ADMB-help,<br><br>I have noticed that estimating selection for the last age group (I mean for the last age group from which selection is fixed=1.0)<br>in an age-based model becomes too large.<br>To get around this I have also fixed it to 1.0, f.e.,<br>
<br>PARAMETER_SECTION<br> init_bounded_dev_vector log_sel1_coff(1,nages-2,-15.,15.,2) // nages-2 = 10 ,, age-12<br> init_bounded_dev_vector log_sel2_coff(1,nages-2,-15.,15.,2) <br> init_bounded_dev_vector log_sel3_coff(1,nages-2,-15.,15.,2)<br>
...............<br><br>for (i=1;i<=nyrs;i++)<br> { <br> if (i<=16)<br> { for (j=1;j<nages-2;j++) <br> {<br> log_sel1(j)=log_sel1_coff(j);<br> }<br> }<br> else<br> {<br>
if (i<=35)<br> { for (j=1;j<nages-2;j++) <br> {<br> log_sel2(j)=log_sel2_coff(j);<br> }<br> }<br>.................................<br><br> log_sel1(nages-2)=0;<br> log_sel1(nages-1)=0;<br>
log_sel1(nages)=0;<br><br> log_sel2(nages-2)=0;<br> log_sel2(nages-1)=0;<br> log_sel2(nages)=0;<br>..................<br><br>From ".cor" file:<br><br> index name value<br> 1 log_sel1_coff -2.2926e+00<br>
2 log_sel1_coff -1.3813e+00<br> 3 log_sel1_coff -1.0400e+00<br> 4 log_sel1_coff -1.1053e+00<br> 5 log_sel1_coff -1.1191e+00<br> 6 log_sel1_coff -1.1294e+00<br> 7 log_sel1_coff -1.0320e+00<br>
8 log_sel1_coff -8.5347e-01<br> 9 log_sel1_coff -8.0637e-01<br><b> 10 log_sel1_coff 1.0760e+01</b><br> 11 log_sel2_coff -3.2898e+00<br> 12 log_sel2_coff -1.6544e+00<br> 13 log_sel2_coff -1.0402e+00<br>
14 log_sel2_coff -7.6604e-01<br> 15 log_sel2_coff -8.2858e-01<br> 16 log_sel2_coff -9.0862e-01<br> 17 log_sel2_coff -9.5011e-01<br> 18 log_sel2_coff -9.6055e-01<br> 19 log_sel2_coff -8.8192e-01<br>
<b> 20 log_sel2_coff 1.1280e+01</b><br>..............................................<br><br>I have tried to shorten estimates for selection so that instead of estimating (nages-2) <br>I end up with (nages.-3) parameters:<br>
<br>PARAMETER_SECTION<br> init_bounded_dev_vector log_sel1_coff(1,nages-3,-15.,15.,2) <br> init_bounded_dev_vector log_sel2_coff(1,nages-3,-15.,15.,2) <br> init_bounded_dev_vector log_sel3_coff(1,nages-3,-15.,15.,2) <br>
.................................<br><br>for (i=1;i<=nyrs;i++)<br> { <br> if (i<=16)<br> { for (j=1;j<nages-2;j++) <br> {<br> log_sel1(j)=log_sel1_coff(j);<br> }<br> }<br>
else<br> {<br> if (i<=35)<br> { for (j=1;j<nages-2;j++) <br> {<br> log_sel2(j)=log_sel2_coff(j);<br> }<br>......................................<br><br> log_sel1(nages-2)=0;<br>
log_sel1(nages-1)=0;<br> log_sel1(nages)=0;<br><br> log_sel2(nages-2)=0;<br> log_sel2(nages-1)=0;<br> log_sel2(nages)=0;<br>..................<br><br>Estimates by this second approach are even more flawed:<br>
<br>index name value <br> 1 log_sel1_coff -5.3813e-01<br> 2 log_sel1_coff -5.1953e-01<br> 3 log_sel1_coff -3.2202e-01<br> 4 log_sel1_coff -2.2866e-01<br> 5 log_sel1_coff -1.1985e-01<br>
6 log_sel1_coff -1.4601e-02<br><b> 7 log_sel1_coff 2.0729e-01<br> 8 log_sel1_coff 5.7745e-01<br> 9 log_sel1_coff 9.5804e-01<br> 10 log_sel2_coff -1.2548e+00<br></b> 11 log_sel2_coff -8.8436e-01<br>
12 log_sel2_coff -3.8221e-01<br> 13 log_sel2_coff 1.0742e-01<br> <b>14 log_sel2_coff 2.8319e-01<br> 15 log_sel2_coff 3.8595e-01<br> 16 log_sel2_coff 4.4048e-01<br> 17 log_sel2_coff 5.7487e-01<br>
18 log_sel2_coff 7.2942e-01<br>....</b>.....................................................<br><br>One question is whether is right or not to let ADMB estimate this last age group (and because of its high value) and set to 1?<br>
On the other hand selection in such an old age group (10=age-12 ) is quite noisy and therefore difficult to estimate<br>(and probably not so bad to fixed it to 1.0)<br><br>Tips and comments are welcome,<br>Thanks in advance,<br>
Luis<br><br>