<html><body><div style="color:#000; background-color:#fff; font-family:times new roman, new york, times, serif;font-size:10pt"><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span>Hi,</span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span><br></span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span>I've been obtaining nan values with my code and I'm trying to understand them. I used <span style="font-style: italic;">cout</span> to check the values of my parameters as they were changed through each function evaluation. Here's the commented output:</span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span><br></span></div><blockquote style="margin: 0px 0px 0px 40px; border: none; padding: 0px; "><div><span><div><span style="font-size: 13px;">---------------
EVALUATION 1 ----------------</span></div><div><span style="font-size: 13px;">The value of SM1 is: 2.4188</span></div><div><span style="font-size: 13px;">The value of SM2 is: -0.0026</span></div><div><span style="font-size: 13px;">The value of S1 is: -0.3135</span></div><div><span style="font-size: 13px;">The value of S2 is: -0.0074</span></div><div><span style="font-size: 13px;">The value of S3 is: 0.5715</span></div><div><span style="font-size: 13px;">The value of S4 is: 0.0005</span></div><div><span style="font-size: 13px;">The value of GS is: 0.6556</span></div><div><span style="font-size: 13px;">The value of G1 is: 0.2917</span></div><div><span style="font-size: 13px;">The value of G2 is: 0.0052</span></div><div><span style="font-size: 13px;">The value of G3 is: 0.9913</span></div><div><span style="font-size: 13px;">The value of G4 is: -0.0007</span></div><div><span style="font-size: 13px;">The value of NS1 is: 11.6095</span></div><div><span
style="font-size: 13px;">The value of NS2 is: -0.2073</span></div><div><span style="font-size: 13px;">The value of NS3 is: -0.78</span></div><div><span style="font-size: 13px;">The value of NS4 is: 0.0147</span></div><div><span style="font-size: 13px;">The value of SSS is: 1.0408</span></div><div><span style="font-size: 13px;">The value of SS1 is: 2.7105</span></div><div><span style="font-size: 13px;">The value of SS2 is: -0.0072</span></div><div><span style="font-size: 13px;">The value of ll is: 35661.1</span><br></div><div><span style="font-size: 13px;"><br></span></div><div><span><div><span style="font-size: 13px;">Initial statistics: 18 variables; iteration 0; function evaluation 0</span></div><div><span style="font-size: 13px;">Function value 3.5661063e+04; maximum gradient component mag -2.9995e+04</span></div><div><span style="font-size: 13px;">Var Value Gradient |Var Value Gradient
|Var Value Gradient </span></div><div><span style="font-size: 13px;"> 1 -0.29118 -7.65878e+03 | 2 -0.16745 -1.94756e+04 | 3 0.22306 -3.82115e+03</span></div><div><span style="font-size: 13px;"> 4 -0.24128 -6.19649e+03 | 5 0.03043 -5.81076e+03 | 6 0.10660 -6.79017e+03</span></div><div><span style="font-size: 13px;"> 7 0.19681 -1.26020e+03 | 8 -0.27356 -5.74639e+03 | 9 0.34814 -1.49102e+04</span></div><div><span style="font-size: 13px;"> 10 -0.15867 -1.89236e+03 | 11 -0.49363 -1.26001e+04 | 12 0.75866 -2.17818e+03</span></div><div><span style="font-size: 13px;"> 13 -0.48566 -2.64172e+04 | 14 -0.24226 -4.22884e+03 | 15 0.40017 -2.99951e+04</span></div><div><span style="font-size: 13px;"> 16 0.88007 -8.13183e+01 | 17 0.50306 7.74543e+01 | 18 -0.51172 6.60299e+01</span></div><div
style="font-size: 13px; "><br></div></span></div></span></div></blockquote><span style="font-size: 13px;">Look that I'm starting with a good negative log-likelihood value: 35661.1.</span><div><span style="font-size: 13px;"><br></span></div><div><span style="font-size: 13px;">After 10 evaluations, ADMB reaches a value of 19672:<br></span><div><span style="font-size: 13px;"><br></span></div><blockquote style="margin: 0 0 0 40px; border: none; padding: 0px;"><div><span><div><span style="font-size: 13px;">--------------- EVALUATION 10 ----------------</span></div><div><span style="font-size: 13px;">The value of SM1 is: 2.43238</span></div><div><span style="font-size: 13px;">The value of SM2 is: -0.00210362</span></div><div><span style="font-size: 13px;">The value of S1 is: -0.29471</span></div><div><span style="font-size: 13px;">The value of S2 is: -0.00709708</span></div><div><span style="font-size: 13px;">The value of S3 is:
0.578347</span></div><div><span style="font-size: 13px;">The value of S4 is: 0.000552595</span></div><div><span style="font-size: 13px;">The value of GS is: 0.657158</span></div><div><span style="font-size: 13px;">The value of G1 is: 0.298574</span></div><div><span style="font-size: 13px;">The value of G2 is: 0.00552995</span></div><div><span style="font-size: 13px;">The value of G3 is: 0.991372</span></div><div><span style="font-size: 13px;">The value of G4 is: -0.000676023</span></div><div><span style="font-size: 13px;">The value of NS1 is: 11.621</span></div><div><span style="font-size: 13px;">The value of NS2 is: -0.19179</span></div><div><span style="font-size: 13px;">The value of NS3 is: -0.776388</span></div><div><span style="font-size: 13px;">The value of NS4 is: 0.0162437</span></div><div><span style="font-size: 13px;">The value of SSS is: 1.04082</span></div><div><span style="font-size: 13px;">The value of SS1 is:
2.71036</span></div><div><span style="font-size: 13px;">The value of SS2 is: -0.0072012</span></div><div><div><span style="font-size: 13px;">The value of ll is: 19672</span></div><div><span style="font-size: 13px;"><br></span></div></div></span></div></blockquote></div><div><div><span><div><div><span style="font-size: 13px;">However, on the next evaluation I get a nan:</span></div></div></span></div></div><div><blockquote style="margin: 0 0 0 40px; border: none; padding: 0px;"><div><span><div><div><span style="font-size: 13px;"><br></span></div><div><span style="font-size: 13px;">--------------- EVALUATION 11 ----------------</span></div><div><span style="font-size: 13px;">The value of SM1 is: 2.5007</span></div><div><span style="font-size: 13px;">The value of SM2 is: 0.000383817</span></div><div><span style="font-size: 13px;">The value of S1 is: -0.203675</span></div><div><span style="font-size: 13px;">The value of S2 is:
-0.00558685</span></div><div><span style="font-size: 13px;">The value of S3 is: 0.611781</span></div><div><span style="font-size: 13px;">The value of S4 is: 0.000807524</span></div><div><span style="font-size: 13px;">The value of GS is: 0.664767</span></div><div><span style="font-size: 13px;">The value of G1 is: 0.332894</span></div><div><span style="font-size: 13px;">The value of G2 is: 0.00701577</span></div><div><span style="font-size: 13px;">The value of G3 is: 0.991727</span></div><div><span style="font-size: 13px;">The value of G4 is: -0.000548353</span></div><div><span style="font-size: 13px;">The value of NS1 is: 11.6752</span></div><div><span style="font-size: 13px;">The value of NS2 is: -0.104003</span></div><div><span style="font-size: 13px;">The value of NS3 is: -0.758469</span></div><div><span style="font-size: 13px;">The value of NS4 is: 0.022198</span></div><div><span style="font-size: 13px;">The value of SSS is:
1.04092</span></div><div><span style="font-size: 13px;">The value of SS1 is: 2.70966</span></div><div><span style="font-size: 13px;">The value of SS2 is: -0.00720709</span></div><div><span style="font-size: 13px;">The value of v is: nan</span><br></div></div><div><span style="font-size: 13px;"><br></span></div></span></div></blockquote><span style="font-size: 13px;">I've been debugging my code and apparently the problem is due to a large exponential value. I'm working on solving that.</span></div><div><span style="font-size: 13px;"><br></span></div><div><span style="font-size: 13px;">However, my problem right now is that I don't understand why on the next evaluation ADMB changes the parameter values so dramatically:</span></div><div><span style="font-size: 13px;"><br></span></div><blockquote style="margin: 0 0 0 40px; border: none; padding: 0px;"><div><span><div><span style="font-size: 13px;">--------------- EVALUATION 12
----------------</span></div><div><span style="font-size: 13px;">The value of SM1 is: 3.20126</span></div><div><span style="font-size: 13px;">The value of SM2 is: 0.00564513</span></div><div><span style="font-size: 13px;">The value of S1 is: 0.567494</span></div><div><span style="font-size: 13px;">The value of S2 is: 0.00982242</span></div><div><span style="font-size: 13px;">The value of S3 is: 0.898156</span></div><div><span style="font-size: 13px;">The value of S4 is: 0.00273254</span></div><div><span style="font-size: 13px;">The value of GS is: 0.738235</span></div><div><span style="font-size: 13px;">The value of G1 is: 0.693724</span></div><div><span style="font-size: 13px;">The value of G2 is: 0.00432671</span></div><div><span style="font-size: 13px;">The value of G3 is: 0.995349</span></div><div><span style="font-size: 13px;">The value of G4 is: 0.000861996</span></div><div><span style="font-size: 13px;">The value of NS1 is:
11.9882</span></div><div><span style="font-size: 13px;">The value of NS2 is: 0.0299506</span></div><div><span style="font-size: 13px;">The value of NS3 is: -0.57073</span></div><div><span style="font-size: 13px;">The value of NS4 is: -0.0243074</span></div><div><span style="font-size: 13px;">The value of SSS is: 1.04191</span></div><div><span style="font-size: 13px;">The value of SS1 is: 2.70261</span></div><div><span style="font-size: 13px;">The value of SS2 is: -0.00726567</span></div><div><span style="font-size: 13px;">The value of ll is: 45840.4</span><br></div></span></div></blockquote><div><div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span><br></span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span style="font-size: 10pt; ">Look that now the value of the negative log-likelihood is 45840.4, a value that is worst that the one from where it started.
So basically a good starting point was pointless...</span></div></div></div><div><div><div style="font-size: 10pt; font-family: 'times new roman', 'new york', times, serif; "><span><div><br></div></span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span>After 41 evaluations, ADMB didn't move from this value:</span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span><br></span></div></div></div><blockquote style="margin: 0 0 0 40px; border: none; padding: 0px;"><div><div><div><span style="font-size: 13px;"><div> - final statistics:</div><div>18 variables; iteration 2; function evaluation 41</div><div>Function value 4.5840e+04; maximum gradient component mag 6.1581e+04</div><div>Exit code = 1; converg criter 1.0000e-04</div><div>Var Value Gradient |Var Value Gradient |Var
Value Gradient </div><div> 1 0.41100 -9.14221e+03 | 2 1.61813 1.73251e+04 | 3 0.57339 -1.19896e+02</div><div> 4 0.32683 -1.58464e+02 | 5 0.56317 -3.73742e+01 | 6 0.72914 -1.49856e+01</div><div> 7 0.31234 -3.68046e+03 | 8 0.25329 1.52480e+03 | 9 1.71514 -2.89367e+03</div><div> 10 0.01483 3.43486e+02 | 11 0.66157 1.11757e+03 | 12 0.95836 -7.59758e+02</div><div> 13 1.93634 6.15814e+04 | 14 0.14545 -9.47917e+03 | 15 3.15020 -1.58681e+04</div><div> 16 0.88752 -3.36327e+01 | 17 0.49596 6.55166e+02 | 18 -0.51777 6.42341e+02</div></span></div></div></div></blockquote><div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span><br></span></div><div style="font-family: 'times new
roman', 'new york', times, serif; font-size: 10pt; "><span><div style="font-size: medium; "><div><div style="font-size: 10pt; "><span><div style="font-size: 10pt; ">I'm running the program with the following parameter limits:</div><div><br></div></span></div></div></div><blockquote style="margin: 0px 0px 0px 40px; font-size: medium; border: none; padding: 0px; "><div><div><span style="font-size: 13px; "> init_bounded_number SM1(2.0,3.5)</span></div><div><span style="font-size: 13px; "> init_bounded_number SM2(-0.01,0.01)</span></div><div><span style="font-size: 13px; "> init_bounded_number S1(-3.0,1.0)</span></div><div><span style="font-size: 13px; "> init_bounded_number S2(-0.02,0.02)</span></div><div><span style="font-size: 13px; "> init_bounded_number S3(0.1,1)</span></div><div><span style="font-size: 13px; "> init_bounded_number S4(-0.003,0.003)</span></div><div><span style="font-size: 13px; ">
init_bounded_number GS(0.01,1.0)</span></div><div><span style="font-size: 13px; "> init_bounded_number G1(0.0,1.0)</span></div><div><span style="font-size: 13px; "> init_bounded_number G2(-0.01,0.01)</span></div><div><span style="font-size: 13px; "> init_bounded_number G3(0.98,1.01)</span></div><div><span style="font-size: 13px; "> init_bounded_number G4(-0.001,0.001)</span></div><div><span style="font-size: 13px; "> init_bounded_number NS1(1,12)</span></div><div><span style="font-size: 13px; "> init_bounded_number NS2(-0.3,0.3)</span></div><div><span style="font-size: 13px; "> init_bounded_number NS3(-1,-0.3)</span></div><div><span style="font-size: 13px; "> init_bounded_number NS4(-0.025,0.025)<span class="Apple-tab-span" style="white-space: pre; "> </span></span></div><div><span style="font-size: 13px; "> init_bounded_number SSS(0.01,1.05)</span></div><div><span style="font-size: 13px; ">
init_bounded_number SS1(1.0,3.0)</span></div><div><span style="font-size: 13px; "> init_bounded_number SS2(-0.01,0.01)</span></div></div></blockquote></span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><span><br></span></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; ">Does anyone know the kind of rules ADMB follows when faced with a nan?</div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "><br></div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; ">Thanks in advance,</div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; "> </div><div style="font-family: 'times new roman', 'new york', times, serif; font-size: 10pt; ">Edgar J. González<br></div></div></div></body></html>