum state of the quantum bit from the readout signal from among multiple possible quantum states while training the linear readout circuit in a calibration process respectively activated by a specific one of each of the multiple quantum states, wherein the training process updates weights within the linear readout circuit by minibatch learning for each of multiple measurement sequences of the calibration process;
generating, by the linear readout circuit, a binary output after the multiple measurement sequences in a post-calibration classification process for a test quantum bit; and
selectively triggering a controller operatively coupled to the linear readout circuit to output a control pulse responsive to the quantum state of the test quantum bit indicated by the binary output.

### Claim 17 (depends on 16)

The method for quantum state classification of claim 16, wherein the calibration process sequentially updates the weights for each of the multiple possible quantum states.

### Claim 18 (depends on 16)

The method for quantum state classification of claim 16, wherein the calibration process comprises switching between calibration modes for each of the multiple possible quantum states.

### Claim 19 (depends on 16)

The method for quantum state classification of claim 16, further comprising determining each of the weights as a function of a gain of at least one readout amplifier comprised in the linear readout circuit.

### Claim 20 (depends on 16)

The method for quantum state classification of claim 16, further comprising tuning the linear readout circuit by adjustment of the weights.

### Claim 21 (depends on 16)

The method for quantum state classification of claim 16, wherein said post-processing step comprises binarizing, by a comparator, a pre-processed signal representative of the quantum bit prior to the post-processing.

### Claim 22

A quantum state classifier, comprising:
an analog reservoir computing circuit for post-processing a quantum bit using an Echo State Network (ESN), at least one delayed feedback device and a non-linear amplifier to obtain a readout signal;
a linear readout circuit, operatively coupled to the reservoir computing circuit, for discriminating a quantum state of the quantum bit from the readout signal from among multiple possible quantum states, the linear readout circuit being trained in a calibration process respectively activated by a specific one of each of the multiple quantum states such that weights within the linear readout circuit are updated by minibatch learning for each of multiple measurement sequences of the calibration process, wherein the linear readout circuit generates a binary output after the multiple measurement sequences in a post-calibration classification process for a test quantum bit; and
a controller, operatively coupled to the linear readout circuit, selectively triggerable to output a control pulse responsive to the quantum state of the test quantum bit indicated by the binary output.

### Claim 23 (depends on 22)

The quantum state classifier of claim 22, wherein the linear readout circuit and the reservoir computing circuit are comprised in a cryostat.

### Claim 24 (depends on 22)

The quantum state classifier of claim 22, wherein the linear readout circuit comprises an analog integrator circuit for outputting a mean activation of each of a plurality of neurons.

### Claim 25 (depends on 22)

The quantum state classifier of claim 22, wherein the Echo State Network is a Leaky Integrator Echo State Network (LI-ESN).
