R1: For a nonsymmetric matrix We will clarify from the start that our method is designed for symmetric matrices

Nov-17-2025, 10:43:51 GMT–Neural Information Processing Systems 

R2: Will the failure cases become more or less common when matrices are large enough?