Closed moustakas closed 7 months ago
@Ragadeepika-Pucha @stephjuneau can you share a couple examples where the lack of this constraint led to some fitting issues?
@moustakas @stephjuneau
There are two different cases where the lack of constraint between forbidden lines is leading to fitting issues. I have primarily focused on comparison of [NII] and Ha widths and its dependence on measuring emission-line fluxes and broad Ha component widths. I am going to add two separate comments, explaining each of these cases.
Case 1 In some sources, the [NII] widths is quite broader than Ha and this eats up some of the existing broad Ha component, leading to a "narrower" broad Ha component, or in some cases from the narrow Ha component. As a result, both narrow and broad Ha components and of [NII] are not accurate, leading to inaccurate BPT positions and also lower black hole (BH) masses. This can be seen from the following plot (stellar mass-BH mass scaling relation), color coded by percentage difference between [NII] and Ha widths:
This plot shows that most of the "low BH mass" sources have [NII] widths >30-50% of Ha widths.
One example comparison between EmFit (my code) and Fastspecfit in [NII]+Ha spectra region is:
Example TARGETIDs: 39628005236015433
, 39628005236015433
, 39633049721244605
, 39628522964128747
Case 2
This case is the reverse, where the [NII] width is much narrower than the Ha width. In such cases, instead on one "narrow" and one "broad" component, fastspecfit is fitting two "broader" Ha components. This leads to very high Ha fluxes (>10 times of what I am measuring using my code) - clear BPT outliers
Example Fit in the [NII]+Ha region:
The fluxes mentioned are those of narrow Ha component.
Example TARGETIDs: 39627735873624156
, 39627752483065233
, 39628129861371535
, 39633127848545072
Based on the above two cases, I think it is important to have a constraint between forbidden lines and narrow Balmer lines. Maybe we can have Balmer line widths to vary within some percent of forbidden line widths. @stephjuneau can say more about it, especially regarding the physical reasoning for it.
Based on the above two cases, I think it is important to have a constraint between forbidden lines and narrow Balmer lines. Maybe we can have Balmer line widths to vary within some percent of forbidden line widths. @stephjuneau can say more about it, especially regarding the physical reasoning for it.
Thanks for posting examples for these two different scenarios, Raga! Another useful piece of information is that you had made plots comparing the widths of the narrow lines measured independently to assess what range is reasonable if we wanted to allow a model to differ and I think that lead to something like a +/- 20% range capturing a lot of the possible intrinsic variation. (Is this value correct?) Based on this, your fitting code first lets the linewidth vary and checks if the difference is within that small range. If not, then it forces the linewidths to be the same for the narrow lines. This two-step fitting may not work well for Fastspecfit but given that the variations are quite small, just fixing them to be the same is probably a good approximation. Physically it would imply that we assume that the narrow-line emitting gas has consistent kinematics, which is reasonable in most cases though maybe less so in AGN hosts. However, the failure modes that you mention also will preferentially affect AGN hosts so I'd anticipate that this fix will be a net positive!
@Ragadeepika-Pucha thanks for these specific examples. In a work-in-progress branch, I now constrain the narrow and forbidden line-widths and velocity shifts during narrow+broad-line fitting.
(During just narrow-line fitting, which impacts most objects, I haven't changed anything; the narrow Balmer+helium lines are tied separately from the forbidden lines.)
Here are your examples with this branch:
39628005236015433
- not a good fit; the code likes a highly redshift broad component but only for Hbeta; the [SII] doublet is particularly poorly fit. I may need to just let this one go...but at least the reduced chi2 is relative high, so we should be able to find these problem children.
39633049721244605
- looks pretty good!
39628522964128747
- not perfect, but probably OK
39627735873624156
- looks great
39627752483065233
- looks good!
39628129861371535
- not bad given it's a quasar...
39633127848545072
- looks great!
After consulting with @stephjuneau and @Ragadeepika-Pucha and looking at several examples together, we've opted to make a small change to the constraints when fitting the broad- + narrow-line model. The change is that the [OIII] 4959,5007 doublet is now fitted separately from the other forbidden + narrow Balmer lines in order to allow for the possibility of a blue-shifted outflow. This change should help decompose the Halpha + [NII] complex (again, only when testing the broad-line hypothesis).
This readout is too much information, but I'm pasting it for completeness (for one particular low-redshift object):
--------------------------
Broad Balmer+helium lines:
--------------------------
h6_broad_amp untied
h6_broad_sigma tied to halpha_broad_sigma with factor 1.0000
h6_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hepsilon_broad_amp untied
hepsilon_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hepsilon_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hdelta_broad_amp untied
hdelta_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hdelta_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hgamma_broad_amp untied
hgamma_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hgamma_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hei_broad_4471_amp untied
hei_broad_4471_sigma tied to halpha_broad_sigma with factor 1.0000
hei_broad_4471_vshift tied to halpha_broad_vshift with factor 1.0000
heii_broad_4686_amp untied
heii_broad_4686_sigma tied to halpha_broad_sigma with factor 1.0000
heii_broad_4686_vshift tied to halpha_broad_vshift with factor 1.0000
hbeta_broad_amp untied
hbeta_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hbeta_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hei_broad_5876_amp untied
hei_broad_5876_sigma tied to halpha_broad_sigma with factor 1.0000
hei_broad_5876_vshift tied to halpha_broad_vshift with factor 1.0000
halpha_broad_amp untied
halpha_broad_sigma untied
halpha_broad_vshift untied
---------------------------
Narrow Balmer+helium lines:
---------------------------
h6_amp untied
h6_sigma tied to nii_6584_sigma with factor 1.0000
h6_vshift tied to nii_6584_vshift with factor 1.0000
hepsilon_amp untied
hepsilon_sigma tied to nii_6584_sigma with factor 1.0000
hepsilon_vshift tied to nii_6584_vshift with factor 1.0000
hdelta_amp untied
hdelta_sigma tied to nii_6584_sigma with factor 1.0000
hdelta_vshift tied to nii_6584_vshift with factor 1.0000
hgamma_amp untied
hgamma_sigma tied to nii_6584_sigma with factor 1.0000
hgamma_vshift tied to nii_6584_vshift with factor 1.0000
hei_4471_amp untied
hei_4471_sigma tied to nii_6584_sigma with factor 1.0000
hei_4471_vshift tied to nii_6584_vshift with factor 1.0000
heii_4686_amp untied
heii_4686_sigma tied to nii_6584_sigma with factor 1.0000
heii_4686_vshift tied to nii_6584_vshift with factor 1.0000
hbeta_amp untied
hbeta_sigma tied to nii_6584_sigma with factor 1.0000
hbeta_vshift tied to nii_6584_vshift with factor 1.0000
hei_5876_amp untied
hei_5876_sigma tied to nii_6584_sigma with factor 1.0000
hei_5876_vshift tied to nii_6584_vshift with factor 1.0000
halpha_amp untied
halpha_sigma tied to nii_6584_sigma with factor 1.0000
halpha_vshift tied to nii_6584_vshift with factor 1.0000
----------------
Forbidden lines:
----------------
nev_3346_amp untied
nev_3346_sigma tied to nii_6584_sigma with factor 1.0000
nev_3346_vshift tied to nii_6584_vshift with factor 1.0000
nev_3426_amp untied
nev_3426_sigma tied to nii_6584_sigma with factor 1.0000
nev_3426_vshift tied to nii_6584_vshift with factor 1.0000
oii_3726_sigma tied to nii_6584_sigma with factor 1.0000
oii_3726_vshift tied to nii_6584_vshift with factor 1.0000
oii_3729_amp untied
oii_3729_sigma tied to nii_6584_sigma with factor 1.0000
oii_3729_vshift tied to nii_6584_vshift with factor 1.0000
neiii_3869_amp untied
neiii_3869_sigma tied to nii_6584_sigma with factor 1.0000
neiii_3869_vshift tied to nii_6584_vshift with factor 1.0000
oiii_4363_amp untied
oiii_4363_sigma tied to nii_6584_sigma with factor 1.0000
oiii_4363_vshift tied to nii_6584_vshift with factor 1.0000
oiii_4959_amp tied to oiii_5007_amp with factor 0.3351
oiii_4959_sigma tied to oiii_5007_sigma with factor 1.0000
oiii_4959_vshift tied to oiii_5007_vshift with factor 1.0000
oiii_5007_amp untied
oiii_5007_sigma untied
oiii_5007_vshift untied
nii_5755_amp untied
nii_5755_sigma tied to nii_6584_sigma with factor 1.0000
nii_5755_vshift tied to nii_6584_vshift with factor 1.0000
oi_6300_amp untied
oi_6300_sigma tied to nii_6584_sigma with factor 1.0000
oi_6300_vshift tied to nii_6584_vshift with factor 1.0000
siii_6312_amp untied
siii_6312_sigma tied to nii_6584_sigma with factor 1.0000
siii_6312_vshift tied to nii_6584_vshift with factor 1.0000
nii_6548_amp tied to nii_6584_amp with factor 0.3399
nii_6548_sigma tied to nii_6584_sigma with factor 1.0000
nii_6548_vshift tied to nii_6584_vshift with factor 1.0000
nii_6584_amp untied
nii_6584_sigma untied
nii_6584_vshift untied
sii_6716_amp untied
sii_6716_sigma tied to nii_6584_sigma with factor 1.0000
sii_6716_vshift tied to nii_6584_vshift with factor 1.0000
sii_6731_sigma tied to nii_6584_sigma with factor 1.0000
sii_6731_vshift tied to nii_6584_vshift with factor 1.0000
oii_7320_amp untied
oii_7320_sigma tied to nii_6584_sigma with factor 1.0000
oii_7320_vshift tied to nii_6584_vshift with factor 1.0000
oii_7330_amp tied to oii_7320_amp with factor 0.8163
oii_7330_sigma tied to nii_6584_sigma with factor 1.0000
oii_7330_vshift tied to nii_6584_vshift with factor 1.0000
siii_9069_amp is NOT FITTED
siii_9069_sigma tied to nii_6584_sigma with factor 1.0000 and FIXED
siii_9069_vshift tied to nii_6584_vshift with factor 1.0000 and FIXED
siii_9532_amp is NOT FITTED
siii_9532_sigma tied to nii_6584_sigma with factor 1.0000 and FIXED
siii_9532_vshift tied to nii_6584_vshift with factor 1.0000 and FIXED
--------------------------
Broad Balmer+helium lines:
--------------------------
h6_broad_amp tied to halpha_broad_amp with factor 1.0000
h6_broad_sigma tied to halpha_broad_sigma with factor 1.0000
h6_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hepsilon_broad_amp tied to halpha_broad_amp with factor 1.0000
hepsilon_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hepsilon_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hdelta_broad_amp tied to halpha_broad_amp with factor 1.0000
hdelta_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hdelta_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hgamma_broad_amp tied to halpha_broad_amp with factor 1.0000
hgamma_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hgamma_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hei_broad_4471_amp tied to halpha_broad_amp with factor 1.0000
hei_broad_4471_sigma tied to halpha_broad_sigma with factor 1.0000
hei_broad_4471_vshift tied to halpha_broad_vshift with factor 1.0000
heii_broad_4686_amp tied to halpha_broad_amp with factor 1.0000
heii_broad_4686_sigma tied to halpha_broad_sigma with factor 1.0000
heii_broad_4686_vshift tied to halpha_broad_vshift with factor 1.0000
hbeta_broad_amp tied to halpha_broad_amp with factor 1.0000
hbeta_broad_sigma tied to halpha_broad_sigma with factor 1.0000
hbeta_broad_vshift tied to halpha_broad_vshift with factor 1.0000
hei_broad_5876_amp tied to halpha_broad_amp with factor 1.0000
hei_broad_5876_sigma tied to halpha_broad_sigma with factor 1.0000
hei_broad_5876_vshift tied to halpha_broad_vshift with factor 1.0000
halpha_broad_amp is NOT FITTED
halpha_broad_sigma is NOT FITTED
halpha_broad_vshift is NOT FITTED
---------------------------
Narrow Balmer+helium lines:
---------------------------
h6_amp untied
h6_sigma tied to halpha_sigma with factor 1.0000
h6_vshift tied to halpha_vshift with factor 1.0000
hepsilon_amp untied
hepsilon_sigma tied to halpha_sigma with factor 1.0000
hepsilon_vshift tied to halpha_vshift with factor 1.0000
hdelta_amp untied
hdelta_sigma tied to halpha_sigma with factor 1.0000
hdelta_vshift tied to halpha_vshift with factor 1.0000
hgamma_amp untied
hgamma_sigma tied to halpha_sigma with factor 1.0000
hgamma_vshift tied to halpha_vshift with factor 1.0000
hei_4471_amp untied
hei_4471_sigma tied to halpha_sigma with factor 1.0000
hei_4471_vshift tied to halpha_vshift with factor 1.0000
heii_4686_amp untied
heii_4686_sigma tied to halpha_sigma with factor 1.0000
heii_4686_vshift tied to halpha_vshift with factor 1.0000
hbeta_amp untied
hbeta_sigma tied to halpha_sigma with factor 1.0000
hbeta_vshift tied to halpha_vshift with factor 1.0000
hei_5876_amp untied
hei_5876_sigma tied to halpha_sigma with factor 1.0000
hei_5876_vshift tied to halpha_vshift with factor 1.0000
halpha_amp untied
halpha_sigma untied
halpha_vshift untied
----------------
Forbidden lines:
----------------
nev_3346_amp untied
nev_3346_sigma tied to oiii_5007_sigma with factor 1.0000
nev_3346_vshift tied to oiii_5007_vshift with factor 1.0000
nev_3426_amp untied
nev_3426_sigma tied to oiii_5007_sigma with factor 1.0000
nev_3426_vshift tied to oiii_5007_vshift with factor 1.0000
oii_3726_sigma tied to oiii_5007_sigma with factor 1.0000
oii_3726_vshift tied to oiii_5007_vshift with factor 1.0000
oii_3729_amp untied
oii_3729_sigma tied to oiii_5007_sigma with factor 1.0000
oii_3729_vshift tied to oiii_5007_vshift with factor 1.0000
neiii_3869_amp untied
neiii_3869_sigma tied to oiii_5007_sigma with factor 1.0000
neiii_3869_vshift tied to oiii_5007_vshift with factor 1.0000
oiii_4363_amp untied
oiii_4363_sigma tied to oiii_5007_sigma with factor 1.0000
oiii_4363_vshift tied to oiii_5007_vshift with factor 1.0000
oiii_4959_amp tied to oiii_5007_amp with factor 0.3351
oiii_4959_sigma tied to oiii_5007_sigma with factor 1.0000
oiii_4959_vshift tied to oiii_5007_vshift with factor 1.0000
oiii_5007_amp untied
oiii_5007_sigma untied
oiii_5007_vshift untied
nii_5755_amp untied
nii_5755_sigma tied to oiii_5007_sigma with factor 1.0000
nii_5755_vshift tied to oiii_5007_vshift with factor 1.0000
oi_6300_amp untied
oi_6300_sigma tied to oiii_5007_sigma with factor 1.0000
oi_6300_vshift tied to oiii_5007_vshift with factor 1.0000
siii_6312_amp untied
siii_6312_sigma tied to oiii_5007_sigma with factor 1.0000
siii_6312_vshift tied to oiii_5007_vshift with factor 1.0000
nii_6548_amp tied to nii_6584_amp with factor 0.3399
nii_6548_sigma tied to oiii_5007_sigma with factor 1.0000
nii_6548_vshift tied to oiii_5007_vshift with factor 1.0000
nii_6584_amp untied
nii_6584_sigma tied to oiii_5007_sigma with factor 1.0000
nii_6584_vshift tied to oiii_5007_vshift with factor 1.0000
sii_6716_amp untied
sii_6716_sigma tied to oiii_5007_sigma with factor 1.0000
sii_6716_vshift tied to oiii_5007_vshift with factor 1.0000
sii_6731_sigma tied to oiii_5007_sigma with factor 1.0000
sii_6731_vshift tied to oiii_5007_vshift with factor 1.0000
oii_7320_amp untied
oii_7320_sigma tied to oiii_5007_sigma with factor 1.0000
oii_7320_vshift tied to oiii_5007_vshift with factor 1.0000
oii_7330_amp tied to oii_7320_amp with factor 0.8163
oii_7330_sigma tied to oiii_5007_sigma with factor 1.0000
oii_7330_vshift tied to oiii_5007_vshift with factor 1.0000
siii_9069_amp is NOT FITTED
siii_9069_sigma tied to oiii_5007_sigma with factor 1.0000 and FIXED
siii_9069_vshift tied to oiii_5007_vshift with factor 1.0000 and FIXED
siii_9532_amp is NOT FITTED
siii_9532_sigma tied to oiii_5007_sigma with factor 1.0000 and FIXED
siii_9532_vshift tied to oiii_5007_vshift with factor 1.0000 and FIXED
Done in #166.
In a recent GQP meeting, an argument was made that in objects with both broad and narrow line-emission, the narrow Balmer+helium line-widths should be tied to the forbidden lines.