Fix cg of admm: warm start, using inversion for non convolution case

benchopt / benchmark_tv_1d

TV Denoising in 1D

2 stars 7 forks source link

Fix cg of admm: warm start, using inversion for non convolution case #29

Closed EnLAI111 closed 2 years ago

EnLAI111 commented 2 years ago

For non-convolution case, when we set tol=1e-12 for cg, ADMM can converge. But in this case, it is faster to use the inversion of matrix and get the exact solution, instead of solving an equation.

EnLAI111 commented 2 years ago

Could I merge this one？