[Rgc_analysis] [EXTERNAL] Re: [Rgc] RGC-analysis meeting tomorrow

Fri Feb 21 15:27:40 EST 2025

Hi Sebastian,
you'll have to do without me then on Tuesday, I'll be landing on Tuesday 
at noon Paris time, so I won't be able to run the RGC meeting.
But are we sure we'll have more information on Tuesday than we have now?
Derek, I doubt you messed up something in the sqlite files. We see in 
your comparison for the vertices a slight improvement after applying 
your constants, so it goes in the correct direction. I spoke with 
Raffaella this morning, and she says that it may be that we saw a 
positive effect in Fall22 due to the bigger size of the raster, and that 
in Spring23, where the raster radius is smaller, we may see that slight 
worsening due to the helix reconstruction algorithm in the CVT. I didn't 
understand all the details, but she says she discussed this with the 
review committee in the closed session after we left on Wednesday, and 
to her it seems like a possible explanation of the effects we observe.
I will ask her to send us an email with the explanation, it may be 
easier for me to understand it in writing :-)
Best regards,
Silvia

On 2025-02-21 12:29, Sebastian Kuhn wrote:
> Due to insufficient enthusiasm, I’d rather NOT try to schedule an
> ad-hoc meeting today - let’s discuss all the evidence that we
> (hopefully) will have collected by Tuesday during our regular RG-C
> Analysis Meeting.
> 
> - Sebastian
> 
>> On Feb 21, 2025, at 10:27, Holmberg, Derek <deholmberg at wm.edu>
>> wrote:
>> 
>> Hi Sebastian,
>> 
>> I'm free this afternoon, so I can certainly meet to discuss the
>> scripts I used to add information for the sqlite files. I really
>> hope I didn't mess up something incredibly basic that's causing
>> these discrepancies...
>> 
>> Sincerely,
>> Derek
>> 
>> -------------------------
>> 
>> From: Sebastian Kuhn <kuhn at jlab.org>
>> Sent: Friday, February 21, 2025 10:08 AM
>> To: Holmberg, Derek <deholmberg at wm.edu>
>> Cc: silvia <silvia at jlab.org>; pilleux <pilleux at jlab.org>; Samy
>> Polcher Rafael <spolcher at jlab.org>; Holmberg, Derek via Rgc_analysis
>> <rgc_analysis at jlab.org>
>> Subject: Re: [Rgc_analysis] [EXTERNAL] Re: [Rgc] RGC-analysis
>> meeting tomorrow
>> 
>> Dear all,
>> 
>> I am still trying to understand why some of the plots shown during
>> the (very successful!) calibration review show a worse resolution
>> for the “sqlite” version of the cooking vs. the “CCDB”
>> version. It would be helpful if someone could check whether
>> Derek’s scripts are working as intended - resulting in sqlite
>> having the exact same calibration constants as CCDB for EVERYTHING
>> other than the raster and beam offset constants? (Yes, the beam
>> offset constants should be set to zero, so that is correct). I am
>> also hoping to get a response from Noemie regarding the exact cuts
>> used in the elastic analysis.
>> 
>> Samy’s response (also quoted below) does not indicate any cuts on
>> vertex variables, so the slight difference in vertex reconstruction
>> doesn’t seem to have played any role. For the inbending
>> comparison, the difference between sqlite and CCDB seems small, but
>> I am still puzzled about the Delta-phi resolution, which seems worse
>> for sqlite. (However, this may be partially a “Trompe-l'œil”
>> because the peak is a little lower - if one were to re-normalize
>> both distributions to the same max peak height, the difference would
>> probably be truly minimal). In any case, for outbending the
>> difference is more significant - sqlite is clearly worse for
>> Delta-phi, while the vertex distributions still look better for both
>> electrons and protons, at least for Vy (Vx not shown). Samy, would
>> it be possible to plot Delta-vx and Delta-vy (difference between p
>> and e), again comparing CCDB and sqlite for both inbending and
>> outbending? Noëmie, would you be able to do the same for elastic
>> events?
>> 
>> If you have time this afternoon and would like to discuss this in a
>> meeting, I could set something up for 1:00 p.m. (until at most 3
>> p.m.) today. Please let me know your preferences! Otherwise, I would
>> suggest we make a final decision on Tuesday based on all information
>> we can collect until then (please send as soon as you have it).
>> 
>> Thanks - Sebastian
>> 
>> ————
>> 
>> Hi Sebastian,
>> 
>> Please find attached a document detailing the event selection,
>> fiducial and exclusivity cuts I use. I've selected the relevant
>> pages from the release note I'm writing, but the QADB selection cuts
>> of course do not apply to the spring23 test runs.
>> 
>> The plots with the separate inbending/outbending and the vertex
>> plots are in the slides attached. The details on the paths ect..
>> used are in the email I sent on Tuesday.
>> 
>> Best,
>> 
>> Samy
>> 
>> On 20/02/2025 17:06, Sebastian Kuhn wrote:
>> 
>>> Hi Samy,
>>> 
>>> thanks for your response. I was also asking explicitly WHAT cuts
>>> you are using
>> 
>>> On Feb 20, 2025, at 10:26, Holmberg, Derek <deholmberg at wm.edu>
>>> wrote:
>>> 
>>> Hi Silvia,
>>> 
>>> To generate the latest version of the sqlite, I used the
>>> following script:
>>> /group/clas12/packages/local/bin/clas12-ccdb-mysql2sqlite.py
>>> ~/work/My_Sqlite/latest.sqlite
>>> 
>>> This script grabs the latest snapshot from ccdb, so the one I
>>> grabbed for the latest raster stuff was probably taken on January
>>> 29th.
>>> 
>>> Just to make sure that I didn't make any other errors, the shell
>>> script that I used to add my raster constants can be found at
>>> 
>>> 
>> 
> /w/hallb-scshelf2102/clas12/holmberg/My_Sqlite/Add_ADC_to_Position_Data_Sp23.sh
>>> 
>>> 
>>> In this script, I set the beam-offset to zero for all runs:
>>> /group/clas12/bin/ccdb add /geometry/beam/position -v
>>> rgc_spring2023 -r 17479-99999 derek_zero_beam_position_sp23.txt
>>> 
>>> Then, within a loop, I add my raster constants to the sqlite
>>> file:
>>> /group/clas12/bin/ccdb add /calibration/raster/adc_to_position -v
>>> rgc_spring2023 -r ${j}-${j}
>>> Text_Files/Run_${j}_ADC_to_Position.txt
>>> 
>>> Could there be an issue with setting the beam offsets to zero? I
>>> thought that setting the offset to zero would be necessary since
>>> the detached vertex tracking method is agnostic to any offset.
>>> 
>>> Please let me know if there's anything else I can help with!
>>> 
>>> Sincerely,
>>> Derek
>>> -------------------------
>>> 
>>> From: silvia <silvia at jlab.org>
>>> Sent: Thursday, February 20, 2025 10:11 AM
>>> To: Sebastian Kuhn <kuhn at jlab.org>
>>> Cc: Holmberg, Derek <deholmberg at wm.edu>; pilleux
>>> <pilleux at jlab.org>; Samy Polcher Rafael <spolcher at jlab.org>
>>> Subject: Re: [Rgc_analysis] [EXTERNAL] Re: [Rgc] RGC-analysis
>>> meeting tomorrow
>>> 
>>> Hello all,
>>> Derek, did you have a chance to double-check your CCDB copying
>>> procedures, to address Sebastian's question?
>>> For the rest, I can say that CJ was for sure the same for the two
>>> cookings. Noemie and Samy can reply to the other questions.
>>> Best regards,
>>> Silvia
>>> 
>>> On 2025-02-19 10:01, Sebastian Kuhn wrote:
>>>> Dear all,
>>>> 
>>>> coming back to the contradictory results for the comparison
>>> between
>>>> sqlite and CCDB (improved vertex resolution, worse phi
>>> resolution for
>>>> elastic and pDVCS), I was wondering whether we are 100% sure
>>> that
>>>> everything else was exactly the same for both versions other
>>> than the
>>>> raster calibration constants. I remember last time we had this
>>> issue,
>>>> we found that there was something else that had changed between
>>> CCDB
>>>> and sqlite (or something that hadn’t been completely copied
>>> over to
>>>> sqlite from CCDB) which confounded the effect of ONLY the raster
>>>> constants.
>>>> 
>>>> In addition, it would be helpful to get as much detail as
>>> possible for
>>>> all conditions for both the elastic and the pDVCS analysis using
>>> both
>>>> versions:
>>>> - COATJAVA version (same?)
>>>> - any and all cuts? (vertex, coincidence, fiducial,…)
>>>> - any background subtractions?
>>>> 
>>>> I am particularly puzzled why Delta-phi for elastics seems
>>> better for
>>>> CCDB than for sqlite - to first order, there should be NO effect
>>> at
>>>> all from the vertex, and if anything a tighter vertex
>>> distribution
>>>> should give an improvement.
>>>> 
>>>> I think we should convince ourselves that this improvement is a
>>> real
>>>> effect before we decide to use CCDB instead of sqlite for the
>>> Winter
>>>> cook.
>>>> 
>>>> - Sebastian
>>>> 
>>>>> On Feb 18, 2025, at 14:50, Holmberg, Derek via Rgc_analysis
>>>>> <rgc_analysis at jlab.org> wrote:
>>>>> 
>>>>> Hi All,
>>>>> 
>>>>> I finished some updated slides for the raster calibration
>>> review. I
>>>>> uploaded the PowerPoint presentation (as a .pptx file) to the
>>> wiki
>>>>> for today's meeting:
>>>>> 
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fclasweb.jlab.org%2Fwiki%2Findex.php%2FRGC_Feb18&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722463341%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=Cv9p6EKcgmU3FIN0VQgmof%2B%2F7fJ8U9FZTo7NXKrI5Qg%3D&reserved=0
>>> [1]
>>>>> 
>>>>> Since the recooked runs were carbon and NH3, I made plots for
>>> one
>>>>> carbon (17485) and one NH3 (17577) run. I used the sidisdvcs
>>> trains
>>>>> for my plots, with each coming from the following directories:
>>>>> 
>>>>> Sqlite:
>>>>> 
>>>> 
>>> 
>> 
> /volatile/clas12/rg-c/production/dst/11.1.0_02_04_25_dst_sqlite/dst/train/sidisdvcs/
>>>>> 
>>>>> CCDB:
>>>>> 
>>>> 
>>> 
>> 
> /volatile/clas12/rg-c/production/dst/11.1.0_01_24_25_dst2/dst/train/sidisdvcs/
>>>>> 
>>>>> 
>>>>> I included plots of the vx, vy, and vz distributions for FD
>>>>> electrons and CD pi+ as well as plots of the polar and
>>> azimuthal
>>>>> angular distributions of each particle. I also included some
>>> plots
>>>>> of the angle between the pi+ and electron tracks and the
>>> delta_Vz
>>>>> distribution for both sqlite and ccdb.
>>>>> 
>>>>> Please let me know if there are any issues with the formatting
>>> or
>>>>> other things that need to be plotted; if you're copying things
>>> from
>>>>> my slides, make sure to copy and paste the entire slide since I
>>> put
>>>>> some extra objects in the slides for the "preliminary"
>>> watermarks as
>>>>> well as to cover up some mislabeled axes (whoops!).
>>>>> 
>>>>> Sincerely,
>>>>> Derek Holmberg
>>>>> 
>>>>> -------------------------
>>>>> 
>>>>> From: Rgc <rgc-bounces at jlab.org> on behalf of Samy Polcher
>>> Rafael
>>>>> via Rgc <rgc at jlab.org>
>>>>> Sent: Tuesday, February 18, 2025 1:39 PM
>>>>> To: silvia <silvia at jlab.org>; Rgc <rgc at jlab.org>; Rgc Analysis
>>>>> <rgc_analysis at jlab.org>
>>>>> Subject: Re: [Rgc] [Rgc_analysis] RGC-analysis meeting tomorrow
>>>>> 
>>>>> Dear all,
>>>>> 
>>>>> I have updated the slides from last week's meeting to add the
>>>>> electron
>>>>> and proton Vx, Vy, Vz distributions for the ccdb and sqlite
>>>>> versions.
>>>>> The .odp version can be found here
>>>>> 
>>>> 
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fclasweb.jlab.org%2Fwiki%2Findex.php%2FFile%3ASpring23_pDVCS.odp&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722492053%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=cbYlud3osZ0FFZgJm%2FXWaoCwrwJUgTWFjlvRYBiHLfg%3D&reserved=0
>>> [2]
>>>>> [1] and a
>>>>> pdf version is attached.
>>>>> 
>>>>> I have re-run on the files in the following directories to
>>> double
>>>>> check:
>>>>> 
>>>>> - ccdb:
>>>>> 
>>> 
>> /volatile/clas12/rg-c/production/dst/11.1.0_01_24_25_dst2/dst/train/
>>>>> 
>>>>> - sqlite:
>>>>> 
>>>> 
>>> 
>> 
> /volatile/clas12/rg-c/production/dst/11.1.0_02_04_25_dst_sqlite/dst/train/
>>>>> 
>>>>> On the vertex distributions the only significant difference I
>>>>> noticed is
>>>>> a small shift on the electron and proton Vy of the sqlite
>>> compared
>>>>> to ccdb.
>>>>> 
>>>>> I then tried to separate inbending and outbending runs and it
>>> shows
>>>>> that
>>>>> most of the difference between the two versions is coming from
>>>>> outbending runs. The exclusivity and vertex variables in
>>> inbending
>>>>> are i
>>>>> think compatible given the statistics available.
>>>>> 
>>>>> Best,
>>>>> 
>>>>> Samy
>>>>> 
>>>>> On 17/02/2025 22:32, silvia via Rgc_analysis wrote:
>>>>>> Dear all,
>>>>>> we'll have our RGC-analysis meeting tomorrow, Tuesday, at
>>> 8:30AM
>>>>> JLab
>>>>>> time, zoom link:
>>>>> 
>>>> 
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fjlab-org.zoomgov.com%2Fj%2F1609368114&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722512842%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=v%2B85kTB8Boht3R9Zu9hc5k2f4PBRxBqXTZpx0w0LPYc%3D&reserved=0
>>> [3]
>>>>> [2] , wiki
>>>>>> page:
>>>>> 
>>>> 
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fclasweb.jlab.org%2Fwiki%2Findex.php%2FRGC_Feb18&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722530199%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=yHbIqpIx8nNrmg5AYPhDVI1pTUgWhJvnqx%2BnYDnGjuY%3D&reserved=0
>>> [1]
>>>>> [3]
>>>>>> Please let me know if you plan to present something.
>>>>>> Best regards,
>>>>>> Silvia
>>>>>> _______________________________________________
>>>>>> Rgc_analysis mailing list
>>>>>> Rgc_analysis at jlab.org
>>>>>> 
>>>>> 
>>>> 
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.jlab.org%2Fmailman%2Flistinfo%2Frgc_analysis&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722546605%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=cIThNrjW2JcBDILJhOLoLnGxemeJDlt23jOQgRW93nY%3D&reserved=0
>>> [4]
>>>>> [4] _______________________________________________
>>>>> Rgc_analysis mailing list
>>>>> Rgc_analysis at jlab.org
>>>>> 
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.jlab.org%2Fmailman%2Flistinfo%2Frgc_analysis&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722563807%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=qwEupVbvPqQ960OOkFdIDx8p%2BgkzrzWd28eZvOBcDH0%3D&reserved=0
>>> [4]
>>>> 
>>>> 
>>>> 
>>>> Links:
>>>> ------
>>>> [1]
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fclasweb.jlab.org%2Fwiki%2Findex.php%2FFile%3ASpring23_pDVCS.odp&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722582143%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=plOVEJeo1GL%2FD3X3XU%2F8QTM3Y1x%2B41q1U6udhAZF3BU%3D&reserved=0
>>> [2]
>>>> [2]
>>>> 
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-3A__jlab-2Dorg.zoomgov.com_j_1609368114%26amp%3Bd%3DDwMFAw%26amp%3Bc%3DCJqEzB1piLOyyvZjb8YUQw%26amp%3Br%3DaSEBncnFTdfouxOejKajYG--Ygz0DFQolIcHUhF20pw%26amp%3Bm%3DDdTzR7_OOnUk3NlPemAID40leS_ymL_FeOHip8Hx4hj_CfDOhUnlavkLCx3aUIAe%26amp%3Bs%3DFlUc4g7EFxM5LR8cX3Bpop6QV-k0oYJLwcAs3byGN14%26amp%3Be%3D&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722600677%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=DPEM5cYe%2BU%2B3ci0OYFJ0q%2FSIfMfjKi4q2mjV03yAHO8%3D&reserved=0
>>> [5]
>>>> [3]
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fclasweb.jlab.org%2Fwiki%2Findex.php%2FRGC_Feb18&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722618671%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=FoppIMjQ8UHxGi6QMb73B6FjoB7m%2FXkkOm2j66e2Rmc%3D&reserved=0
>>> [1]
>>>> [4]
>>> 
>> 
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.jlab.org%2Fmailman%2Flistinfo%2Frgc_analysis&data=05%7C02%7Cdeholmberg%40wm.edu%7C1037694749364a7bf38108dd51c0cea4%7Cb93cbc3e661d40588693a897b924b8d7%7C0%7C0%7C638756610722635673%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=0QmFsWT5sBeJPmOUQSDScYtood%2FaBcB4YfQ2HWeGDN0%3D&reserved=0
>>> [4]
> 
> 
> 
> Links:
> ------
> [1] https://clasweb.jlab.org/wiki/index.php/RGC_Feb18
> [2] https://clasweb.jlab.org/wiki/index.php/File:Spring23_pDVCS.odp
> [3] 
> https://urldefense.proofpoint.com/v2/url?u=https-3A__jlab-2Dorg.zoomgov.com_j_1609368114&d=DwMF-g&c=CJqEzB1piLOyyvZjb8YUQw&r=cvGaa-sH-cjAi4BvycGtCA&m=R1DsuFlzGo0ridKhy2du4lkTdYtnDQ3vzP3v-8FTWKMXRBNvGkS-0wxmPsoN3Luu&s=2JiXr_GluYbW_qZpcvmixXkaMTfhLAepL4bhe7mN1wQ&e=
> [4] https://mailman.jlab.org/mailman/listinfo/rgc_analysis
> [5] 
> https://urldefense.proofpoint.com/v2/url?u=https-3A__jlab-2Dorg.zoomgov.com_j_1609368114&amp;d=DwMFAw&amp;c=CJqEzB1piLOyyvZjb8YUQw&amp;r=aSEBncnFTdfouxOejKajYG--Ygz0DFQolIcHUhF20pw&amp;m=DdTzR7_OOnUk3NlPemAID40leS_ymL_FeOHip8Hx4hj_CfDOhUnlavkLCx3aUIAe&amp;s=FlUc4g7EFxM5LR8cX3Bpop6QV-k0oYJLwcAs3byGN14&amp;e=