MUMBAI, India, Jan. 9 -- Intellectual Property India has published a patent application (202541133671 A) filed by Indian Institute Of Technology, Hyderabad, Telangana, on Dec. 30, 2025, for 'hardware system and method for computing svd using two-sided jacobi-based iterative decomposition.'
Inventor(s) include Anjan Banerjee; Jonnala Sai Maneesh Kumar; and Amit Acharyya.
The application for the patent was published on Jan. 9, under issue no. 02/2026.
According to the abstract released by the Intellectual Property India: "The present invention discloses a hardware system and method for computing the Singular Value Decomposition (SVD) of real-valued and complex-valued matrices using a two-sided Jacobi-based iterative decomposition technique implemented entirely within programmable or reconfigurable logic. The system includes an input and configuration interface, an angle calculation subsystem for computing left and right Jacobi rotation angles independently of matrix rotation, an angle storage subsystem for reusing computed rotation angles across multiple matrix updates, and a unified rotation engine for applying orthogonal transformations to update the input matrix and accumulate transformations in the left and right singular vector matrices. A deterministic, hardware-controlled permutation-based sweep scheduling strategy executes a predefined number of Jacobi sweeps without participation of a general-purpose processor during sweep execution. The invention supports complex-valued matrices by transforming each complex element into an equivalent real-valued structured representation compatible with two-sided Jacobi rotations."
Disclaimer: Curated by HT Syndication.
