Abstract: Implementations for performing workload-aware space reclamation are provided. One example includes a computing system comprising: processing circuitry and memory storing instructions that, during execution, causes the processing circuitry to: track extent usage characteristics for each of a plurality of extents storing data for a graph database; and perform memory reclamation to reclaim storage areas corresponding to invalid data, wherein performing the memory reclamation comprises: generating a list of extents to be reclaimed based on the tracked extent usage characteristics; and reclaiming memory space on the plurality of extents in order based on the list of extents.

Abstract: Implementations for I/O efficient synchronization on a graph database are provided. One example includes a computing system comprising: processing circuitry and memory storing instructions that, during execution, causes the processing circuitry to: implement the graph database using at least: a read-write node and a read-only node operating in cache memory; and a shared storage for data persistence, wherein the shared storage stores a mapping table of the graph database; receive an update query; update the read-write node based on the update query; write, using the read-write node, a set of logs corresponding to the update query to a write-ahead log journal in the shared storage; and retrieve, using the read-only node, the set of logs from the write-ahead log journal.

Abstract: Implementations for a space-optimized graph database system are provided. One implementation includes a computing system comprising: processing circuitry and memory storing instructions that causes the processing circuitry to: store a graph database comprising an initial tree graph storing a plurality of data entries, each data entry comprising a respective field identifier; receive a query to update the graph database, wherein the query comprises a request to add a new data entry; determine a splitting event to perform based on one or more predetermined criteria; generate a new tree graph corresponding to a field identifier of the new data entry by splitting off a subset of the plurality of data entries of the initial tree graph, wherein the subset comprises all data entries of the initial tree graph that correspond to the field identifier of the new data entry; and update the new tree graph in accordance with the query.

Abstract: A method of manufacturing a memory device at least includes the following steps. A first interconnect and a first dielectric layer are formed on a substrate. A first chemical mechanical polishing process is performed on the first dielectric layer. A stack structure is formed over the first dielectric layer and a staircase structure is formed in the stack structure. A second dielectric layer is formed on the substrate to cover the stack structure and the staircase structure. A second chemical mechanical polishing process is performed on the second dielectric layer. A depth of second grooves of a second polishing pad used in the second chemical mechanical polishing process is smaller than a depth of first grooves of a first polishing pad used in the first chemical mechanical polishing process. The memory device may be a 3D NAND flash memory with high capacity and high performance.

Abstract: The present disclosure is related to a file processing method, electronic apparatus and storage medium, the file processing method including: updating a file list according to file access information for a file in a storage device, wherein the file list contains file inode numbers of files to be defragmented; defragmenting the files to be defragmented corresponding to the file inode numbers in the storage device, according to the updated file list, in response to the occurrence of a triggering event.

Abstract: The invention provides a method for constructing a target prediction model in a multicenter small sample scenario and a prediction method. In combination with the idea of transfer learning, a training set of a new node is predicted by directly using knowledge of a trained node, and a prediction error sample is used to reflect a difference between the new node and the trained node. The difference is used as supplementary knowledge. Model knowledge of the new node is quickly acquired, to avoid training of a new node from scratch each time. Finally, parallel integration of incremental subclassifiers is implemented by using a ridge regression method, so that the deployment time and costs are greatly reduced. The generalization of models is ensured through sharing of historical knowledge and a knowledge discarding mechanism, good classification effect can also be achieved for a node with a small sample size.

Abstract: A computing device and a computing system are provided. The computing device comprises: a plurality of computing modules; and serial communication paths between/among the plurality of computing modules. Each computing module comprises: an internal circuit for performing an operation on a signal received from a corresponding serial communication path; and an extension circuit for receiving a signal from the internal circuit as an input signal. The extension circuit comprises: a delay module for delaying the input signal, the delay module comprising one or more delay units; one or more extension select modules for selectively performing a level extension on the input signal through the signal delayed by corresponding one or more delay units to generate one or more respective level-extended signals; and an output module for outputting one or more of the one or more level-extended signals.

Abstract: Aspects described herein relate to transmitting an indication of a capability to support concurrent or time division multiplexed uplink transmission of at least one of multiple uplink channels across multiple component carriers. In addition, a configuration to transmit over the at least one of the multiple uplink channels over the multiple component carriers using concurrent transmission or time division multiplexed transmission can be received based on the indication.

Abstract: Various aspects of the present disclosure generally relate to wireless communication. In some aspects, a user equipment (UE) may determine a starting symbol of a radio frequency (RF) retuning gap between a first plurality of repetitions of an uplink transmission in a first frequency hop and a second plurality of repetitions of the uplink transmission in a second frequency hop. The starting symbol of the RF retuning gap is based at least in part on a starting symbol of the second plurality of repetitions and a quantity of symbols included in the RF retuning gap. The UE may determine an ending symbol of the RF retuning gap based at least in part on the starting symbol of the second plurality of repetitions. The UE may perform RF retuning during the RF retuning gap. Numerous other aspects are provided.

Abstract: A motor, a compressor, and a refrigeration device are provided. The motor includes a stator assembly and a rotor assembly. The stator assembly includes a stator core provided with a stator slot. The rotor assembly includes a rotor core and a permanent magnet. The stator core is sleeved outside of the rotor core or vice versa. The permanent magnet is arranged on the rotor core. By limiting the relationship between the distance between the stator core and the rotor core, the length of the permanent magnet in its own magnetization direction, the number of the stator slots, and the intrinsic coercivity of the permanent magnet, it is possible to adjust the strength of the demagnetization reverse magnetic field generated by energizing the motor.

Abstract: This application relates to a frame and an electronic device. The frame includes a first body and a second body. The first body includes a sealing surface, and the second body includes a flow-blocking part sunken toward one side. An extension surface of the sealing surface intersects the flow blocking part in a first direction. Because the flow blocking part is designed, in a plastic injection process, a pressure of injected plastic on the first body is reduced, to ensure that, after sealing processing is performed on the sealing surface of the first body, the plastic does not continuously flow in the first direction. This ensures sealing reliability of the sealing surface, improves sealing effect of the first body, enables the first body and the second body to be bound well, and ensures smooth manufacturing of the frame.

Abstract: A preparation method of a modified bismaleimide resin includes the following steps. Maleic anhydride is mixed with a bisamine compound to obtain a corresponding mixture. The mixture is heated. The bisamine compound has a main chain structure or fragment represented by Formula A, Formula B, or Formula C as described in the description.

Abstract: A projection device includes multiple optical fiber mounting mechanisms. Each of the optical fiber mounting mechanisms includes an optical fiber extending along a first direction as an axis direction, a signal circuit extending along the first direction, and a mounting structure. The optical fiber includes an engaging section. The mounting structure surrounds the engaging section of the optical fiber and the signal circuit. The mounting structure includes an installation portion extending radially relative to the first direction. The installation portion includes a surface and multiple elements exposed from the surface. The surface includes a normal direction parallel with the first direction. A first length of the engaging section of the optical fiber protruding from the surface of the installation portion is longer than lengths of the elements protruding from the surface.

Abstract: An optimization method for data flow includes converting a plurality of extract-transform-load (ETL) scripts into a data flow diagram, wherein the data flow diagram includes a plurality of data lineages, with each data lineage including a plurality of data nodes; performing a first merging process, which repeatedly identifies at least two data lineages with the same nodes and merges them into a data family until any two of the data lineages do not have the same nodes; performing a second merging process, which calculates a column similarity of a pair of node, wherein the pair of nodes includes two adjacent data nodes within a data lineage; when the column similarity exceeds a threshold, merging the two data nodes; when the column similarity of the pair of nodes falls below the threshold, the data flow diagram is optimized, which is then converted into ETL scripts accordingly.

Abstract: An image capture device is disclosed that includes a body and a door assembly that is configured for removable connection to the body. The door assembly includes a door body; a locking mechanism that is slidable in relation to the door body between a locked position and an unlocked position; and at least one biasing member that is configured for engagement (contact) with the door body and the locking mechanism to automatically move the locking mechanism into the locked position upon closure of the door assembly.

Abstract: Certain aspects of the present disclosure provide techniques for using subband precoding for uplink transmissions. In some cases, a UE may transmit sounding reference signals (SRS) to a network entity, receive information indicating at least one of a set of one or more frequency domain (FD) bases and linear combination coefficients, determined based on the SRS transmission, determine subband precoding based at least in part on linear combinations of the FD bases based on the linear combination coefficients, and transmit a physical uplink shared channel (PUSCH) with the subband precoding.

Abstract: An operating method of a fully homomorphic encrypted neural network model is provided, wherein the fully homomorphic encrypted neural network model includes a plurality of layers, and the method performed by a processor includes: for one of the plurality of layers, encrypting a plaintext input with a first encryption algorithm to generate a ciphertext vector, performing a convolution operation according to the ciphertext vector to generate a result vector, transforming the result vector into a plurality of result ciphertexts adopting a second encryption algorithm, inputting the plurality of result ciphertexts into an activation function to generate a plurality of encrypted activation values, and repacking the plurality of encrypted activation values to generate an output vector adopting the first encryption algorithm.

Abstract: Provided is a fusion protein containing GLP-1 and FGF21 linked through an immunoglobulin Fc portion. Each of the GLP-1 and FGF21 can be substituted with its analogues or variants and, the Fc portion, which can be derived from IgG4, can include substitutions that further enhance the fusion protein's stability and activity. These fragments can be linked together through peptide linkers. The fusion protein can be used clinically to reduce glucose and lipid levels and body weight.

Abstract: A manufacturing method of a memory device may be applied to a three-dimensional NAND memory device with high capacity and high performance. In a manufacturing process of the three-dimensional NAND memory device, a material of a control gate (word line) is tungsten. The forming method of a tungsten layer includes nucleation and bulk formation performed. In at least one of the nucleation and the bulk formation, hydrogen flow is between 1000 and 20000 sccm. At least one time of soak with nitrogen may also be performed after the nucleation. A tungsten grain size in the tungsten layer is 70 nm or more.

Abstract: This invention discloses a method for comprehensively apportioning the source contribution to fine particular matter (PM2.5) based on the receptor model and the chemical transport model, which comprises the receptor model calculation steps, chemical transport model calculation steps and comprehensive source apportionment steps; the said comprehensive source apportionment step comprises the following sub-steps: according to the principle of inverse proportionality between uncertainty and weight coefficient, the first (receptor model) uncertainty and the second (chemical transport model) uncertainty are normalized to obtain their respective weight coefficients; the comprehensive source apportionment results are calculated based on the apportionment results of the receptor model and the chemical transport model, and their respective weight coefficient.