Closed vivianmiranda closed 5 months ago
Perhaps this was fixed with https://github.com/CobayaSampler/cobaya/pull/344 - can you check on latest master?
The DESI Y1 likelihood in the current master (from DESI team) returns a chi2 of about 21 on the test point.
I will check that.
On another note - the cover matrix of Pantheon+ (and dv) looks like a javascript code
https://github.com/CobayaSampler/sn_data/blob/master/PantheonPlus/Pantheon%2BSH0ES_STAT%2BSYS.cov
And I am getting reading errors. Is that the correct file?
Also DES-Y5 is missing (STAT+SYS.txt)
No it’s not, should be same as
Antony
From: Vivian Miranda @.> Sent: Monday, April 8, 2024 7:24 PM To: CobayaSampler/cobaya @.> Cc: Antony Lewis @.>; Mention @.> Subject: Re: [CobayaSampler/cobaya] Interpolation error creating a delta chi2 = 20 on DESI likelihood (Issue #352)
I will check that.
On another note - the cover matrix of Pantheon+ (and dv) looks like a javascript code
https://github.com/CobayaSampler/sn_data/blob/master/PantheonPlus/Pantheon%2BSH0ES_STAT%2BSYS.cov
And I am getting reading errors. Is that the correct file?
— Reply to this email directly, view it on GitHubhttps://github.com/CobayaSampler/cobaya/issues/352#issuecomment-2043284330, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AA6GQI7UQEAYPEOY6YUANSDY4LHETAVCNFSM6AAAAABF4BI7IKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANBTGI4DIMZTGA. You are receiving this because you were mentioned.Message ID: @.***>
PantheonPlus should be fixed now. (I don't have a test value to add a unit test)
Also DES Y5 (thanks, looks like an inconsistency in the names in the original repo)
Is DESI OK for you now on master, can we close?
Hi @cmbant @JesusTorrado
In both CAMB and CLASS (older version of Cobaya - so I have not tested if this problem is absent on new versions). I was getting chi2 ~40 for DESI and horrible predictions on bin z = 1.4
Then I made this change (picture attached)
and chi2 drop to 20.
With the fix (random cosmology - see the z=1.49 prediction and DESI chi2):
Without the fix