WO2022123269A1

WO2022123269A1 - Quantum computing system and methods therefor

Info

Publication number: WO2022123269A1
Application number: PCT/GB2021/053247
Authority: WO
Inventors: Iain Mcintosh Hunter; Zak David ROMASZKO
Original assignee: Universal Quantum Ltd
Priority date: 2020-12-11
Filing date: 2021-12-10
Publication date: 2022-06-16
Also published as: WO2022123269A8; GB2602004A; GB202019589D0

Abstract

A system and methods for quantum computing are described, the system comprising a processor (110) in communication with at least one or more electrode driver circuit (230) connected to a respective electrode for applying an analogue signal, the each or at least one electrode driver circuit (23) comprising a Digital Analogue (DAC) (260) connected with a local memory (240) and a local processor (250) for controlling the DAC (260), and wherein the processor (110) is adapted to provide one or more predefined procedure calls to the local processor (250) of the each or at least one electrode driver circuit (230).

Description

Quantum Computing System and Methods therefor Introduction The present application relates to quantum computing systems, including for example, but not limited by, such systems utilizing ions and ion traps, and relates to apparatus and methods for operating such systems. Background Quantum computing in general, unlike so-called “classical computing”, relies on the quantum mechanical properties of particles or matter to produce or alter data. The data may be represented by quantum bits or “qubits”, which may involve the superposition of quantum states, or for example entanglement in which the state of one particle or atom is influenced by another particle or atom. Quantum mechanical qubits are able to encode information as combinations of zeros and ones simultaneously. Such properties open numerous complex numerical applications that are traditionally difficult for classical computers. Examples include artificial Intelligence, image processing and recognition, cryptography, or secure communications and so on. Surface ion traps may be used to create quantum computers. This involves the trapping of super cooled ions on or above surface patterned electrodes forming an ion trap, exciting other ions so that they interact or superpose with another ion, and manipulating or moving the ions on the electrodes by altering the surface electrode fields and “reading out” the result from the qubit. The surface electrode potentials of an ion-trap are preferably controlled by Digital to Analogue Converters (DACs) as described for example in: “Lekitsch, Bjoern & Weidt, Sebastian & Fowler, Austin & Mølmer, Klaus & Devitt, Simon & Wunderlich, Christof & Hensinger, W.. (2017). Blueprint for a microwave trapped ion quantum computer. Science Advances. 3. e1601540. 10.1126/sciadv.1601540.” In the above paper, numerous multi-channel-DACs (AD5370) are driven using Field Programmable Gate Arrays (FPGAs) and memory, and all DAC data originates from FPGA processing or from stored data in the memory. All data for all DACs must be sent via a bus from the central controlling computer, the bus being for example a Serial Processing Interface bus, in order to control the ion in real time. This has a disadvantage in that updating all DACs synchronously is slow with respect to the update rate of a single DAC. A further problem lies in the scalability of such an architecture (and hence “computing power”) owing to the disadvantage. Superconducting qubits can be used to create quantum computers. This involves providing a plurality of independent IQ control waveforms each at a unique frequency. The generation of the unique IQ controls involves mixing a carrier signal with two separately generated and arbitrarily specifiable waveforms to produce a single “complex” pulse waveform. Whilst, in a theoretical ideal, this single “complex” pulse waveform can be provided to all qubits, in practice the variability in fabrication and the presence of local defects result in a requirement to customise the pulse waveform for each qubit. This overhead scales with the number of qubits present and therefore this architecture encounters the same issues in relation to scalability as the ion trap architecture previously mentioned. Further relevant information is set out in the review paper “A Quantum Engineer’s Guide to Superconducting Qubits” from P. Krantz; M. Kjaergaard; F. Yan; T.P. Orlando, S. Gustavsson and W. D. Oliver dated 9 July 2021 and available as https://arxiv.org/pdf/1904.06560.pdf. There is therefore a desire to provide improved quantum computing systems and methods for operation of a variety of different architectures including, but not limited to, ion trap and superconducting architectures. Summary According to a first aspect there is provided a system for quantum computing comprising a processor in communication with at least one or more electrode driver circuit connected to a respective electrode for applying an analogue signal, the each or at least one electrode driver circuit comprising a Digital Analogue Converter (DAC) connected with a local memory and a local processor for controlling the DAC, and wherein the processor is adapted to provide one or more predefined procedure calls to the local processor of the each or at least one electrode driver circuit. According to another aspect there is provided a method for quantum computing, comprising providing a computer with a plurality of predefined procedural calls defining operations on a quantum computer, providing said procedure calls to at least one or more electrode driver circuits comprising a Digital Analogue Converter (DAC) connected with a local memory and a local processor for controlling the DAC. Advantageously, the at least one electrode driver circuit is configured to receive procedural calls or commands from the central processor. In an embodiment, the local processor is adapted to select and output an operation waveform in dependence on the received procedure call and data stored in the local memory. In another embodiment, the electrode driver circuit processor generates parameterized waveforms from functions stored in the local memory. In a further embodiment, the electrode driver circuit processor generates parameterized waveforms from functions stored in the local memory wherein the functions comprise one or more of a ramp, ¼ sine wave, or descending or any mixture of these. In yet another embodiment, a system is provided further comprising a plurality of surface electrodes, each connected with a respective electrode driver circuit to provide procedural calls defining waveforms for local storage and execution on said surface electrodes. In an embodiment, a system is provided further comprising a plurality of surface electrodes, each connected with a respective electrode driver circuit to provide procedural calls defining waveforms for local storage and execution on said surface electrodes, and wherein the electric potential of the or each surface electrodes is controlled by the respective DAC output of the electrode driver circuit in accordance with the selected or generated waveform. In an embodiment according to the second aspect, there is provided a method for quantum computing, comprising providing a computer with a plurality of predefined procedural calls defining operations on a quantum computer, providing said procedure calls to at least one or more electrode driver circuits comprising a Digital Analogue Converter (DAC) connected with a local memory and a local processor for controlling the DAC, and further comprising the electrode driver circuit retrieving from the local memory, under control of the local processor, an operation associated with the provided procedure call, and controlling the DAC in dependence on the operation. Furthermore, according to the present invention, there is provided a method comprising the electrode driver circuit generating from the local memory, under control of the local processor, an operation associated with the provided procedure call, and controlling the DAC in dependence on the operation. In another aspect, there is provided a carrier for quantum computing carrying a plurality of electrode driver circuits in connection with respective row and column drivers for use with a control processor. In yet another aspect, there is provided computer code for controlling an ion in a quantum computer comprising instructions, which when executed by a processor, cause said processor to perform the steps of the second aspect. In the abovementioned aspects the electrode driver circuit architecture provides a form of local processing and memory for each electrode driver circuit and therefore the quantum computer controller may only need to send very small amounts of predefined control or procedural data to the distributed local electrode driver circuit processors to execute very large sequences of DAC value up-dates, thereby providing a scale-able architecture. Advantageously, in one example, the local memory may store data defining multiple waveforms defining ion operations. In another example, the local processor may retrieve data from the memory associated with that local processor, the data defining a waveform and the processor may provide the data to the input of the DAC to control the surface electrode potential of an ion trap to achieve the desired operation. In yet a further example, the local memory may store code instructions comprising the assembly or generation of multiple waveforms, advantageously enabling “on electrode driver circuit” generation of a complex procedure thereby decoupling the generation and provision of input data to the DAC from the control computer and enhancing scalability. In another example the control computer may provide global timing and synchronisation signals in addition to the predefined procedure calls. In yet another example, the control computer may provide one or more of the predefined procedure calls relating to the waveform to be applied to the DAC, in a stack or consecutive loop. This advantageously decouples the provision of the waveforms from the central computer and enables faster loading and computation. The present invention will now be described, by way of example only, with reference to the accompanying drawings, in which: Figure 1 shows a quantum computing system, Figure 2 illustrates a quantum computing system according to an aspect of the present disclosure, Figure 3 shows an electrode driver circuit in accordance with the present disclosure, Figure 4 is an example of a table of procedure calls, and their associated operations, Figure 5 is a diagram illustrating example waveforms and electrical potential timing to move or perform a “shuttle” operation on an ion from left to right across the electrode, Figure 6 is a block diagram illustrating steps in a method according to an aspect of the present disclosure. Detailed Description Figure 1 shows an example quantum computer (100) in which surface ions or other matter may be used to provide qubits for the quantum computer (100). Ion movement may be subjected to applied surface electrode potentials, instructed by a central processor (110) and memory (120) provided via a bus (130) to each Digital to Analogue Converter DAC (140). For example, traditional techniques of driving multiple DC electrodes of a surface ion based quantum computer (100) may be typically achieved using bussed (serial or parallel) communication (130) between a centralized controller/processor (110) and Digital to Analogue Converters (DACs) 140, thereby controlling the DC electrode voltage and hence surface potential to influence the ion in question. In this architecture the central processor (110) may have to generate large amounts of data in real time. The central processor (110) may also typically access data stored in a central memory (120) in communication with the central processor (110). An example of such memory storage associated with a central processor (110) is, as will be appreciated DDR (120), and all DACs (140) must share that memory. This creates memory (120) access bottlenecks and may further invoke bus line (130) delays. Furthermore, as will be appreciated by those skilled in the art, the quantum computer (100) may require substantive parallel wiring (130) between the central processor (110) and the multiple DACs (140) leading to noise. Figure 2 depicts an embodiment of a quantum computing architecture (200) according to the present disclosure. There is provided a substrate (215), which may be silicon, in which the processing in any computer (110) and memory (120) is not unconstrained and arbitrary. The processor 110 is configured in order to enable simultaneous control of the plurality of electrode driver circuits 230 or cells. In the context of this invention, the term simultaneous is used to define a procedure call that is put out to all electrode driver circuits 230 at substantially the same time. This effectively provides a clocked process wherein the commencement of unit operations across an array of electrode driver circuits 230 is synchronized. The synchronization across all of the electrode driver circuits ensures that all unit operations from all electrode driver circuits are complete before a new procedure call is issued from the processor 110. In the case of an empirical system, the electrode driver circuits 230 each have a flag which is set to either “ready” or “busy” and the processor 110 reads the flags and provides commands to any subset of the driver units that show their flag as “ready.” Once a full set of “ready” flags is displayed across the array, a new procedure call can be issued by the processor 110. In the case of a calculated system, the processor 110 stores information in the memory 120 relating to the time taken for the commands initiated by each of the waveforms stored in each of the local memories 240 in each electrode driver circuit 230. The processor 110 can use this information to create a workflow in which all electrode driver circuits 230 commence a task at T=0. However, if the first task executed by each of the electrode driver circuits 230 is not the same, then the time taken for each of the driver units to execute the task will differ. The processor 110, through the information stored in the memory 120, is aware of when the last of the electrode driver circuits 230 should have finished its task and therefore the time at which the system is ready for a new procedure call to be issued by the processor 110. In this embodiment, the substrate or wafer (215) comprises a column (210) and a row (220) arrangement to address each cell (230) of the matrix (215) such obtained. In another embodiment the substrate or carrier may be in the form of an Application Specific Integrated circuit (ASIC) or a custom IC or chip. In the example where the qubits are provided in the form of ion traps, the substrate or wafer 215 does not necessarily need to operate at cryogenic temperatures, although it will need to be compatible with an ultra high vacuum (UHV) environment. Although this is challenging for packaged components, a bare-die form silicon is compatible with an UHV environment and therefore can provide the wafer 215 in the example illustrated in Figure 2. As illustrated in Figure 3, each cell (230) or electrode driver circuit (230) comprises its own local processor (250) or microcontroller (250) connected to a local memory store (240) in communication with a DAC (260) for that electrode. By providing an electrode driver circuit (230) the central processor (110) and memory (120) are decoupled from the current electrode voltage waveforms applied to the DAC (260) for each electrode function. Hence noise is reduced. With reference to Figure 4, each particular procedure or waveform representing such may be simply instructed by the central processor (110) and the local processor (250) may look up from local memory (240) and provide the required procedure or waveform for the required ion movement. This may be repeated for each electrode having an associated electrode driver circuit (230) or cell. This has the advantage of decoupling the control processor (110) and memory (120) from the DAC (260) responsible for the electrode waveform or potential applied to the electrode to move, trap or read out an ion. Figure 4 illustrates example procedures (410a, b) and waveforms (420) or partial waveforms (snippets) that may be realized or useful. For example, and with reference to Figure 4, each waveform (420) or snippet or portion of a waveform, or a sequence of snippets or portions may be considered a procedure. For instance, example procedures may comprise at least (See Figure 4): o Move single ion Right/Left o Merge 2 ions -Left and Right –to form crystal, o Split 2 ion crystal to form LEFT-RIGHT ion pair, o Move LEFT ion pair to the left, o … The waveforms (420) and associated procedures (410a,b) are stored, with reference to each cell or electrode driver circuit (230) in the local memory (240) of that cell or electrode driver circuit (230) and available for lookup via the local cell processor (250). Hence a hardware architecture is provided that, provides a central control (110, 120) that is capable of addressing all of the electrode driver circuits simultaneously. Because each of the electrode driver circuits is provided with its own memory, processor and DAC, each driver unit is capable of acting independently and simultaneously on receipt of a command from the controller, thereby overcoming the problem of multiplexing memory access (120, 130) and reducing the data transfer overhead on the central processor (110). To further illustrate this, and with reference to Figure 5, distributed processing at the DAC level of hierarchy is achieved by decoupling the central processor (110) from the DAC (260). Figure 5 illustrates, on the left, the waveforms applied over time to move or shuffle the trapped ion to the right on an electrode, so as to be measured or otherwise at a Gate Trap or Measurement Zone (right of the figure). The local memory (240) may contain a finite number of waveform snippets that can be conjoined, without discontinuity, to provide synced or looped waveform operation. The output of the memory (240) is preferably directly connected to the DAC (260) input. Direct connection between the memory and the DAC is beneficial when the waveforms are stored as intensity v time plots and no interpretation is required of the information stored in the memory. Conversely, the processor 250 may be provided between the memory 240 and the DAC 260. This configuration is preferable in circumstances wherein the waveforms are stored as a list of coefficients of a polynomial and the processor 250 is required to interpolate the polynomial in order to identify the values of the points that should be provided to the DAC 260. This configuration enables more efficient use of memory and therefore enables more waveforms to be stored within a memory of a given size. This therefore increases the range of commands that can be carried out by the driver unit. Those skilled in the art will appreciate that the number and configuration of DC electrodes and their respective spacing may vary and the waveforms stored or supplied to each DAC (260) via local memory (240) will vary in dependence on the geometry and design of the electrodes. Each co-processor (250) may be programmed to synthesize or generate parameterized waveforms from locally stored mathematical functions – such as increasing or decreasing ramp, ¼ sine wave and so on. The communication between central processor (110) and DAC co-processors (250) consists of a protocol that calls for the local processor (250) to perform a command from a finite set of available commands. At the lowest level, commands are the identification of which waveform or waveform snippets are to be applied to the DC electrodes. Advantageously, only the order and time at which the snippet or waveform should be applied need be transmitted by the central processor (110). A stack (410a, 410b) of such commands may be stored in the same local memory (240) as waveform (420) snippets or formulae for generation in yet another embodiment. The waveform (420) may be preloaded to the DAC (260) local memories (240) during periods when quantum processing is not occurring, such as in an initialization phase. This pre-loading may occur on the same bus as the commands. The bus would use a protocol that allows data or commands to be sent using the same physical bus. In one embodiment, the memory (240) is dual port, allowing simultaneous read and write. This local memory (240) of the electrode driver circuit (230) may then be updated at the same time as a procedure (410a, 410b) is being executed (250). In this fashion the DAC (260) and the electrode it outputs too is not limited in the sense that a procedure or waveform may be updated whilst another is executing. In an embodiment the array of micro-processors (250) are addressed so that the central computer (110) updates the local memories (250). This is achieved, with reference to Figure 2, by using row (220) and column (210) addressing in the same fashion as TFT displays. Data or procedure commands from the central computer (110) are input to the column drive (210), which may be for example a shift register with Nc columns. The Row drive (220) may then be a N to 2N decoder or a single bit shift resister. The row output (220) defines which row is active to receive data. Data is then parallel loaded into the local data processor (250). The local processor (250) updates the local memory (240) with respect to the data it receives. The central processor (110) may this thus provide global signals (130) indicating timing for synchronization purposes. These may include, for example, a global reset, a global start program clock, a global DAC update clock, a snippet update frame clock. By way of example, in the above embodiments a large array of DACs (260) control the position of multiple ions on a smaller array of surface ion traps – for example, cross or X-junctions. As the processing power of the quantum computer increases the number of DACs (260) needed to be controlled increases leading to the requirement of scale-able system architectures for increased computing power. The inventors have realised that using a central processor and memory to provide data for these DACs is not scale-able. Hence, in the above having central control (110) of distributed architectures (210, 220, 230) including a local (to each electrode) driver circuit (230) comprising local processing (250) and memory (240) enables large arrays (215) to be built and utilised and therefore provides a scaling solution for quantum computing. The distributed processors (250) may be used to do low level data retrieval of waveforms stored in local memory (240) – the waveforms are used to move ions on the trap surface. The memory can be pre-programmed ROM, may usefully be RAM. The local processor may also generate parameterised waveform-snippet sequences. It is noted that the number of waveform or waveform-snippets required to be stored in local memory (240) is finite as there is only a finite number of ion movement states that are controlled by each DAC (260). In the foregoing embodiments, a snippet or waveform comprises a number of DAC values and the times for which they are applied. The snippet defines the shape of the waveform, for example, sigmoid, ¼ sine wave, exponential or linear. The snippet can additionally be parameterized, for example stretch in time, stretch in amplitude or offset in value. It can also be an arbitrary waveform with no underlying mathematical function. When applied (250) to the DAC (260) the snippet would perform a task on the ion, see for example Figure 4. By way of example only, and with Reference to Figures 5 and 6 there now follows embodiments illustrating the efficient movement of ions using the aforementioned embodiments. In general, the electric field generated by DC electrodes controls the position and movement of the one or more trapped ions. Each ion may then be shuttled between trapping zones, measurement zones and quantum gate zones (See Figure 5 right hand side for a X-junction arrangement with such zones). The trapping zone is usually where the ion is first captured and held in position from the background noise. The ion is then moved to a quantum gate zone where a quantum state can be imposed on the ion using for example microwave radiation. To perform a quantum gate operation, the ion is subsequently transported to a gate zone. Here it may interact with another ion. Subsequently, the ion would be moved to a measurement zone where the final quantum state of the ion may be read using a laser of suitable wavelength, for example. In Figure 5 the left side graphs show examples of the applied potential waveforms along the x-junction arm. The graphs are labelled so that a graph line 504 is the potential due to the voltage applied to one of the electrode 504a 504b and the waveform 501 is the potential at one of the electrodes 501a or 501b. The y-axis represents electric field strength and the x-axis represents ion position along the arm of the x-junction. For example in the first graph a potential, indicated by line 504 has been applied to electrode 504a. The three graphs on the left, depict example waveforms applied sequentially at Time.0, Time.1 and Time.2. As can be seen, during these time intervals the ion has been “shuttled” from a position at the left-hand electrode (502a) to a position of the right hand electrode (502b). The right-hand figure shows the layout of the x-junction and the ion position at the 2^nd time interval (Time.1) as the ion moves to the right. During this period the electrode 501a starts at a high potential (Time=0) and falls to a lower potential at Time=1. Finally the electrode 501b returns to a high potential. All the DAC values between time =0 and time =2 form a waveform and this is called a snippet. That snippet defines, for each electrode the waveform required to shuttle the ion from its left-hand side 501a to its right-hand side 501b. Other snippets shuttle the ion from right to left, or two ions left to right. In the context of some embodiments we use the term distributed processing to refer to the fact that control of the data flow from memory to the DAC is local to the DAC and that that local processor can be programmed using the local memory and synchronised with global timing. Hence a central control computer provides local waveform data to the electrode driver circuits, or at least the memory associated with each. Following this and each electrode driver circuit, via timing impulses, provides the analogue output required in respect of the waveform. Figure 6 is a technical block diagram illustrating steps of the method. A central processor 610 in communication with memory 620 retrieves and provides a procedure call or command 630 to a local electrode driver circuit comprising a local processor 640 and a DAC. At step 650 the local processor controls the DAC of its respective electrode driver circuit to output the required waveform to the surface electrode. It should be noted that in some embodiments the central processor (110) has loaded the local memories (240) with waveform and program data and subsequently has no role in the actual transfer of data from the local memory (240) to the local DAC (260) other than via synchronization or timing pulses. Any function applied to the waveform such as stretch, amplitude gain, interpolate and so on may be carried out by the local processor (250). This functionality may further, in another embodiment, be achieved by parametrizing the waveform or snippet with those parameters also held in the local memory (240). Hence there is provided a system in which local tasks are effectively distributed via each electrode driver circuit thereby leading to a local-distributed system that enables scaling in quantum computing. Various further aspects and embodiments of the present invention will be apparent to those skilled in the art in view of the present disclosure. “and/or” where used herein is to be taken as specific disclosure of each of the two specified features or components with or without the other. For example “A and/or B” is to be taken as specific disclosure of each of (i) A, (ii) B and (iii) A and B, just as if each is set out individually herein. Unless context dictates otherwise, the descriptions and definitions of the features set out above are not limited to any particular aspect or embodiment of the invention and apply equally to all aspects and embodiments which are described. It will further be appreciated by those skilled in the art that although the invention has been described by way of example with reference to several embodiments. It is not limited to the disclosed embodiments and that alternative embodiments could be constructed without departing from the scope of the invention as defined in the appended claims.

Claims

CLAIMS 1. A system for quantum computing comprising a processor (110) in communication with a plurality of electrode driver circuits, each driver circuit (230) connected to a respective electrode (501a, b – 504a, b) for applying an analogue signal comprising one or more waveforms, each driver circuit (230) comprising a Digital Analogue Converter (DAC) (260) connected to a local memory (240) for storing the waveforms and a local processor (250) for controlling the DAC (260), and wherein the processor (110) is adapted to provide one or more predefined procedure calls to the local processor (250) of each electrode driver circuit (230).

2. A system according to claim 1, wherein each electrode driver circuit is adapted to receive procedure calls from the central processor (110).

3. A system according to claims 1 or 2, wherein the local processor is adapted to control the selection and output an operation waveform in dependence on the received procedure call and data stored in the local memory.

4. A system according to claim 1, wherein each electrode driver circuit processor generates parameterized waveforms from functions stored in the local memory.

5. A system according to claim 4, wherein the functions comprise one or more of a ramp, ¼ sine wave, descending.

6. A system according to any preceding claim, wherein the electrodes are surface electrodes (510a, b).

7. A system according to claim 6, wherein the electric potential of each surface electrode is controlled by the respective DAC output of the electrode driver circuit in accordance with the selected or generated waveform.

8. A method for quantum computing on a quantum computer system comprising a central processor (110) and a plurality of electrode driver circuits (230) each comprising a Digital Analogue Converter (DAC), a local memory and a local processor, the method comprising the steps of: providing to the memory of each of the plurality of electrode driver circuits (230) one or more predefined waveforms each defining an operation on the quantum computer system, providing, from the central processor (110), at least one procedure calls to each electrode driver circuit such that the local processor accesses a corresponding waveform from the local memory and outputs this via the Digital Analogue Converter (DAC).

9. A method according to claim 8, comprising the electrode driver circuit retrieving from the local memory, under control of the local processor, an operation associated with the provided procedure call, and controlling the DAC in dependence on the operation.

10. A method according to claim 8 or 9, comprising the electrode driver circuit generating from the local memory, under control of the local processor, an operation associated with the provided procedure call, and controlling the DAC in dependence on the operation.

11. An electrode driver circuit according to claim 1, wherein the local processor (250) is adapted to generate a waveform (420) in dependence on the received procedure call (410a, b).

12. A carrier carrying a plurality of electrode driver circuits in connection with respective row and column drivers.

13. Computer code for controlling an ion in a quantum computer comprising instructions which, when executed by a processor, cause said processor to perform the steps of any of method claims 8, 9 or 10.