In
|
Jd .= Jz * d #lsmr doesn't return this information |
We are computing J * d, while I am suspecting this could be provided by Krylov's method.
Additionally, we are also computing the norm of Jd and d afterwards
|
Pred = T(0.5) * (normcz^2 - norm(Jd)^2) # T(0.5) * (normcz^2 - norm(Jd + cz)^2) |
In
DCISolver.jl/src/dci_feasibility.jl
Line 311 in d45e79f
We are computing
J * d, while I am suspecting this could be provided by Krylov's method.Additionally, we are also computing the norm of
JdanddafterwardsDCISolver.jl/src/dci_feasibility.jl
Line 76 in d45e79f