Hi, I get confused for size of vectors, parameters in dogleg function, and size of tauall for this problem.
For size of vectors, why rhoall, dlall, and tauall are vectors of size (it+1)1/2? I think we can only get values for such three variables only if there is an iteration. Hence, I suppose that these variables are of size it1/2.
For dogleg function, since step p only depends on gfx, hfx, and Delta , why do we need parameters fun, x, and fx?
For variable tauall, I wonder what is the dimension of tauall. At first, tauall is a variable of size (it+1)*2; but later, it seems that we only need to store one value for tauall in each iteration(dummy value or larger of the roots).
So, should there be a typo in the question? Or simply I did this question in a wrong way.
The questions are well-justified.
Indeed rhoall, dlall, and tauall are defined only when
there is an iteration.
But to make my output simpler, I appended at the end a dummy value,
but I understand that this may be confusing.
So I updated the relevant description in A3.
The dogleg function does not use fun, x, and fx.
I had a different implementation where dogleg was calling fun
and passing values to the trust region, but I do not recommend it,
so I updated A3 in that part too.
For tauall, indeed at most one such value is needed in the algorithm,
the one which is between 1 and 2, if there is one,
(and the larger between 1 and 2, if both between 1 and 2),
but I ask that you output both.
The reason is debugging and marking. I did not change A3 in that part.
Also, for the dummy cases, set both taus to the same dummy value.