Observation group contribution to objective function = 0 in PEST_HP #348
-
|
Hi All, Our setup of PEST_HP includes observations that are all in a single group. Strangely, the contribution of that group to the objective is always zero while the objective is something like 3.0e+9. It seems like the contribution of the observation group to the objective should be 3.0e+9 since all observations are in one group. Has anyone experienced this and know of a misplaced setting in the control file that can cause this to happen? Maybe only having one observation group defined causes a breakdown in PEST_HP's mathematical formulation? Thanks, |
Beta Was this translation helpful? Give feedback.
Replies: 4 comments 4 replies
-
|
Can you post the rec file that shows group phi report? I don't know off hand any reason why the sum of the group phis wouldn't equal the total phi... |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for your reply. I pasted some example lines from my .rec file in my reply to @wkitlasten. I wonder if having zero values of observations can cause this (@wkitlasten questions got me wondering this). I could share the .rec file with you offline if that would be of any interest. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
I don't think zero-weighted obs should be a problem...and with non-zero residuals, there ought to be a non-zero phi...feel free to send me that rec file |
Beta Was this translation helpful? Give feedback.
-
|
Only 2 things I can think of off the top of my head, only b cause I’ve made
the mistakes myself at some point:
1) are the obsvals actual obsvals, or are they simulated equivalents
accidentally left over from building the pest files? What are the residuals?
2) are the weights non-zero?
3) are you doing anything to “rebalance” the objective function?
"Perfect spheres are pointless."
…On Tue, 20 May 2025 at 12:38 PM, sjepsen395 ***@***.***> wrote:
Hi All,
Our setup of PEST_HP includes observations that are all in a single group.
Strangely, the contribution of that group to the objective is always zero
while the objective is something like 3.0e+9. It seems like the
contribution of the observation group to the objective should be 3.0e+9
since all observations are in one group. Has anyone experienced this and
know of a misplaced setting in the control file that can cause this to
happen? Maybe only having one observation group defined causes a breakdown
in PEST_HP's mathematical formulation?
Thanks,
Steve
—
Reply to this email directly, view it on GitHub
<#348>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ADSJXRDDRBSO3XTQVAUONFL27J2PXAVCNFSM6AAAAAB5PBNSJWVHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZYGM2DONZSG4>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Thanks for your questions.
|
Beta Was this translation helpful? Give feedback.
-
|
Cool, thanks. I'll follow up with that .rec file.
…On Tue, May 20, 2025 at 8:20 AM J Dub ***@***.***> wrote:
I don't think zero-weighted obs should be a problem...and with non-zero
residuals, there ought to be a non-zero phi...feel free to send me that rec
file
—
Reply to this email directly, view it on GitHub
<#348 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/BDQWLIXEROGOIJEBWRP2ZC327NB4LAVCNFSM6AAAAAB5PBNSJWVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMRQHA3TAMQ>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
I don’t recall is pest_hp throws an error when obs names are too long or
does something weird, but aren’t those obs names a bit long for pest_hp?
"Perfect spheres are pointless."
…On Wed, 21 May 2025 at 4:23 AM, sjepsen395 ***@***.***> wrote:
Hi Jeremy,
The model has not been approved for public release, so I want to make sure
that I don't post anything that general public can access. Can I send you
that .rec file privately but then keep the dialogue public on Github? If
so, how would I send you that file? To your INTERA email?
Nice meeting you at CWEMF!
Steve
On Tue, May 20, 2025 at 8:20 AM J Dub ***@***.***> wrote:
> I don't think zero-weighted obs should be a problem...and with non-zero
> residuals, there ought to be a non-zero phi...feel free to send me that
rec
> file
>
> —
> Reply to this email directly, view it on GitHub
> <
#348 (reply in thread)>,
> or unsubscribe
> <
https://github.com/notifications/unsubscribe-auth/BDQWLIXEROGOIJEBWRP2ZC327NB4LAVCNFSM6AAAAAB5PBNSJWVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMRQHA3TAMQ>
> .
> You are receiving this because you authored the thread.Message ID:
> ***@***.***>
>
—
Reply to this email directly, view it on GitHub
<#348 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ADSJXREHNEAPXP2RBMRGLU327NJHDAVCNFSM6AAAAAB5PBNSJWVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTGMRQHEZDIMY>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
The observation name lengths are as high as 20 characters. That is PEST_HP's limit. I don't know if PEST_HP throws an error or just truncates the names. It is a good observation though. @jtwhite79 kindly had a look at my .rec file and didn't seem to find any glaring issues. It was thus concluded that the issue described in my original post is probably only a reporting error in PEST_HP and can be ignored. Thank you all for your input. Steve |
Beta Was this translation helpful? Give feedback.
The observation name lengths are as high as 20 characters. That is PEST_HP's limit. I don't know if PEST_HP throws an error or just truncates the names. It is a good observation though.
@jtwhite79 kindly had a look at my .rec file and didn't seem to find any glaring issues. It was thus concluded that the issue described in my original post is probably only a reporting error in PEST_HP and can be ignored.
Thank you all for your input.
Steve