Re: [Gretl-devel] NLS settings inside arima

Thank you! That's why I asked With my "knowledge" of C it coulld have taken hours to extract it from code Dear Aliin, I think, for starting values, something like 10^-5 is enough NLS must not converge at the same point as MLE: that's why attaining true convergence for NLS for obtaining initial values is aimless exercise May I ask: why Hannan Rissanen method used only q+Q >=2 For q=1 even 1-st IGLS iteration I applied gives much better results for arma(0,1), arma(1,1) for arma(2,1) the gain is moderate It seems with growing p several IGLS iterations needed , or Hannan Rissanen Also if tрis is not convincing I even do not know what to do next wooldridge       Total     no-conv1     no-conv2       34284            0           94 stock_watson_2       Total     no-conv1     no-conv2        1992            0            0 stock_watson       Total     no-conv1     no-conv2        4800            0           21 koop       Total     no-conv1     no-conv2        5976            0           11 ETM       Total     no-conv1     no-conv2       10800            0           30 dougherty       Total     no-conv1     no-conv2        3168            0           18 ashenfelter       Total     no-conv1     no-conv2         504            0            0 gujarati       Total     no-conv1     no-conv2       17124            0          107 POE4       Total     no-conv1     no-conv2       23112            0           30 verbeek       Total     no-conv1     no-conv2       23160            0           12 gretl/data/(Ramanathan)       Total     no-conv1     no-conv2       45024            0          157 gretl/data/greene       Total     no-conv1     no-conv2        9576            0           42 gretl/data/misc       Total     no-conv1     no-conv2       98364            0          437 TOTAL  Total     no-conv1     no-conv2      277884            0          959   Oleh 24 листопада 2018, 23:05:58, від "Allin Cottrell" <cottrell(a)wfu.edu&gt;: On Sat, 24 Nov 2018, oleg_komashko(a)ukr.net wrote: We are in fact setting a sloppy tolerance when we use NLS in arma_init, since quite recently -- see arma_init.c, line 1401: libset_set_double(NLS_TOLER, 1.0e-5); That can be compared with the default, namely pow(DBL_EPSILON, .75), which is about 1.82e-12. Perhaps it should be sloppier still. But gretl is open source, no need to guess what we're doing or not doing! Allin