Leveraging Operation-Aware Error Rates for Effective Quantum Circuit Mapping on NISQ-Era Quantum Computers
Total Page：16
File Type：pdf, Size：1020Kb
UREQA: Leveraging Operation-Aware Error Rates for Effective Quantum Circuit Mapping on NISQ-Era Quantum Computers Tirthak Patel Baolin Li Rohan Basu Roy Devesh Tiwari Quantum Computing is Coming! What is a Qubit (Quantum Bit)? A classical bit has two states: A quantum bit or qubit can be in a superposition of the two basis states: Upon measurement, the qubit superposition collapses, and the qubit can be found in one of the two basis states. Manipulating Qubit States A qubit can be put in a desired superposition by applying quantum operations which can be represented as rotations on the Bloch sphere. Manipulating Qubit States A qubit can be put in a desired superposition by applying quantum operations which can be represented as rotations on the Bloch sphere. Initially, the qubit is in the ground state. Then, it first gets manipulated by an H gate in an equal superposition state, then by a Rz gate. Multi-qubit Gate Operations Basis states of a two-qubit system can be expressed as Multi-qubit Gate Operations Two qubits can be entangled using two-qubit gates. E.g., Bell State ≠ In 2-qubit gates (CH, CRx, CRy and CRz), one qubit is the control qubit and the other is the target qubit. The respective 1-qubit gate is applied to the target qubit depending on the superposition of the control qubit. All quantum algorithm circuits can be broken down into one- and two- qubit basis gates. Engineering a Quantum Computing Device Readout/Control Resonators Coupling Resonators Qubits NISQ Devices are Highly Erroneous! Errors in applying microwave pulses cause 1-qubit gate errors. Coupling resonators can be highly erroneous causing 2-qubit gate errors. The readout resonators are also highly error-prone and cause readout errors. T1 coherence time: energy decay to the ground state. T2 coherence time: phase damping due to env. factors. Execution Flow on a Quantum Computer quantum algorithm. For example, if the estimated er- Table 1: IBM QX quantum computers. ror rate of each qubit is significantly different than the Online Date ComputersQuantum (Num. Qubits) Circuit Maps Nov 06, 2018 Melbourne (14), Yorktown (5) actual error rate when the circuit map is executed, then Every quantum computers is composed of multiple qubits – each with Jul 03, 2019 Ourense (5), Vigo (5) quantum algorithm. For example, ifthese the differences estimated add uper- over the circuit map executionTable 1: IBM QX quantumpotentially computers. different number of qubits and topological structure and result in a significantly inaccurate outcome. There- Online Date Computers (Num. Qubits) ror rate of each qubit is significantlyfore, different previous works than have the focused on estimating the error actual error rate when the circuit maprates is accurately executed, and then using that to find the optimalNov circuit 06, 2018 MelbourneMelbourne (14), YorktownYorktown (5) Ourense & Vigo map [3, 15, 18, 25, 26, 28]. Jul 03, 2019 Figure 2:Ourense LayoutA of single (5), IBM quantum Vigo quantum (5) algorithm computers. can be The “mapped” circles in rep- different ways on the these differences add up over the circuit map execution resent qubits. Thesame arrows quantum show computer possible – each 2-qubit mapping gates: is referred the di- as “circuit map”. What is Missing from Existing Solutions? Current ap- and result in a significantly inaccurate outcome. There- rection points fromCircuit control map toA target qubit. proaches use a single number to characterize the error for a 3-qubit Circuit map B fore, previous works have focused on estimatingrate of a given the qubit error irrespective of the different quantum mized circuit mapalgorithm selection achieves up to 15% reduc- rates accurately and using that to findoperations the optimal being circuit performed on the qubit [25, 26,Melbourne 28, 31]. tion in error rateYorktown for a quantumOurense algorithm, & Vigo compared to For the first time, we show that error rate is not only the current approaches which rely on a single aggregated map [3, 15, 18, 25, 26, 28]. qubit-specific, but also operation-specificFigure 2: (as Layout explained of IBMnumber quantum for error computers. rate estimation The based circles on historical rep- data. in Sec. 2, a 1-qubit gate can performresent different qubits. types The of arrows show possible 2-qubit gates: the di- UREQA’s quantum error prediction model and What is Missing from Existing Solutions?quantum operationsCurrent on ap- a single qubit). We show that rection points from controlcircuit to target mapping qubit. framework is open-sourced at quantum error rates can vary significantly depending on proaches use a single number to characterize the error https://github.com/GoodwillComputingLab/UREQA. rate of a given qubit irrespective of thethe different specific quantum quantum operation thatmized is being circuit performed, map selection achieves up to 15% reduc- even if other conditions are kept constant (i.e., the phys- 2UREQA: The Solution operations being performed on the qubitical qubit [25, and 26, the 28, machine). 31]. Sometion qubits in error with low rate ag- for a quantum algorithm, compared to For the first time, we show that errorgregate rate average is not error only rate mightthe experience current high approaches error Background. which relyThis on study a is single performed aggregated on the IBM Quan- rate for specific quantum operations. Hence, these qubits tum Experience (QX) - a public cloud service. We use qubit-specific, but also operation-specificshould (asbe avoided explained for a circuit-mapnumber selection for if a error partic- ratethe estimation IBM QX machines based listed on historical in Table 1. They data. cover a in Sec. 2, a 1-qubit gate can performular different circuit consists types of of many such specific quantum op- diverse range of quantum architectures in terms of error- UREQA’s quantumrates, topology, error and prediction time of introduction model (Fig. and 2). quantum operations on a single qubit).erations. We The show reason that for this phenomena is the unstable nature of current NISQ technologycircuit where qubits mapping are erro- frameworkQuantum operations is on open-sourced these computers include at both quantum error rates can vary significantly depending on the gate and readout operations. Primary 1-qubit gates neous and do not have consistenthttps://github.com/GoodwillComputingLab/UREQA properties as different . the specific quantum operation that isqubits being interact performed, differently with external control and the include the Hadamard (H) gate which puts the two ba- sis states into equal superposition and the x-, y-, and z- even if other conditions are kept constantenvironmental (i.e., the features. phys-This is the first work to discover and leverage the above insight2U to chooseREQA better circuit: Therotation Solution gates (Rx, Ry, and Rz, respectively) which rotate ical qubit and the machine). Some qubitsmaps that with lower low the impact ag- of quantum errors, and push the qubit about the x-axis, y-axis and z-axis on the Bloch the state-of-the-art in improvingBackground. the efficiency of quan-This studySphere, is respectively. performed The on Bloch the Sphere IBM is Quan- a unit sphere gregate average error rate might experience high error with the 0 state represented as a vector pointing toward tum algorithm execution on NISQ computers. | i rate for specific quantum operations. Hence, these qubits tum Experience (QX)the - positive a public z-axis cloud and the service.1 state is represented We use on the 1 | i UREQA Solution. UREQA buildsthe a IBM data driven QX model machinesnegative listed z-axis. in The Table other 1. two They axis represent cover the a qubit should be avoided for a circuit-map selectionfor correctly if estimating a partic- the error rates of operations on diverse range of quantumphase. The architectures qubit state vector in terms can point of anywhere error- on the ular circuit consists of many such specificdifferent quantum qubits of a quantum op- computer, and then, lever- Bloch Sphere, but upon readout, it collapses to the posi- ages this information to find therates, most optimized topology, circuit and timetive ( 0 of) or introduction negative ( 1 ) z-axis. (Fig. As 2). an example, Fig. 3 erations. The reason for this phenomena is the unstable | i | i for a quantum algorithm. UREQA builds its error rate uses the Bloch Sphere to show the state changes after ap- nature of current NISQ technology whereprediction qubits model are by erro- performing aQuantum large number operations of ex- on these computers include both plying a H gate followed by a Rz gate with p rotation to a neous and do not have consistent propertiesperiments as on real different IBM NISQ computers.the gate Our and evaluation readoutsingle operations. qubit. When Primary the H gate is 1-qubit applied, the gates qubit state qubits interact differently with externalshows control that these and error the rate predictioninclude methods the are Hadamard more vector (H points) gate toward which the positive puts the x-axis two and the ba- qubit is accurate than current state-of-the-art approaches of sim- equally probable to be measured as 0 or 1 . This prob- ply using a general error rate numbersis states periodically into pub- equal superposition and the x-,| y-,i and| i z- environmental features. This is the first work to discover ability of measurement remains the same even after a Rz and leverage the above insight to chooselished by better the quantum circuit computingrotation platform provider. gates (Rx, Ryrotation, and isR applied,z, respectively) except the qubit which has a negative rotate phase. To demonstrate UREQA’s effectiveness, we evaluate the qubit about the x-axis,All 1-qubit y-axis gates and have z-axis 2-qubit on variants the ( BlochCH, CRx, CRy maps that lower the impact of quantumUREQA errors,for a and diverse push set of quantum benchmarks, con- and CRz) where one qubit is the control and the other is the state-of-the-art in improving the efficiencyducting experiments of quan- over more thanSphere, 50 days respectively.