markliuchina
commented
1 year ago I used to be encountered with the same issue and I also noticed the issue #9 and then checked the filter files one by one.
After that, nothing is going to be fixed.
The question may also come from the fitting component parameters setting.
I think it may be caused by the parameter range value.
At that time, I was trying to imitate one of Adam's work (Carnal et al. 2020, published on MNRAS, staa1535).
I set the double-power-law SFH component, following the instructions by Adam's paper (strictly input the parameter, without thinking too much about it, my faults), but the jupyter-notebook failed, in the way of your feedback.
I checked the example file, which describes the "double-power-law SFH component" fitting and compared carefully these two codes line by line and I found where is the problem.
(Carnall et al. 2020)
Please look at this screenshot, the stellar mass range is (1, 10^13) and we wish it in log space.
As a result, you may set the input parameter in this way. (for log(1) ==0, log(10^13) == 13)
dblplaw["massformed"] = (0., 13.)
And when you run the fitting, the code fails.
By testing, the lower limit of some parameter with a LOG prior should not be 0
Maybe a value closer to 0 but not equals to 0 is more plausible?
so I set :
dblplaw["massformed"] = (0.00001, 13.) #因为无法把这个质量下限的log_10取0，所以近似取0.00001，防止kernel崩溃。
dblplaw["metallicity"] = (0.1, 2.5) #对数空间下的取值应该是-1到0.4之间，对应就是0.1到2.5
dblplaw["massformed_prior"] = "log_10"
dblplaw["metallicity_prior"] = "log_10"And then the problem fixed! (But I am not sure if this will bias the fitting result) I suggest that you check the parameter range before fitting. Here is the component setup code I used before:
dblplaw = {}
dblplaw["tau"] = (0.1, 15.)
dblplaw["tau_prior"] = "uniform"
dblplaw["massformed"] = (1., 13.)
dblplaw["massformed_prior"] = "log_10"
dblplaw["metallicity"] = (0.1, 2.5)
dblplaw["metallicity_prior"] = "log_10"
dblplaw["alpha"] = (0.01, 1000.)
dblplaw["beta"] = (0.01, 1000.)
dblplaw["alpha_prior"] = "log_10"
dblplaw["beta_prior"] = "log_10"
dust = {}
dust["type"] = "Calzetti"
dust["delta"] = (-0.3, 0.3)
dust["delta_prior"] = "Gaussian"
dust["delta_prior_mu"] = 0.
dust["delta_prior_sigma"] = 0.1
dust["B"] = (0., 5.)
dust["B_prior"] = "uniform"
dust["Av"] = (0., 8.)
dust["Av_prior"] = "uniform"
dust["qpah"] = 2.
dust["umin"] = 1.
dust["gamma"] = 0.01
nebular = {}
nebular["logU"] = -3.
fit_info = {}
fit_info["redshift"] = (0., 10.)
fit_info["redshift_prior"] = "uniform"
fit_info["dblplaw"] = dblplaw
fit_info["dust"] = dust
fit_info["nebular"] = nebularYou could just paste it in your notebook and see whether the problem is the the same as I describe above. Hope it help!
Mingfeng Liu
ACCarnall
elNava
For example, you set your parameter like this:
Hope it fix things.
Hi,
I am having the same error detailed in issue #9. In that case they mention that the filter curve was the problem, but I am using NIRCam filters taken directly from SVO. I am following the steps in the notebooks, this is my input:
And the error:
Which is pretty similar to the issue previous mentioned. I already checked that the SVO filters are in Angstrom, I do not know what else can I check.
Thanks in advance. Cheers, Benjamin