Not yet Degenerate meaning

[danieloliveira56]

Hello,

I'm trying to understand the meaning of the Nhj info string better.

I noticed Nhj always show up at iteration 2 when the problem behaves well and factorization suceeds.

Can we be sure this will always happen? Is there any meaning to it?

Checking the PDPerturbationHandler::finalize_test() function, what I think is happening is that iteration 1 starts with

• test_status_ == NO_TEST
• jac_degenerate_ == NOT_YET_DETERMINED
• hess_degenerate_ == NOT_YET_DETERMINED,

so when finalize_test() is called by PDPerturbationHandler::ConsiderNewSystem, the function just returns, without changing any variable or setting any info string. But at the end of iteration 1, test_status_ will have changed.

So at the start of iteration 2, we have:

• test_status_ == TEST_DELTA_C_EQ_0_DELTA_X_EQ_0,
• jac_degenerate_ == NOT_YET_DETERMINED,
• hess_degenerate_ == NOT_YET_DETERMINED,

and finalize_test() will set both jac_degenerate_ and hess_degenerate_ to NOT_DEGENERATE.

So any subsequent call of finalize_test() will not result in any info string, unless jac_degenerate_ or hess_degenerate_ are reset to NOT_YET_DETERMINED.

So maybe Nhj at the end of iteration 2 is a "false alarm"?

[svigerske]

Maybe Nhj is not meant to be alarming. Not being degenerate should be a good thing.

If Ipopt calls PerturbForSingularity() or PerturbForWrongInertia(), then I would think you can get other info strings in finalize_test()