Hi Sergei, hope all is well. just have 2 quick ?

1. i've noticed i don't get any gain in speed when starting the optimisation of an alternate (unconstrained) model at the optimised parameter values for the null model. i would've expected this? is this perhaps unique to my problem, or is the optimise function hard-coded to do at least x cycles?

2. does ClearConstraints work with variables? i've seen it work with cat vars, but can't seem to get it to work with standard variables.
eg:

global c = Random(0,1);
c :< 1;
ClearConstraints(c);

does not remove the constraint c:<1?

thanks. ./w

thanks.

1. i did this manually, storing parameter estimates and then setting starting values to those stored in array. ie.

global c = somearray [ i ]; 

does hyphy still guess in this case. should i rather set USE_LAST_RESULTS=1? this probably won't work since null and alt are being optimised on different nodes and thus there are no last results per se?

2. mmm. so i had

global c:<1;

to constrain less than 1. i reset manually with

global c:<1000;

thus allowing an upper limit which includes values > 1. however i get some strange optimisations where the parameter estimates are exactly 1000. almost as if there is a fatal attraction to the upper limit? i seem to have solved this by changing the parameter names to c and c_alt for the two models, thus removing the need to remove constraints. although it makes the code a little less elegant. cheers ./w