Displaying 1 result from an estimated 1 matches for "compare23_21te061221".
2007 Jan 10
2
problems with optim, "for"-loops and machine precision
...res[s,10]=M21$par[3]
res[s,11]=M21$par[4]
res[s,12]=M21$par[5]
res[s,13]=M21$par[6]
res[s,14]=M21$par[7]
res[s,15]=M21$value
res[s,16]=M23$value-M21$value
res[s,17]=M23$convergence
res[s,18]=M21$convergence
write.table(res,"compare23_21TE061221.txt")
rm(M23,M21)
}
}
For some strange reason the results do not correspond to what I expect:
about 10% of the solutions have a difference of NLL smaller than 0. I have
verified the optimisation of these results manually and found that a minimal
NLL was ignored and a higher NLL...