I think what I wrote is correct, but why is it giving this error?
Don’t use quiet
. Then I think you’ll see that one of the iterations did not find an optimal solution, so L_new
was populated with NaN
, so in the next iteration, L
and therefore L_pre
was NaN
, resulting in CVX considering expressions having NaN
in the input data to be invalid
.
You can look at posts in Search results for 'sca unreliable order:latest' - CVX Forum: a community-driven support forum to see the hazards of crude SCA.