Abstract: A full-link reconciliation method based on a snowflake algorithm, an apparatus, a device, and a medium are provided.

Abstract: Disclosed is an operation and maintenance method and system for automatically and uniformly managing nodes of bastion host. No matter distribution of user permission, daily deployment of machine monitoring and network monitoring or batch management of daily operation and maintenance, the operation and maintenance method and system can be abstracted as follows: master control dispatches a Master of a certain area node to issue and execute a certain task, and unified management is naturally achieved; the design concept can be continued subsequently.

Abstract: Disclosed is a privacy-preserved method for constructing an aggregate thermal dynamic model of buildings, including the steps of: establishing a thermal dynamic model of one building region, and establishing an aggregate thermal dynamic model of buildings based on an aggregation equation; performing parameter estimation using a least square method based on a measurement equation, introducing a regular term to solve a sparsity problem, and obtaining a parameter estimation model in a compact form for the aggregate thermal dynamic model of buildings; and establishing a privacy-preserved parameter estimation method for the aggregate thermal dynamic model of buildings. Based on the technique, the aggregate modeling is performed on numerous buildings by a building load aggregator to participate in the operation and control of an energy system while preserving privacy of building users, promoting the mining of thermal inertia of buildings and enhancing the flexibility of operation and regulation of a power system.

Abstract: A large language model, previously pre-trained on multiple source code modeling tasks, is pre-trained, through curriculum learning, to learn to predict a code execution trace given a source code program. The model is pre-trained using a variety of pre-training datasets consisting of pairs of a source code sample and a corresponding execution trace. The curriculum pre-training starts with a pre-training dataset of single line executions and adds in additional pre-training datasets with more increasing complex behaviors. The pre-training datasets include mutation-augmented source code samples and their corresponding execution traces.

Abstract: A deep learning model is pre-trained with a large-scale of unsupervised data of code review tasks in order to learn the relationships between code changes and a code review. The pre-trained deep learning model predicts a code review given a code diff hunk in a code diff format. The code diff hunk includes the changed code and its surrounding context. The pre-trained deep learning model may then be fine-tuned with supervised data in order to make predictions for several code review activities, such as, code change quality estimation and code refinement.

Abstract: Disclosed is an analytical method for dynamic analysis of a natural gas network in the field of energy system modeling and operational analysis, which includes establishing adynamic model of natural gas transmission according to the conservation equations, and reconstructing the dynamic model into the equations in a heat conduction equation form. The present disclosure directly constructs an analytical method for dynamic analysis of a natural gas network, avoiding approximation errors, numerical dispersion, and dissipation compared with the traditional numerical methods. The discretization process is avoided during the solution, greatly improving the computational efficiency and solution accuracy of dynamic analysis of the natural gas network.

Abstract: A routing-policy-based global user compliance access method and an apparatus are provided. The method includes steps of: initiating, by a user, a login request to a unified user gateway if there is a user login behavior; sending, by an authentication unit, an authenticated user request to a local user center for processing, and to the global user center for global query if the user is a non-local user; initiating, by the user, a business request to the unified user gateway if there is a user business operation behavior; routing the request to an application program interface after being passed by the authentication unit; sending, by the application program interface, the business request to the local user center for business processing, and to a global business center for routing policy query and redirecting to the remote user center to which the user belongs if the user is the non-local user.

Abstract: An operation and maintenance method and system for automatically and uniformly managing nodes of bastion host can be abstracted as follows: master control dispatches a Master of a certain area node to issue and execute a certain task, and unified management is naturally achieved; the design concept can be continued subsequently. A Master host in the node serves as a master controller of the node, and related contents including a tool script library, a crontab task and a configuration file are preset in a Redis of the node in advance; when a new machine is accessed to a certain node, the new machine performs Salt-Master access management of the node where the new machine is located, and the corresponding machine is controlled to complete the corresponding task through a takeover program; therefore, unified and automatic management is realized.

Abstract: A static voltage stability margin evaluation method and system, and a terminal device are related to the field of integrated energy system operation. The method includes the following steps: establishing a thermal dynamic model of a heating system; establishing a thermoelectric coupling device model; establishing a static voltage stability margin model of an electric power system that considers thermal dynamics of the heating system; and solving the model to obtain a voltage stability margin. In the present invention, a static voltage stability margin that considers thermal dynamics of a heating system can be obtained, and a Pareto boundary of the static voltage stability margin that considers the thermal dynamics can be obtained through a dual-objective nonlinear optimization method, so that an impact of thermoelectric coupling on voltage stability and an impact of thermal inertia of the heating system on a voltage stability margin can be revealed.

Abstract: A comprehensive management method and apparatus for application traffic, and an electric device are provided. The method includes: marking a traffic request with a label, and transmitting the traffic request with the label to a user access layer gateway; managing and configuring rules; receiving the traffic request; analyzing the traffic request based on a traffic analyzing rule; identifying and filtering the traffic request based on a traffic filtering rule; performing secondary processing on the traffic request based on a traffic processing rule; and forwarding the traffic request after secondary processing to an application proxy gateway based on a traffic forwarding rule; and performing an in-traffic proxy and an out-traffic proxy.

Abstract: A neural transformer model with attention is trained to predict candidates to complete a line of source code with a zero-inference capability. The model is trained on an unsupervised training dataset that includes features from source code written in multiple programming languages. The features include a file-level context and a local context, where the file-level context includes a global context, a class context, a function context, and/or a method context for each class, function and/or method of the source code programs used in the training dataset. The local context includes method bodies, function bodies, and/or stand-alone code of main method routines. From these features, the model is able to learn to predict an ordered sequence of code elements that complete a line of source code in a programming language seen and not seen during training.

Abstract: A large language model, previously pre-trained on multiple source code modeling tasks, is pre-trained, through curriculum learning, to learn to predict a code execution trace given a source code program. The model is pre-trained using a variety of pre-training datasets consisting of pairs of a source code sample and a corresponding execution trace. The curriculum pre-training starts with a pre-training dataset of single line executions and adds in additional pre-training datasets with more increasing complex behaviors. The pre-training datasets include mutation-augmented source code samples and their corresponding execution traces.

Abstract: A dispatching method for an electric-hydrogen energy system considering flexible hydrogen demand includes establishing an electric load flexibility equation, a power purchase and sale constraint equation, a renewable energy output constraint equation, a hydrogen load flexibility equation, an electricity-to-hydrogen production safety operation constraint equation and an electric power balance constraint equation, establishing an electric-hydrogen energy system dispatching model with the lowest operating cost of the electric-hydrogen energy system within the dispatching cycle as an objective function, and solving the electric-hydrogen energy system dispatching model to obtain an optimal dispatching result. As compared with the prior art, the present invention can effectively solve the problem of coordination between electric and hydrogen energy flows, while taking into account the flexibility of electric and hydrogen loads, further providing additional flexibility to the operation of the system.

Abstract: A retrieval-augmented code completion system uses the context of a partially-formed source code snippet of a source code program and a hint to predict the source code tokens needed to complete the partially-formed source code snippet. The hint is a source code segment that completes a semantically-similar source code segment of the partially-formed source code snippet. The hint is found in a retrieval source code database using a hybrid retrieval technique. A deep learning decoder model uses the context of the partially-formed source code snippet and the hint to predict the most likely candidate sequence of source code tokens to complete the partially-formed source code snippet.

Abstract: A deep learning model is pre-trained with a large-scale of unsupervised data of code review tasks in order to learn the relationships between code changes and a code review. The pre-trained deep learning model predicts a code review given a code diff hunk in a code diff format. The code diff hunk includes the changed code and its surrounding context. The pre-trained deep learning model may then be fine-tuned with supervised data in order to make predictions for several code review activities, such as, code change quality estimation and code refinement.

Abstract: A neural transformer model with attention is trained to predict candidates to complete a line of source code with a zero-inference capability. The model is trained on an unsupervised training dataset that includes features from source code written in multiple programming languages. The features include a file-level context and a local context, where the file-level context includes a global context, a class context, a function context, and/or a method context for each class, function and/or method of the source code programs used in the training dataset. The local context includes method bodies, function bodies, and/or stand-alone code of main method routines. From these features, the model is able to learn to predict an ordered sequence of code elements that complete a line of source code in a programming language seen and not seen during training.

Abstract: A retrieval-augmented code completion system uses the context of a partially-formed source code snippet of a source code program and a hint to predict the source code tokens needed to complete the partially-formed source code snippet. The hint is a source code segment that completes a semantically-similar source code segment of the partially-formed source code snippet. The hint is found in a retrieval source code database using a hybrid retrieval technique. A deep learning decoder model uses the context of the partially-formed source code snippet and the hint to predict the most likely candidate sequence of source code tokens to complete the partially-formed source code snippet.

Abstract: A neural transformer model with attention is trained to predict candidates to complete a line of source code with a zero-inference capability. The model is trained on an unsupervised training dataset that includes features from source code written in multiple programming languages. The features include a file-level context and a local context, where the file-level context includes a global context, a class context, a function context, and/or a method context for each class, function and/or method of the source code programs used in the training dataset. The local context includes method bodies, function bodies, and/or stand-alone code of main method routines. From these features, the model is able to learn to predict an ordered sequence of code elements that complete a line of source code in a programming language seen and not seen during training.

Abstract: A neural transformer model with attention is trained to predict candidates to complete a line of source code with a zero-inference capability. The model is trained on an unsupervised training dataset that includes features from source code written in multiple programming languages. The features include a file-level context and a local context, where the file-level context includes a global context, a class context, a function context, and/or a method context for each class, function and/or method of the source code programs used in the training dataset. The local context includes method bodies, function bodies, and/or stand-alone code of main method routines. From these features, the model is able to learn to predict an ordered sequence of code elements that complete a line of source code in a programming language seen and not seen during training.