High-throughput calculation method and system adaptive to multiple scientific calculation software

Abstract

The invention discloses a high-throughput calculation method and system adaptive to multiple scientific calculation software, and relates to the technical field of computational chemistry, electrochemistry and material science. According to the method, parameter recognition and summarization are carried out on multiple scientific calculation software input files, an execution template and a calculation work station matched with target software are generated, parameters are classified in a differentiated mode and combined to generate a complete parameter configuration set so as to automatically render the target input files, task scheduling and result structured conversion and storage are achieved by means of a unified scheduling platform, and the target input files are automatically rendered. The method effectively solves the problems of high learning operation cost and difficulty in reuse caused by the fact that input files are manually compiled and modified in an existing calculation process and lack of unified standards, avoids repeated labor during parameter or model change, unifies output data formats to simplify analysis management, and improves the efficiency. According to the method, operations such as input file generation and task submission in high-throughput calculation do not need to be manually processed one by one, the error risk caused by tedious operation is greatly reduced, unification of input file automatic generation, task automatic submission, result analysis and high-throughput batch execution is achieved, and scientific research efficiency and data reuse capacity are remarkably improved.

Description

Technical Field

[0001] This invention relates to the fields of computational chemistry, electrochemistry, and materials science, specifically to a high-throughput computing method and system adapted to multi-scientific computing software. Background Technology

[0002] With the development of computational chemistry, electrochemistry, and materials science, scientific research increasingly relies on quantitative calculations and numerical simulations. Commonly used software includes VASP, Gaussian, CP2K, and QuantumESPRESSO. These software programs are powerful, but complex to use, have numerous parameters, and task configuration heavily depends on human experience. However, existing computational workflows suffer from the following problems: First, researchers typically need to manually write or modify input files, which vary significantly across different tasks and lacks a unified standard. For researchers without a programming background, the learning and operational costs are high, and each computation task often involves one-time script or file configurations, making direct reuse across different systems difficult. Second, when computational parameters or models change, scripts and input files often need to be readjusted, resulting in a large amount of repetitive work. Furthermore, inconsistent output formats across different computational software complicate data parsing and management, hindering large-scale result reuse and automated analysis. Finally, if researchers wish to perform high-throughput computations (such as batch calculations on hundreds of materials or parameter combinations), they usually need to manually generate input files, create directories, name tasks, and submit computations one by one. Each time the structure or parameters are modified, this process must be repeated, tedious, and prone to errors.

[0003] In summary, existing computing workflows have significant shortcomings in terms of automation, high throughput, and intelligence. There is an urgent need for a unified computing framework that can automatically generate input files, submit tasks, parse results, and perform high-throughput batch execution without relying on manual programming, in order to improve research efficiency and data reuse capabilities. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a high-throughput computing method and system adapted to multiple scientific computing software, thereby achieving unified abstraction and standardization of input parameters from different computing software, automatic generation of templated input files, task scheduling and result parsing, and high-throughput automated computing based on process templates.

[0005] To achieve the above objectives, the present invention provides a high-throughput computing method adaptable to multiple scientific computing software, comprising the following steps: The input files of various scientific computing software are identified and summarized to obtain a parameter list; Input the parameter list into the selected target computing software, generate an execution template based on the input file template of the target computing software, and set up the corresponding computing workstation; The computing workstation performs differential classification processing on the parameter list based on the set parameter configuration and structure file to obtain the classified parameters; The user input parameters are merged according to parameter priority, and the parameters after classification and the default parameters of the input file template are used to generate a complete parameter configuration set. The complete parameter configuration set is then substituted into the placeholders in the input file template to complete the template rendering and generate a target input file that meets the format requirements of the target computing software. The generated target input file and computation tasks are submitted to the unified scheduling platform for computation task scheduling, and the computation results are obtained and converted into structured data. Store structured data in a database.

[0006] Furthermore, the process of identifying parameters from the input files of various scientific computing software and summarizing them to obtain a parameter list is as follows: The input file provided by the user is parsed according to the syntax rules and structural specifications of the corresponding scientific computing software, and the location of configurable content in the input file is identified. If the configurable content meets any of the set conditions, then the corresponding configurable content is determined to be a candidate parameter item that has been successfully identified. Semantic analysis is performed on the successfully identified candidate parameter items, and the actual meaning of the candidate parameter items is extracted by combining industry-standard terminology, official computer software specifications, and preset semantic mapping rules. Candidate parameter items should be uniformly named based on their actual meaning; The named candidate parameter items are summarized and merged, and parameters with the same semantics or equivalent functions are merged. During the merging process, a unique parameter identifier is determined based on the scope of application of the candidate parameter item, the parameter type, and the degree of influence on the calculation results, and its correspondence in different calculation software is preserved; Duplicate candidate parameter items are deduplicated to form a unified parameter list.

[0007] Furthermore, the configurable content may take one or more of the following forms: a1. Replaceable numeric, enumerated, or string parameters; b1. Keywords with clear semantic identifiers and their values; c1. A structural data block representing molecular or crystal structure; d1. Control parameters related to computing resources, operating environment, or computing accuracy; The setting conditions include: a2. The input file contains key-value pairs and conforms to the syntax definition of the target computing software; b2. The input file template has a fixed structure but the content can be changed; c2. Different value requirements exist in multiple calculation tasks; d2. It is directly related to the computational objective, computational accuracy, or computational scale; The naming rules for uniformly naming candidate parameter items based on their actual meaning include: a3. Prioritize using industry-standard professional terms as parameter names; b3. For parameters that are synonymous across multiple software programs, a unified standard parameter name shall be adopted. c3. For parameters that are only applicable to specific computing software, use the standard parameter name + software identifier for naming; d3. For structured data block class parameters, use parameter names that can characterize the structure type and purpose.

[0008] Furthermore, the process of inputting the parameter list into the selected target computing software, generating an execution template based on the input file template of the target computing software, and setting up the corresponding computing workstation is as follows: Based on the target computing software and computing objectives selected by the user, an input file module that matches the target computing software and computing objectives is determined from a pre-built template management library; The parameter list is associated with the input file module. The correspondence between each parameter in the parameter list and the input file module is determined according to the parameter name, type and applicable scope. Based on the preset parameter adaptation rules, the parameters are format converted, position matched or structured to form a template-parameter binding relationship that can be used for subsequent calculation tasks. Based on the completed associated input file template, the parameters in the parameter list are substituted into the corresponding placeholders in the input file template, the input file template is rendered, and an execution template that meets the input format requirements of the target computing software is generated. Create a computing workstation corresponding to the execution template in the self-describing system.

[0009] Furthermore, the binding relationship when the computing workstation is created includes the target computing software identifier and computing target information, the corresponding execution template, and the associated parameter list and parameter configuration rules.

[0010] Furthermore, the computing workstation performs differential classification processing on the parameter list according to the set parameter configuration and structure file, and the process of obtaining the classified parameters is as follows: If a parameter in the parameter list is used to describe a calculation method, theoretical model, or general control option and can be directly replaced in different calculation tasks without depending on the context calculation results, then it is classified as a basic configuration parameter. If a parameter in the parameter list is used to describe a molecular structure or crystal structure, or to replace the corresponding structural description region in the input file, then it is classified as a structural data parameter. If a parameter in the parameter list needs to be calculated or corrected in combination with structural data, computational scale or context information before its value can be determined, or if it needs to be processed by preset calculation rules or algorithms before generating the target input file, then it is classified as a derived calculation parameter. If a parameter in the parameter list is only applicable to a specific scientific computing software and does not have cross-software general semantics, it is classified as a software-specific parameter.

[0011] Furthermore, the process of merging user input parameters according to parameter priority, classifying parameters, and default parameters of the input file template to generate a complete parameter configuration set is as follows: The user input parameters, categorized parameters, and default parameters of the input file template are sorted according to a preset parameter priority rule to form sorted parameters. The sorted parameters are grouped and arranged according to parameter category. Each parameter after arrangement contains one or more of the following: parameter name, parameter type, applicable software identifier, and default value information. The parameters are stored in a structured parameter format after being grouped and arranged. Each parameter has a unique parameter identifier, resulting in a complete parameter configuration set.

[0012] Furthermore, the process of generating the target input file that conforms to the format requirements of the target computing software is as follows: The input file template is parsed to identify pre-set parameter placeholders in the input file template; Based on the parameter naming rules, the parameter identifiers of each parameter in the complete parameter configuration set are matched one by one with the parameter placeholders in the input file template to determine the insertion position of the parameter in the input file template; For the basic configuration parameters in the complete parameter configuration set, their values ​​are directly filled into the corresponding placeholder positions in the input file template according to the corresponding relationship. For the structural data parameters in the complete parameter configuration set, the user-provided structural file is read from the pre-agreed input file, the structural data in the structural file is parsed, and the parsed structural data is converted or reorganized according to the input format requirements of the target calculation software to generate the target structural data block. The generated target structural data block is then filled into the corresponding structural data placeholder position in the input file template. For derived calculation parameters in the complete parameter configuration set, based on naming rules and parameter context relationships, preset calculation rules or constraints are invoked to calculate the basic configuration parameters, structural data parameters or calculation scale information to obtain the final value of the derived calculation parameter. After the value of the derived calculation parameter is determined, it is filled into the corresponding placeholder position in the input file template. After filling in all the parameters, the input file template is rendered as a whole through the template rendering mechanism to generate a target input file that conforms to the input format specifications of the target computing software.

[0013] Furthermore, the process of submitting the generated target input file and computation tasks to a unified scheduling platform for computation task scheduling, obtaining computation results, and converting them into structured data is as follows: In the computing workstation, the generated target input file and the corresponding computing task information are encapsulated to form a computing task request; The computing task request is submitted to the computing resource provider through the computing service interface encapsulated by the unified scheduling platform. The unified scheduling platform pre-maintains the running status information of multiple computing resource nodes and obtains the current running status of each computing resource node through the computing resource interface. The unified scheduling platform performs load assessment on multiple computing resource nodes based on the aforementioned operating status information, and allocates computing tasks to computing resource nodes with lower loads or that meet the execution conditions according to a preset scheduling strategy. When there are multiple computing tasks to be executed, the unified scheduling platform performs parallel scheduling of computing tasks based on scheduling strategies. During the execution of computing tasks, the unified scheduling platform periodically queries the execution status of computing tasks through the computing resource interface and synchronizes the execution status to the background management system. When the unified scheduling platform detects that a computing task is in an abnormal state or has failed to submit, it determines that the computing task has failed. For the computing task that has failed, the unified scheduling platform performs operations according to the preset fault handling rules. After the computing task is completed, the unified scheduling platform obtains the computing result file through the computing resource interface. After the computing result file is obtained, the platform cleans up the temporary directory or intermediate files on the computing resource side to release computing resources, and parses the obtained computing result file to convert the result data into a structured data format.

[0014] A high-throughput computing system adaptable to multiple scientific computing software includes: The parameter summarization module is used to identify parameters in the input files of various scientific computing software and summarize them to obtain a parameter list; The execution module is used to input the parameter list into the selected target computing software, generate an execution template based on the input file template of the target computing software, and set up the corresponding computing workstation. The parameter classification module is used by the workstation to perform differential classification processing on the parameter list based on the set parameter configuration and structure file, and obtain the classified parameters. The input file determination module is used to merge user input parameters, categorized parameters, and default parameters of the input file template according to parameter priority, generate a complete parameter configuration set, substitute the complete parameter configuration set into the placeholders of the input file template, complete the template rendering, and generate a target input file that meets the format requirements of the target computing software. The task scheduling module is used to submit the generated target input file and computing tasks to the unified scheduling platform, schedule computing tasks, obtain computing results and convert them into structured data; The storage module is used to store structured data into the database.

[0015] The present invention has the following beneficial effects: This invention effectively solves the problems of high learning and operation costs and difficulty in reuse caused by manually writing and modifying input files and lacking unified standards in existing computing processes. It achieves parameter identification and summarization of input files from multiple scientific computing software, generates execution templates and computing workstations adapted to the target software, classifies parameters differently and merges them to generate a complete parameter configuration set for automatic rendering of target input files, and realizes task scheduling, structured conversion and storage of results with the help of a unified scheduling platform. It avoids repetitive work when parameters or models change, and at the same time unifies the output data format to simplify parsing and management. It eliminates the need for manual processing of input file generation, task submission and other operations in high-throughput computing, greatly reducing the risk of errors caused by cumbersome operations. It realizes the unification of automatic input file generation, automatic task submission, result parsing and high-throughput batch execution, and significantly improves scientific research efficiency and data reuse capabilities.

[0016] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the high-throughput computing method adapted to multiple scientific computing software of the present invention.

[0018] Figure 2 This is a flowchart illustrating the process of generating a complete parameter configuration set for the high-throughput computing method adapted to multiple scientific computing software according to the present invention. Detailed Implementation

[0019] Please see Figure 1This invention provides a technical solution: a high-throughput computing method adapted to multiple scientific computing software, comprising the following steps: S1. Parameter identification is performed on the input files of each scientific computing software, and the parameters are summarized to obtain a parameter list.

[0020] It should be noted that scientific computing software includes, but is not limited to, ORCA, Gaussian, NWChem, QE, and VASP. The specific process for obtaining the parameter list is as follows: The system parses the user-provided input file according to the syntax rules and structural specifications of the corresponding scientific computing software, and identifies the location of configurable content in the input file.

[0021] If any configurable content meets any of the set conditions, then the corresponding configurable content is determined to be a successfully identified candidate parameter item.

[0022] Semantic analysis is performed on the successfully identified candidate parameter items, and the actual meaning of the candidate parameter items is extracted by combining industry-standard terminology, official computer software specifications, and preset semantic mapping rules.

[0023] Candidate parameter items are uniformly named based on their actual meaning.

[0024] The named candidate parameters are aggregated, and parameters with the same semantics or equivalent functions are merged.

[0025] During the merging process, a unique parameter identifier is determined based on the applicable scope, parameter type, and degree of influence of the candidate parameter item on the calculation results, and its correspondence in different calculation software is preserved.

[0026] Duplicate candidate parameter items are deduplicated to form a unified parameter list.

[0027] By clearly defining the complete process of parameter identification, filtering, semantic analysis, unified naming, and deduplication, this method achieves unified abstraction and standardization of input parameters for multiple scientific computing software, effectively solving the problems of numerous, ambiguous, and redundant synonymous parameters across different software in existing technologies. By parsing software syntax rules, extracting the actual meaning of parameters, and establishing cross-software correspondences, it eliminates the chaos in parameter management, allowing researchers without a programming background to clearly understand the purpose of parameters. This also lays the foundation for subsequent multi-software adaptation and parameter reuse, significantly reducing the learning and operational costs of parameter configuration.

[0028] For example, a uniform parameter list format is shown in Table 1: Table 1 Parameter List ; Furthermore, the configurable content can take one or more of the following forms: a1. Replaceable numeric, enumerated, or string parameters; b1. Keywords with clear semantic identifiers and their values; c1. A structural data block representing molecular or crystal structure; d1. Control parameters related to computing resources, operating environment, or computing accuracy; The settings include: a2. The input file contains key-value pairs and conforms to the syntax definition of the target computing software.

[0029] b2. The input file template has a fixed structure but the content can be changed.

[0030] c2. There are different value requirements in multiple calculation tasks.

[0031] d2. It is directly related to the calculation objective, calculation accuracy, or calculation scale.

[0032] The naming rules for uniformly naming candidate parameter items based on their actual meaning include: a3. Prioritize using industry-standard professional terms as parameter names.

[0033] b3. For parameters that are synonymous in multiple software programs, use a unified standard parameter name.

[0034] c3. For parameters that are only applicable to specific computing software, use the standard parameter name + software identifier for naming.

[0035] d3. For structured data block class parameters, use parameter names that can characterize the structure type and purpose.

[0036] This paper clarifies the specific forms of configurable content, parameter filtering conditions, and unified naming rules, precisely solving the problems of adaptation difficulties and confusion caused by the lack of standard parameter identification and inconsistent naming in existing technologies. The defined configurable content covers core types such as numerical values, keywords, and structured data blocks. The filtering conditions ensure that only parameters with practical significance for calculation are extracted, while the naming rules balance industry universality with software specificity. This avoids cross-software parameter confusion and makes the purpose of special parameters such as structured data blocks intuitively identifiable, significantly improving the accuracy, consistency, and adaptation efficiency of parameter identification. It also provides a standardized parameter foundation for subsequent template binding and input file generation.

[0037] S2. Input the parameter list into the selected target computing software, generate the execution template based on the input file template of the target computing software, and set up the corresponding computing workstation.

[0038] It should be noted that the specific process is as follows: Based on the target computing software and computing objectives selected by the user, an input file module matching the target computing software and computing objectives is determined from a pre-built template management library. The input file template is a standardized template constructed by a combination of manual configuration and automatic generation, based on the input specifications of the target computing software and combined with preset computing objective rules.

[0039] The parameter list is associated with the input file module. The correspondence between each parameter in the parameter list and the input file module is determined according to the parameter name, type and applicable scope. Based on the preset parameter adaptation rules, the parameters are format converted, position matched or structured to form a template-parameter binding relationship that can be used for subsequent calculation tasks.

[0040] Based on the completed associated input file template, the parameters in the parameter list are substituted into the corresponding placeholders in the input file template, the input file template is rendered, and an execution template that meets the input format requirements of the target computing software is generated. The execution template is stored and managed as the basic template for generating target input files for subsequent computing tasks.

[0041] Create a computing workstation corresponding to the execution template in the self-describing system.

[0042] This document outlines the process for generating execution templates and setting up computing workstations, specifically addressing the pain points of existing technologies, such as disorganized input file templates, cumbersome manual editing and adjustments, and scattered computing configurations. It rapidly matches target software with computing targets through a pre-built template management library, achieving precise binding and format adaptation of parameters and templates, and automatically rendering compliant execution templates without requiring the manual creation of complex input files. The constructed computing workstation provides a unified entry point for computing configuration, integrating core elements such as parameters, templates, and software identifiers, reducing computational failures caused by template incompatibility and incorrect parameter placement, and significantly improving the standardization of input file generation and the efficiency of computing configuration.

[0043] Furthermore, the binding relationships during the creation of the computing workstation include the target computing software identifier and computing target information, the corresponding execution template, and the associated parameter list and parameter configuration rules.

[0044] Computing workstations provide users with a unified entry point for computing configuration and serve as the configuration carrier for subsequent computing task submission and scheduling. Created computing workstations are registered in the backend management system, forming workstation instances that can be accessed and reused by users. While not directly executing computing tasks, computing workstations support the management of templates, parameters, and computing configurations, and provide self-describing information for triggering actual computing tasks through the unified scheduling platform.

[0045] This invention clarifies the core binding relationships during workstation creation, resolving issues such as scattered computing configuration information, difficulties in workstation reuse, and a lack of clear scheduling criteria in existing technologies. It endows workstations with complete self-descriptive attributes, clearly associating them with target software, computing objectives, execution templates, parameter lists, and configuration rules. This not only facilitates users in quickly identifying suitable workstation scenarios and supports workstation reuse and centralized management, but also provides precise configuration criteria for the subsequent unified scheduling platform. This avoids task scheduling errors caused by missing or disorganized configuration information, ensuring the continuity and reliability of the computing process.

[0046] S3. The computing workstation performs differential classification processing on the parameter list according to the set parameter configuration and structure file to obtain the classified parameters.

[0047] Furthermore, the process of differentiated classification is as follows: If a parameter in the parameter list describes a calculation method, theoretical model, or general control option and can be directly replaced in different calculation tasks without depending on the context calculation results, then it is classified as a basic configuration parameter.

[0048] If a parameter in the parameter list is used to describe a molecular structure or crystal structure, including data blocks extracted and converted from a user-provided structure file, or used to replace the corresponding structure description region in the input file, then it is classified as a structure data parameter.

[0049] If a parameter in the parameter list requires secondary calculation or correction based on structural data, computational scale, or contextual information before its value can be determined, or if it needs to be processed by preset calculation rules or algorithms before generating the target input file, then it is classified as a derived calculation parameter.

[0050] If a parameter in the parameter list is only applicable to a specific scientific computing software and does not have cross-software general semantics, it is classified as a software-specific parameter.

[0051] By categorizing parameters according to their functions and characteristics, this approach effectively solves the problem of disorganized and difficult-to-target parameters in existing technologies. It divides parameters into four categories: basic configuration, structured data, derived calculations, and software-specific parameters, allowing for precise adaptation and processing of different parameter types: basic configuration parameters can be directly reused, structured data parameters are easily converted and recombined, derived calculation parameters can be automatically corrected through preset rules, and software-specific parameters avoid cross-software misuse. This classification method improves the targeting and efficiency of parameter processing, reduces configuration errors caused by parameter type confusion, and provides a clear logical basis for subsequent parameter filling and template rendering, facilitating the automated and accurate generation of input files.

[0052] S4. Merge user input parameters according to parameter priority, classify parameters and default parameters of input file template to generate a complete parameter configuration set, substitute the complete parameter configuration set into the placeholders of input file template to complete template rendering, and generate a target input file that meets the format requirements of the target computing software.

[0053] like Figure 2 As shown, the specific process for generating a complete parameter configuration set is as follows: The user input parameters, categorized parameters, and default parameters of the input file template are sorted according to the preset parameter priority rules to form sorted parameters.

[0054] The parameter priority is determined based on at least the following factors: User-defined input parameters have higher priority than template default parameters.

[0055] Derived calculation parameters have higher priority than basic configuration parameters.

[0056] Structural data parameters have a higher priority than ordinary control parameters.

[0057] The sorted parameters are grouped and arranged according to parameter category. Each parameter after arrangement includes one or more of the following: parameter name, parameter type, applicable software identifier, and default value information.

[0058] The parameters are stored in a structured parameter format after being grouped and arranged. Each parameter has a unique parameter identifier, resulting in a complete parameter configuration set.

[0059] By prioritizing parameters and grouping them for storage to generate a complete parameter configuration set, this technology resolves the issues of parameter conflicts and incomplete configurations from different sources found in existing technologies. Its preset priority rules clearly define the priorities of user-input parameters, categorized parameters, and default parameters, avoiding value conflicts. Grouping and assigning unique identifiers ensures clear and orderly parameter configurations, facilitating retrieval and adjustment. The complete parameter configuration set ensures the integrity of parameters in the target input file, reducing the tedious manual parameter addition and avoiding calculation deviations caused by missing parameters or confused priorities, providing reliable parameter support for template rendering and input file generation.

[0060] The process of generating a target input file that conforms to the format requirements of the target computing software is as follows: The input file template is parsed to identify pre-set parameter placeholders in the input file template.

[0061] Based on the parameter naming rules, the parameter identifiers of each parameter in the complete parameter configuration set are matched one by one with the parameter placeholders in the input file template to determine the insertion position of the parameter in the input file template.

[0062] For the basic configuration parameters in the complete parameter configuration set, the computing workstation directly fills the parameter values ​​into the corresponding placeholder positions in the input file template according to the corresponding relationship, thus completing the rendering process of the basic configuration parameters.

[0063] For the structural data parameters in the complete parameter configuration set, the computing workstation reads the user-provided structural file from the pre-agreed input file, parses the structural data in the structural file, performs format conversion or reorganization on the parsed structural data according to the input format requirements of the target computing software, generates the target structural data block, and fills the corresponding structural data placeholder position in the input file template with the generated target structural data block.

[0064] For derived calculation parameters in the complete parameter configuration set, the computing workstation calls the preset calculation rules or constraints based on the naming rules and parameter context relationship to calculate the basic configuration parameters, structural data parameters or calculation scale information to obtain the final value of the derived calculation parameter. After the value of the derived calculation parameter is determined, it is filled into the corresponding placeholder position in the input file template.

[0065] After filling in all the parameters, the input file template is rendered as a whole through the template rendering mechanism to generate a target input file that conforms to the input format specification of the target computing software. The target input file is stored and used as the input file for subsequent computing tasks to be submitted and executed.

[0066] This paper describes the process of generating target input files, specifically addressing the pain points of existing technologies, such as reliance on manual generation, susceptibility to formatting errors, and difficulties in adapting to different software. It achieves automated and accurate generation of input files by parsing template placeholders and filling parameters according to their type (direct filling of basic parameters, format conversion of structured data parameters, and automatic calculation of derived parameters). This process eliminates the need for manual writing or adjusting of file formats, adapting to the input specifications of different target software while avoiding oversights from manual operations. This significantly improves the efficiency and accuracy of input file generation in high-throughput computing, while lowering the operational threshold for researchers without a programming background.

[0067] S5. Submit the generated target input file and computation tasks to the unified scheduling platform for computation task scheduling, obtain computation results and convert them into structured data.

[0068] It should be noted that the specific process is as follows: In the computing workstation, the generated target input file and the corresponding computing task information are encapsulated to form a computing task request. The standardized computing task request includes at least the target input file, the target computing software identifier, and the computing task parameters and running configuration.

[0069] The computing task requests are submitted to the computing resource provider through the computing service interface encapsulated by the unified scheduling platform. The unified scheduling platform pre-maintains the running status information of multiple computing resource nodes and obtains the current running status of each computing resource node through the computing resource interface. The running status information includes at least one of the following: the number of currently queued tasks; the number of tasks currently being executed; and the availability of computing resources.

[0070] The unified scheduling platform assesses the load of multiple computing resource nodes based on the operational status information and allocates computing tasks to computing resource nodes with lower loads or those that meet the execution conditions according to a preset scheduling strategy.

[0071] When multiple computational tasks are pending execution, the unified scheduling platform performs parallel scheduling of these tasks based on scheduling policies. Parallel scheduling includes at least one of the following methods: allocating multiple computational tasks to different computing resource nodes for parallel execution; or executing multiple computational tasks in queue order within the same computing resource node.

[0072] During the execution of computational tasks, the unified scheduling platform periodically queries the execution status of the tasks through the computing resource interface and synchronizes the execution status to the backend management system. The execution status includes at least: task queuing status, task running status, task completion status, and task exception status.

[0073] When the unified scheduling platform detects that a computing task is in an abnormal state or has failed to submit, it determines that the computing task has failed. For the failed computing task, the unified scheduling platform performs operations according to preset fault handling rules, including at least one of the following operations: Re-upload the target input file and resubmit the calculation task.

[0074] The computing tasks are rescheduled to other available computing resource nodes.

[0075] Mark abnormal tasks and record fault information.

[0076] Once the computation task is completed, the unified scheduling platform obtains the computation result file through the computation resource interface. After obtaining the computation result file, it cleans up the temporary directory or intermediate files on the computation resource side to release computation resources. It then parses and processes the obtained computation result file, converts the result data into a structured data format, and returns the converted structured data to the computation workstation for subsequent data storage, analysis, or display.

[0077] This system standardizes the processes of task scheduling, result acquisition, and structured data conversion, effectively solving the problems of cumbersome task submission, unreasonable resource allocation, and complex result parsing in existing high-throughput computing technologies. It automates task submission through a unified scheduling platform, improves computing resource utilization based on resource node load assessment and parallel scheduling strategies, and shortens the high-throughput computing cycle. Real-time status monitoring and fault handling mechanisms reduce task failure rates and ensure stable computing processes. The system automatically parses computation results into structured data, unifies the output formats of different software, solves the problems of complex data parsing and management, facilitates result reuse and automated analysis, significantly reduces the workload of manual task submission, status monitoring, and data processing, and significantly improves research efficiency.

[0078] S6. Store the structured data in the database.

[0079] A high-throughput computing system adaptable to multiple scientific computing software includes: The parameter summarization module is used to identify parameters in the input files of various scientific computing software and summarize them to obtain a parameter list.

[0080] The execution module is used to input the parameter list into the selected target computing software, generate an execution template based on the input file template of the target computing software, and set up the corresponding computing workstation.

[0081] The parameter classification module is used by the workstation to perform differential classification processing on the parameter list based on the set parameter configuration and structure file, and obtain the classified parameters.

[0082] The input file determination module is used to merge user input parameters, categorized parameters, and default parameters of the input file template according to parameter priority, generate a complete parameter configuration set, substitute the complete parameter configuration set into the placeholders of the input file template, complete the template rendering, and generate a target input file that meets the format requirements of the target computing software.

[0083] The task scheduling module is used to submit the generated target input file and computation tasks to the unified scheduling platform for computation task scheduling, obtain computation results and convert them into structured data.

[0084] The storage module is used to store structured data into the database.

[0085] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] This invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0088] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0089] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0090] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A high-throughput computing method adaptable to multiple scientific computing software, characterized in that, Includes the following steps: The input files of various scientific computing software are identified and summarized to obtain a parameter list; Input the parameter list into the selected target computing software, generate an execution template based on the input file template of the target computing software, and set up the corresponding computing workstation; The computing workstation performs differential classification processing on the parameter list based on the set parameter configuration and structure file to obtain the classified parameters; The user input parameters are merged according to parameter priority, and the parameters after classification and the default parameters of the input file template are used to generate a complete parameter configuration set. The complete parameter configuration set is then substituted into the placeholders in the input file template to complete the template rendering and generate a target input file that meets the format requirements of the target computing software. The generated target input file and computation tasks are submitted to the unified scheduling platform for computation task scheduling, and the computation results are obtained and converted into structured data. Store structured data in a database.

2. The high-throughput computing method adapted to multiple scientific computing software according to claim 1, characterized in that, The process of identifying parameters from the input files of various scientific computing software and summarizing them to obtain a parameter list is as follows: The input file provided by the user is parsed according to the syntax rules and structural specifications of the corresponding scientific computing software, and the location of configurable content in the input file is identified. If the configurable content meets any of the set conditions, then the corresponding configurable content is determined to be a candidate parameter item that has been successfully identified. Semantic analysis is performed on the successfully identified candidate parameter items, and the actual meaning of the candidate parameter items is extracted by combining industry-standard terminology, official computer software specifications, and preset semantic mapping rules. Candidate parameter items should be uniformly named based on their actual meaning; The named candidate parameter items are summarized and merged, and parameters with the same semantics or equivalent functions are merged. During the merging process, a unique parameter identifier is determined based on the scope of application of the candidate parameter item, the parameter type, and the degree of influence on the calculation results, and its correspondence in different calculation software is preserved; Duplicate candidate parameter items are deduplicated to form a unified parameter list.

3. The high-throughput computing method adapted to multiple scientific computing software according to claim 2, characterized in that: The configurable content is one or more of the following forms: a1. Replaceable numeric, enumerated, or string parameters; b1. Keywords with clear semantic identifiers and their values; c1. A structural data block representing molecular or crystal structure; d1. Control parameters related to computing resources, operating environment, or computing accuracy; The setting conditions include: a2. The input file contains key-value pairs and conforms to the syntax definition of the target computing software; b2. The input file template has a fixed structure but the content can be changed; c2. Different value requirements exist in multiple calculation tasks; d2. It is directly related to the computational objective, computational accuracy, or computational scale; The naming rules for uniformly naming candidate parameter items based on their actual meaning include: a3. Prioritize using industry-standard professional terms as parameter names; b3. For parameters that are synonymous across multiple software programs, a unified standard parameter name shall be adopted. c3. For parameters that are only applicable to specific computing software, use the standard parameter name + software identifier for naming; d3. For structured data block class parameters, use parameter names that can characterize the structure type and purpose.

4. The high-throughput computing method adapted to multiple scientific computing software according to claim 1, characterized in that, The process of inputting the parameter list into the selected target computing software, generating an execution template based on the input file template of the target computing software, and setting up the corresponding computing workstation is as follows: Based on the target computing software and computing objectives selected by the user, an input file module that matches the target computing software and computing objectives is determined from a pre-built template management library; The parameter list is associated with the input file module. The correspondence between each parameter in the parameter list and the input file module is determined according to the parameter name, type and applicable scope. Based on the preset parameter adaptation rules, the parameters are format converted, position matched or structured to form a template-parameter binding relationship that can be used for subsequent calculation tasks. Based on the completed associated input file template, the parameters in the parameter list are substituted into the corresponding placeholders in the input file template, the input file template is rendered, and an execution template that meets the input format requirements of the target computing software is generated. Create a computing workstation corresponding to the execution template in the self-describing system.

5. A high-throughput computing method adapted to multiple scientific computing software according to claim 4, characterized in that, The binding relationship when the computing workstation is created includes the target computing software identifier and computing target information, the corresponding execution template, and the associated parameter list and parameter configuration rules.

6. The high-throughput computing method adapted to multiple scientific computing software according to claim 1, characterized in that, The computing workstation performs differential classification processing on the parameter list based on the set parameter configuration and structure file, and the process of obtaining the classified parameters is as follows: If a parameter in the parameter list is used to describe a calculation method, theoretical model, or general control option and can be directly replaced in different calculation tasks without depending on the context calculation results, then it is classified as a basic configuration parameter. If a parameter in the parameter list is used to describe a molecular structure or crystal structure, or to replace the corresponding structural description region in the input file, then it is classified as a structural data parameter. If a parameter in the parameter list needs to be calculated or corrected in combination with structural data, computational scale or context information before its value can be determined, or if it needs to be processed by preset calculation rules or algorithms before generating the target input file, then it is classified as a derived calculation parameter. If a parameter in the parameter list is only applicable to a specific scientific computing software and does not have cross-software general semantics, it is classified as a software-specific parameter.

7. A high-throughput computing method adapted to multiple scientific computing software according to claim 1, characterized in that, The process of merging user input parameters according to parameter priority, classifying the parameters, and generating a complete parameter configuration set based on the default parameters of the input file template is as follows: The user input parameters, categorized parameters, and default parameters of the input file template are sorted according to a preset parameter priority rule to form sorted parameters. The sorted parameters are grouped and arranged according to parameter category. Each parameter after arrangement contains one or more of the following: parameter name, parameter type, applicable software identifier, and default value information. The parameters are stored in a structured parameter format after being grouped and arranged. Each parameter has a unique parameter identifier, resulting in a complete parameter configuration set.

8. A high-throughput computing method adapted to multiple scientific computing software according to claim 1, characterized in that, The process of generating a target input file that conforms to the format requirements of the target computing software is as follows: The input file template is parsed to identify pre-set parameter placeholders in the input file template; Based on the parameter naming rules, the parameter identifiers of each parameter in the complete parameter configuration set are matched one by one with the parameter placeholders in the input file template to determine the insertion position of the parameter in the input file template; For the basic configuration parameters in the complete parameter configuration set, their values ​​are directly filled into the corresponding placeholder positions in the input file template according to the corresponding relationship. For the structural data parameters in the complete parameter configuration set, the user-provided structural file is read from the pre-agreed input file, the structural data in the structural file is parsed, and the parsed structural data is converted or reorganized according to the input format requirements of the target calculation software to generate the target structural data block. The generated target structural data block is then filled into the corresponding structural data placeholder position in the input file template. For derived calculation parameters in the complete parameter configuration set, based on naming rules and parameter context relationships, preset calculation rules or constraints are invoked to calculate the basic configuration parameters, structural data parameters or calculation scale information to obtain the final value of the derived calculation parameter. After the value of the derived calculation parameter is determined, it is filled into the corresponding placeholder position in the input file template. After filling in all the parameters, the input file template is rendered as a whole through the template rendering mechanism to generate a target input file that conforms to the input format specifications of the target computing software.

9. A high-throughput computing method adapted to multiple scientific computing software according to claim 1, characterized in that, The process of submitting the generated target input file and computation tasks to the unified scheduling platform for computation task scheduling, obtaining computation results, and converting them into structured data is as follows: In the computing workstation, the generated target input file and the corresponding computing task information are encapsulated to form a computing task request; The computing task request is submitted to the computing resource provider through the computing service interface encapsulated by the unified scheduling platform. The unified scheduling platform pre-maintains the running status information of multiple computing resource nodes and obtains the current running status of each computing resource node through the computing resource interface. The unified scheduling platform performs load assessment on multiple computing resource nodes based on the aforementioned operating status information, and allocates computing tasks to computing resource nodes with lower loads or that meet the execution conditions according to a preset scheduling strategy. When there are multiple computing tasks to be executed, the unified scheduling platform performs parallel scheduling of computing tasks based on scheduling strategies. During the execution of computing tasks, the unified scheduling platform periodically queries the execution status of computing tasks through the computing resource interface and synchronizes the execution status to the background management system. When the unified scheduling platform detects that a computing task is in an abnormal state or has failed to submit, it determines that the computing task has failed. For the computing task that has failed, the unified scheduling platform performs operations according to the preset fault handling rules. After the computing task is completed, the unified scheduling platform obtains the computing result file through the computing resource interface. After the computing result file is obtained, the platform cleans up the temporary directory or intermediate files on the computing resource side to release computing resources, and parses the obtained computing result file to convert the result data into a structured data format.

10. A high-throughput computing system adaptable to multiple scientific computing software, characterized in that, include: The parameter summarization module is used to identify parameters in the input files of various scientific computing software and summarize them to obtain a parameter list; The execution module is used to input the parameter list into the selected target computing software, generate an execution template based on the input file template of the target computing software, and set up the corresponding computing workstation. The parameter classification module is used by the workstation to perform differential classification processing on the parameter list based on the set parameter configuration and structure file, and obtain the classified parameters. The input file determination module is used to merge user input parameters, categorized parameters, and default parameters of the input file template according to parameter priority, generate a complete parameter configuration set, substitute the complete parameter configuration set into the placeholders of the input file template, complete the template rendering, and generate a target input file that meets the format requirements of the target computing software. The task scheduling module is used to submit the generated target input file and computing tasks to the unified scheduling platform, schedule computing tasks, obtain computing results and convert them into structured data; The storage module is used to store structured data into the database.

Images