add configs for prod 20240716_v0.10.7_src10_dec3476_tuned #470
Conversation
Hi @ToraTherese Unfortunately this is failing because is empty 😬 The dataset had been separated in train/test ones: So you could start from these and do the dl1ab stage without the train/test split stage, e.g.: etc... opening an issue... |
| }, | ||
| "image_modifier": { | ||
| "increase_nsb": true, | ||
| "extra_noise_in_dim_pixels": 1.327, | ||
| "extra_bias_in_dim_pixels": 0.41, | ||
| "transition_charge": 8, | ||
| "extra_noise_in_bright_pixels": 0.0 | ||
| } |
There was a problem hiding this comment.
this production seems quite similar to 20240430_v0.10.4_src9_dec3476_tuned (same dec line and somewhat close image modifier parameter). @ToraTherese have you tried that production already (even if later more fine-tuning of the NSB is needed)?
There was a problem hiding this comment.
It is indeed very similar, just different image modifier parameter. @morcuended, what do you mean by using a later fine-tuning of the NSB? I was under the impression that requesting a re-analysis was the intended method of getting the required NSB tuning
There was a problem hiding this comment.
I mean that if you have tried that prod already before producing the new one
There was a problem hiding this comment.
Hi @ToraTherese
I think @morcuended is actually right here.
The difference between production 20240430_v0.10.4_src9_dec3476_tuned and the one you are requesting is
"image_modifier": {
"increase_nsb": true,
"extra_noise_in_dim_pixels": 1.577,
"extra_bias_in_dim_pixels": 0.463,
"transition_charge": 8,
"extra_noise_in_bright_pixels": 1.892
}
vs
"image_modifier": {
"increase_nsb": true,
"extra_noise_in_dim_pixels": 1.327,
"extra_bias_in_dim_pixels": 0.41,
"transition_charge": 8,
"extra_noise_in_bright_pixels": 0.0
}
Using directly the trained model from 20240430_v0.10.4_src9_dec3476_tuned for src10 would probably lead to very similar results - have you tried already?
There was a problem hiding this comment.
As a reference, you may have a look at https://indico.cta-observatory.org/event/5532/contributions/45108/attachments/25665/37594/DL2_auto.pdf
The difference of nsb here (0.25 or 16%) is quite small and differences will most probably be hidden by systematics.
There was a problem hiding this comment.
Hi @ToraTherese
Have you been able to work with existing production 20240430_v0.10.4_src9_dec3476_tuned ?
Should #473 be closed ?
New Prod config
Self check-list:
az_telinstead ofsin_az_telif data to be analyzed have been produced with lstchain <= v0.9.7Prod_ID
20240716_v0.10.7_src10_dec3476_tuned
Short description of the config
Config for reprocessing of Gamma Diffuse MC at declination line of 34.76 deg, based on 20240430_v0.10.4_src9$
NSB tuning adapted to the observed data of src10.
Why this config is needed
Need diffuse MC with tuning for new source: src10