Data Analysis Question
Dear Fermipy helper,
I met some questions on understanding the meaning of TS changes and wonder whether you may give some advice on it.
I follow the tutorial of IC443. After step 5, gta.optimize(), if I run print_model, I got the following:
sourcename offset norm eflux index ts npred free
3FGL J0617.2+2234e 0.000 0.332 0.000196 2.22 34178.20 10883.6
3FGL J0619.4+2242 0.536 0.245 4.03e-06 2.57 47.51 280.8
3FGL J0609.3+2131 2.105 1.334 1.9e-06 3.91 44.85 193.8
3FGL J0609.2+2051c 2.524 0.937 1.82e-06 2.87 29.06 142.8
3FGL J0621.0+2514 2.804 0.346 2.26e-06 2.58 83.65 155.4
3FGL J0611.5+1957 2.931 0.361 1.37e-06 2.38 15.83 76.4
3FGL J0603.8+2155 3.166 0.575 4.74e-06 1.81 37.08 57.5
3FGL J0628.4+2429 3.214 0.976 1.59e-06 2.21 1.23 15.7
3FGL J0605.9+2039c 3.263 3.368 1.39e-05 2.36 20.80 125.4
3FGL J0601.5+2309 3.664 0.460 6.78e-07 2.48 4.13 31.1
3FGL J0603.3+2042 3.725 0.664 2.01e-06 1.50 nan 0.6
3FGL J0631.2+2019 3.954 0.000 3.49e-11 2.46 -0.00 0.0
3FGL J0620.4+2644 4.224 0.000 1.79e-09 1.65 -0.00 0.0
3FGL J0610.6+1728 5.336 11.340 4.9e-05 4.85 2.11 36.9
isodiff --- 1.110 0.0232 2.24 61.89 1047.6
galdiff --- 1.078 0.14 -0.02 54128.23 19456.1
If I free_sources((distance=1.0,pars='norm'), run gta.fit(), then I got the model:
sourcename offset norm eflux index ts npred free
3FGL J0617.2+2234e 0.000 0.330 0.000194 2.22 28589.78 10801.1 *
3FGL J0619.4+2242 0.536 0.253 4.16e-06 2.57 47.68 290.0 *
3FGL J0609.3+2131 2.105 1.334 1.9e-06 3.91 44.85 193.8
3FGL J0609.2+2051c 2.524 0.937 1.82e-06 2.87 29.06 142.8
3FGL J0621.0+2514 2.804 0.346 2.26e-06 2.58 83.65 155.4
3FGL J0611.5+1957 2.931 0.361 1.37e-06 2.38 15.83 76.4
3FGL J0603.8+2155 3.166 0.575 4.74e-06 1.81 37.08 57.5
3FGL J0628.4+2429 3.214 0.976 1.59e-06 2.21 1.23 15.7
3FGL J0605.9+2039c 3.263 3.368 1.39e-05 2.36 20.80 125.4
3FGL J0601.5+2309 3.664 0.460 6.78e-07 2.48 4.13 31.1
3FGL J0603.3+2042 3.725 0.664 2.01e-06 1.50 nan 0.6
3FGL J0631.2+2019 3.954 0.000 3.49e-11 2.46 -0.00 0.0
3FGL J0620.4+2644 4.224 0.000 1.79e-09 1.65 -0.00 0.0
3FGL J0610.6+1728 5.336 11.340 4.9e-05 4.85 2.11 36.9
isodiff --- 0.743 0.0156 2.24 12.32 701.8 *
galdiff --- 1.102 0.143 -0.02 48163.98 19886.8 *
Note the TS of 3FGL J0617.2+2234e changed from 34178 to 28589,
the TS of galdiff changed from 54128 to 48163, while the loglike values are similar.
I wonder which one is the better fitting, the initial optimize() or the later fit()? what caused the changes of those TS values, which should I trust? And should I take the TS seriously?
Similar TS changes happened, say, after gta.localize(), If I run gta.fit() again, the TS sometimes decrease. I am confused with which is the better fitting?
Best regards,
Jiren
Data Analysis Question
Dear Fermipy helper,
I met some questions on understanding the meaning of TS changes and wonder whether you may give some advice on it.
I follow the tutorial of IC443. After step 5, gta.optimize(), if I run print_model, I got the following:
sourcename offset norm eflux index ts npred free
3FGL J0617.2+2234e 0.000 0.332 0.000196 2.22 34178.20 10883.6
3FGL J0619.4+2242 0.536 0.245 4.03e-06 2.57 47.51 280.8
3FGL J0609.3+2131 2.105 1.334 1.9e-06 3.91 44.85 193.8
3FGL J0609.2+2051c 2.524 0.937 1.82e-06 2.87 29.06 142.8
3FGL J0621.0+2514 2.804 0.346 2.26e-06 2.58 83.65 155.4
3FGL J0611.5+1957 2.931 0.361 1.37e-06 2.38 15.83 76.4
3FGL J0603.8+2155 3.166 0.575 4.74e-06 1.81 37.08 57.5
3FGL J0628.4+2429 3.214 0.976 1.59e-06 2.21 1.23 15.7
3FGL J0605.9+2039c 3.263 3.368 1.39e-05 2.36 20.80 125.4
3FGL J0601.5+2309 3.664 0.460 6.78e-07 2.48 4.13 31.1
3FGL J0603.3+2042 3.725 0.664 2.01e-06 1.50 nan 0.6
3FGL J0631.2+2019 3.954 0.000 3.49e-11 2.46 -0.00 0.0
3FGL J0620.4+2644 4.224 0.000 1.79e-09 1.65 -0.00 0.0
3FGL J0610.6+1728 5.336 11.340 4.9e-05 4.85 2.11 36.9
isodiff --- 1.110 0.0232 2.24 61.89 1047.6
galdiff --- 1.078 0.14 -0.02 54128.23 19456.1
If I free_sources((distance=1.0,pars='norm'), run gta.fit(), then I got the model:
sourcename offset norm eflux index ts npred free
3FGL J0617.2+2234e 0.000 0.330 0.000194 2.22 28589.78 10801.1 *
3FGL J0619.4+2242 0.536 0.253 4.16e-06 2.57 47.68 290.0 *
3FGL J0609.3+2131 2.105 1.334 1.9e-06 3.91 44.85 193.8
3FGL J0609.2+2051c 2.524 0.937 1.82e-06 2.87 29.06 142.8
3FGL J0621.0+2514 2.804 0.346 2.26e-06 2.58 83.65 155.4
3FGL J0611.5+1957 2.931 0.361 1.37e-06 2.38 15.83 76.4
3FGL J0603.8+2155 3.166 0.575 4.74e-06 1.81 37.08 57.5
3FGL J0628.4+2429 3.214 0.976 1.59e-06 2.21 1.23 15.7
3FGL J0605.9+2039c 3.263 3.368 1.39e-05 2.36 20.80 125.4
3FGL J0601.5+2309 3.664 0.460 6.78e-07 2.48 4.13 31.1
3FGL J0603.3+2042 3.725 0.664 2.01e-06 1.50 nan 0.6
3FGL J0631.2+2019 3.954 0.000 3.49e-11 2.46 -0.00 0.0
3FGL J0620.4+2644 4.224 0.000 1.79e-09 1.65 -0.00 0.0
3FGL J0610.6+1728 5.336 11.340 4.9e-05 4.85 2.11 36.9
isodiff --- 0.743 0.0156 2.24 12.32 701.8 *
galdiff --- 1.102 0.143 -0.02 48163.98 19886.8 *
Note the TS of 3FGL J0617.2+2234e changed from 34178 to 28589,
the TS of galdiff changed from 54128 to 48163, while the loglike values are similar.
I wonder which one is the better fitting, the initial optimize() or the later fit()? what caused the changes of those TS values, which should I trust? And should I take the TS seriously?
Similar TS changes happened, say, after gta.localize(), If I run gta.fit() again, the TS sometimes decrease. I am confused with which is the better fitting?
Best regards,
Jiren