Test what happens when the starting point is bad

[zaikunzhang]

Use the following code to generate the starting point, and test the performance of different algorithms.

r = randn(n,1);
r = r / norm(r);
x0 = p.x0 + 10 * max(1, norm(x0)) * r;

The same as tests with noise, make sure the following.

1. Conduct a few random experiments and take the average of the performance profiles.
2. For each random experiment, different solvers receive the same starting point.

[Lht97]

I have tested locally, and I will close this issue until it works on GitHub actions.

[zaikunzhang]

I have tested locally, and I will close this issue until it works on GitHub actions.

Your answer should be "I have tested it. I observed the following. ...."

Your current answer does not contain any information.