# US 11,443,171 B2 · Pulse generation for updating crossbar arrays > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,443,171 B2 - Document ID: US11443171B2 - Status: Granted - Kind: B2 - Application number: 16/929,168 - Filed: 2020-07-15 - Published: 2022-01-20 - Granted: 2022-09-13 - Earliest claimed: 2020-07-15 - Assignees: International Business Machines Corporation - Applicants: International Business Machines Corporation - Inventors: Seyoung Kim; Oguzhan Murat Onen; Tayfun Gokmen; Malte Johannes Rasch - Web page: https://patentgrep.com/patent/US11443171B2 ## CPC - G06F 7/5443: {Sum of products (for applications thereof, see the relevant places, e.g. G06F17/10, H03H17/00)} - G06N 3/049: Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs - G06F 9/3001: {Arithmetic instructions} - G06F 9/30036: {Instructions to perform operations on packed data, e.g. vector, tile or matrix operations} - G06N 3/065: Analogue means - G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections - G06F 2207/4802: Special implementations - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Provided are embodiments for a computer-implemented method, a system, and a computer program product for updating an analog crossbar array. Embodiment include receiving a number used in matrix multiplication to represent using pulse generation for a crossbar array, and receiving a bit-length to represent the number. Embodiments also include selecting pulse positions in a pulse sequence having the bit length to represent the number, performing a computation using the selected pulse positions in the pulse sequence, and updating the crossbar array using the computation. ## Claims ### Claim 1 A computer-implemented method for pulse generation for updating analog crossbar arrays, the computer-implemented method comprising: receiving, by a processor, a number used in matrix multiplication to represent using pulse generation for a crossbar array; receiving, by the processor, a bit-length to represent the number; selecting, by the processor, pulse positions in a pulse sequence having the bit-length to represent the number; performing, by the processor, a computation using the selected pulse positions in the pulse sequence; and updating, by the processor, the crossbar array using the computation; wherein the selected pulse positions comprise at least one of a first pulse sequence or a second pulse sequence; and wherein there is a pulse coincidence between the first pulse sequence and the second pulse sequence. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein the computation comprises performing an outer product operation with the number used in the matrix multiplication and a second number used in the matrix multiplication. ### Claim 3 (depends on 1) The computer-implemented method of claim 1, wherein the selected pulse positions in the pulse sequence are based at least in part on the bit-length. ### Claim 4 (depends on 1) The computer-implemented method of claim 1, wherein the bit-length is equal to a number of different elements for a weight matrix. ### Claim 5 (depends on 1) The computer-implemented method of claim 1 further comprising generating a heat map, wherein the generated heat map is based at least in part on the pulse coincidence between the first pulse sequence and the second pulse sequence. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein the at least one first pulse sequence corresponds to the number and the second pulse sequence corresponds to the second number. ### Claim 7 (depends on 1) The computer-implemented method of claim 1 further comprising comparing a true multiplication map to the pulse coincidence using the selected pulse positions, wherein the true multiplication map performs an element-wise multiplication for each element. ### Claim 8 (depends on 7) The computer-implemented method of claim 7 further comprising: determining an error between the true multiplication map and the pulse coincidence; and displaying the error between the true multiplication map and the pulse coincidence. ### Claim 9 The computer-implemented method 1, wherein updating the crossbar array comprises updating a conductance value of one or more memristive devices of the crossbar array based at least in part on the computation. ### Claim 10 A system for implementing pulse generation for updating crossbar arrays, the system comprising: a crossbar array comprising one or more memristive devices; a processor configured to: receive a number used in matrix multiplication to represent using pulse generation for the crossbar array; receive a bit-length to represent the number; select pulse positions in a pulse sequence having the bit-length to represent the number; perform a computation using the selected pulse positions in the pulse sequence; and update the crossbar array using the computation, wherein updating the crossbar array changes weights of the one or more memristive devices; wherein the selected pulse positions comprise at least one of a first pulse sequence or a second pulse sequence; and wherein there is a pulse coincidence between the first pulse sequence and the second pulse sequence. ### Claim 11 (depends on 10) The system of claim 10, wherein the computation comprises performing an outer product operation with the number used in the matrix multiplication and a second number used in the matrix multiplication. ### Claim 12 (depends on 10) The system of claim 10, wherein the selected pulse positions in the pulse sequence are based at least in part on the bit-length. ### Claim 13 (depends on 10) The system of claim 10, wherein the bit-length is equal to a number of different elements for a weight matrix. ### Claim 14 (depends on 10) The system of claim 10, wherein the computation comprises determining the outer product of a first vector and a second vector for the neural network. ### Claim 15 (depends on 10) The system of claim 10, wherein the processor is further configured to generate a heat map, wherein the generated heat map is based at least in part on the pulse coincidence between the first pulse sequence and the second pulse sequence. ### Claim 16 (depends on 10) The system of claim 10 further comprising comparing a true multiplication map to the pulse coincidence using the selected pulse positions, wherein the true multiplication map performs an element-wise multiplication for each element. ### Claim 17 (depends on 16) The system of claim 16, wherein the process is further configured to: determine an error between the true multiplication map and the pulse coincidence; and display the error between the true multiplication map and the pulse coincidence. ### Claim 18 (depends on 10) The system of claim 10, wherein updating the crossbar array comprises updating a conductance value of the one or more memristive devices of the crossbar array based at least in part on the computation. ### Claim 19 A computer program product for pulse generation for updating analog crossbar arrays, the computer program product comprising: a computer readable storage medium having stored thereon first program instructions executable by a processor to cause the processor to: receive a number used in matrix multiplication to represent using pulse generation for a crossbar array; receive a bit-length to represent the number; select pulse positions in a pulse sequence having the bit-length to represent the number; perform a computation using the selected pulse positions in the pulse sequence; and update the crossbar array using the computation, wherein updating the crossbar array comprises updating a conductance value of one or more memristive devices of the crossbar array based at least in part on the computation; wherein the selected pulse positions comprise at least one of a first pulse sequence or a second pulse sequence; and wherein there is a pulse coincidence between the first pulse sequence and the second pulse sequence generating a heat map. ### Claim 20 (depends on 19) The computer program product of claim 19, wherein the computation comprises performing an outer product operation with the number used in the matrix multiplication and a second number used in the matrix multiplication. ### Claim 21 (depends on 19) The computer program product of claim 19, wherein the bit-length is equal to a number of different elements for a weight matrix. ### Claim 22 (depends on 19) The computer program product of claim 19, wherein the instructions are further executable by a processor to cause the processor to determine the outer product of a first pulse sequence and a second pulse sequence for the neural network. ### Claim 23 (depends on 19) The computer program product of claim 19, wherein the instructions are further executable by a processor to cause the processor to generate a heat map, wherein the generated heat map is based at least in part on the pulse coincidence between the first pulse sequence and the second pulse sequence. ### Claim 24 (depends on 19) The computer program product of claim 19, wherein the instructions are further executable by a processor to cause the processor to compare a true multiplication map to the pulse coincidence using the selected pulse positions, wherein the true multiplication map performs an element-wise multiplication for each element. ### Claim 25 (depends on 19) The computer program product of claim 19, wherein the instructions are further executable by a processor to cause the processor to: determine an error between the true multiplication map and the pulse coincidence; and display the error between the true multiplication map and the pulse coincidence.