Processing sparse linear systems using distributed resources

Patent metadata

Field	Value
Publication number	US-2024320033-A1
Application number	US-202318187172-A
Country	US
Kind code	A1
Filing date	Mar 21, 2023
Priority date	Mar 21, 2023
Publication date	Sep 26, 2024
Grant date	—

Abstract

Solving linear systems by sending matrix data from a first computer to a second computer, directing the second computer in determining a solution to a parallel computing task for the matrix data, receiving the solution by the first computer, determining a solution to a non-parallel computing task for the matrix data using the first computer, and providing the solution to the non-parallel computing task.

First claim

What is claimed is:

1 . A method for solving linear systems, the method comprising:
sending, by one or more computer processors, matrix data from a first computer to a second computer; directing, by the one or more computer processors, the second computer in determining a solution to a parallel computing task for the matrix data using the second computer; receiving, by the one or more computer processors, the solution by the first computer; determining, by the one or more computer processors, a solution to a non-parallel computing task for the matrix data using the first computer; and providing, by the one or more computer processors, the solution to the non-parallel computing task.

2 . The computer implemented method according to claim 1 , further comprising masking, by the one or more computer processors, the matrix data; and sending the masked matrix data to the second computer.

3 . The computer implemented method according to claim 2 , wherein the masking comprises applying a linear scaling to the matrix data.

4 . The computer implemented method according to claim 2 , wherein the masking comprises scaling the matrix data using a matrix mean non-zero entry.

5 . The computer implemented method according to claim 2 , wherein the masking comprises applying a unitary transformation to the matrix data.

6 . The computer implemented method according to claim 1 , further comprising encrypting, by the one or more computer processors, the matrix data.

7 . The computer implemented method according to claim 1 , further comprising providing, by the one or more computer processors, the solution to the non-parallel problem to the second computer; and receiving, by the one or more computer processors, a solution based upon the solution to the non-parallel problem from the second computer.

8 . A computer program product for solving linear systems, the computer program product comprising one or more computer readable storage media and collectively stored program instructions on the one or more computer readable storage media, the stored program instructions which, when executed, cause one or more computer systems to:
send matrix data from a first computer to a second computer; direct the second computer to determine a solution to a parallel computing task for the matrix data using the second computer; receive the solution to the first computer; determine a solution to a non-parallel computing task for the matrix data using the first computer; and provide the solution to the non-parallel computing task.

9 . The computer program product according to claim 8 , the stored program instruction further comprising program instructions which, when executed cause the one or more computer systems to mask the matrix data; and program instructions to send the masked matrix data to the second computer.

10 . The computer program product according to claim 9 , wherein the masking comprises applying a linear scaling to the matrix data.

11 . The computer program product according to claim 9 , wherein the masking comprises scaling the matrix data using a matrix mean non-zero entry.

12 . The computer program product according to claim 9 , wherein the masking comprises applying a unitary transformation to the matrix data.

13 . The computer program product according to claim 8 , the stored program instructions further comprising program instructions which, when executed cause the one or more computer systems to encrypt the matrix data.

14 . The computer program product according to claim 8 , the stored program instructions further comprising program instructions which, when executed cause the one or more computer systems to provide the solution to the non-parallel problem to the second computer; and program instructions to receive a solution based upon the solution to the non-parallel problem from the second computer.

15 . A computer system for solving linear systems, the computer system comprising:
one or more computer processors; one or more computer readable storage devices; and stored program instructions on the one or more computer readable storage devices for execution by the one or more computer processors, the stored program instructions which, when executed, cause the one or more computer processors to:
send matrix data from a first computer to a second computer;
direct the second computer to determine a solution to a parallel computing task for the matrix data using the second computer;
receive the solution to the first computer;
determine a solution to a non-parallel computing task for the matrix data using the first computer; and
provide the solution to the non-parallel computing task.

16 . The computer system according to claim 15 , the stored program instruction further comprising program instructions which, when executed cause the one or more computer processors to mask the matrix data; and program instructions to send the masked matrix data to the second computer.

17 . The computer system according to claim 16 , wherein the masking comprises applying a linear scaling to the matrix data.

18 . The computer system according to claim 16 , wherein the masking comprises scaling the matrix data using a matrix mean non-zero entry.

19 . The computer system according to claim 16 , wherein the masking comprises applying a unitary transformation to the matrix data.

20 . The computer system according to claim 15 , the stored program instructions further comprising program instructions which, when executed cause the one or more computer processors to encrypt the matrix data.

Assignees

• IBM

Inventors

• Horesh Lior
• Kalantzis Vasileios
• Ubaru Shashanka

Classifications

• G06F2209/509

  Offload · CPC title

• G06F9/5066

  Algorithms for mapping a plurality of inter-dependent sub-tasks onto a plurality of physical CPUs (mappping at compile time, see G06F8/451) · CPC title

• G06F9/5072

  Grid computing · CPC title

• G06F17/16

  Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)} · CPC title

• G06F9/3555

  using scaling, e.g. multiplication of index · CPC title