From: Kaike Pan Subject: Re: [sloan-25mlog 4435] Re: 2.5m obslog 20190319 (SJD 58562) Date: Wed, 20 Mar 2019 11:42:12 -0600 --Apple-Mail=_D040F9B9-76DB-461D-B018-3AE474083FC1 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=utf-8 We ran some tests with 20 point data sets and found that the fitting rms will be larger if we allow rotator move during the data collection. However, with the rotator axis, we had pointing models with fitting rms at ~ 2 arc seconds, but the telescope could be off set by more than 15 arc seconds when goto field (before last month's twist on the instrument block, we saw 30+ arc-seond pointing occasionally).=20 We know that star images will fall on slightly different spots on CCD = for different rotator angle (due to the rotator center doesn=E2=80=99t = exactly overlap with the instrument center) for the same AZ/Alt coordinates. Because the = actual=20 point errors are much large than the point model fitting rms, I feel = that part of the inconsistency might be due to leaving the rotator axis out in = slews of the data collection. Therefore, I decided to let rotator move (slews = will be longer and more out-close-loop to deal with) last night. The hope is = that even the fitting rms is worse, the actual pointing may be better. Last = night,=20 we observed two fields after the new point model, both has guiding stars = on=20 gotoField. It is a good sign, but only the time will tell. I think that all fitting are on new data. The fitting in the first step = depends on the input model, but the final model should not. In the past, we always = fit=20 pointing model data with the same empirical terms. By the Tpoint = documents, some of these terms are not significant, should not be there (they may = reduce fitting rms, but could actually increase pointing error). Last month, we = use a different fitting approach in aiming to pick up =E2=80=9Cright=E2=80=9D = empirical terms (except for must have physical terms), which seems worked well. With the point = model from the new fitting process, we see the results in the log. If I fit the last night data with the =E2=80=9Cold=E2=80=9D fitting = process and use the pointing model from the =E2=80=9Cold=E2=80=9D fitting process, we do see changes in = different steps as show below: * fix * fit coeff change value sigma 1 IE -2.64 2 IA +96.44 3 PZZ1 +123.40 4 AN -9.75 5 AW +14.40 6 CA -21.26 7 HWCA2 -0.81 8 HWSA2 +0.29 9 HNCA1 -0.82 10 HNSA1 +0.53 11 ACEC +0.29 12 ACES -1.39 13 TX +3.46 Sky RMS =3D 8.62 Popn SD =3D 8.62 * use ie ia pzz1 an aw ca * fit coeff change value sigma 1 IE -3.786 -6.43 0.744 2 IA +1.008 +97.45 1.216 3 PZZ1 -0.156 +123.25 0.937 4 AN -0.118 -9.87 0.347 5 AW -4.030 +10.37 0.347 6 CA +5.230 -16.03 0.825 7 HWCA2 -0.81 8 HWSA2 +0.29 9 HNCA1 -0.82 10 HNSA1 +0.53 11 ACEC +0.29 12 ACES -1.39 13 TX +3.46 Sky RMS =3D 3.69 Popn SD =3D 3.77 Observation #54 is a weak outlier candidate. * use * fit coeff change value sigma 1 IE -2.563 -8.99 0.963 2 IA -0.529 +96.92 1.154 3 PZZ1 -9.502 +113.75 3.391 4 AN +0.374 -9.49 0.408 5 AW +0.056 +10.43 0.410 6 CA +0.653 -15.38 0.856 7 HWCA2 +1.071 +0.26 0.542 8 HWSA2 -0.483 -0.19 0.540 9 HNCA1 +1.617 +0.80 0.642 10 HNSA1 -0.842 -0.31 0.643 11 ACEC -0.316 -0.02 0.773 12 ACES -1.142 -2.54 0.778 13 TX +4.777 +8.24 1.664 Sky RMS =3D 3.42 Popn SD =3D 3.58 Observation #41 is a weak outlier candidate. If I fit the new data with auto fitting process by using the pointing model from the =E2=80=9Cold=E2=80=9D process, I got: coeff change value sigma 1 IA +0.000 +97.12 1.085 2 IE -0.000 -8.65 0.873 3 & ACES -0.000 -2.46 0.647 4 & HECASE5 +0.000 +1.67 0.563 5 & HECA2CE +0.000 +1.29 0.563 6 NPAE +0.00 7 CA -0.000 -15.62 0.736 8 AN +0.000 -9.49 0.352 9 AW -0.000 +10.41 0.310 10 & PZZ1 -0.000 +113.13 3.271 11 TX +0.000 +8.60 1.606 Sky RMS =3D 3.29 Popn SD =3D 3.41 Observation #24 is a weak outlier candidate. If I fit the new data set using the new fitting process and use the = point model that we used last month (from the new fitting process), I got a very similar = model. coeff change value sigma 1 IA +0.000 +97.12 1.085 2 IE +0.000 -9.02 1.328 3 & ACES -0.000 -2.46 0.647 4 & HECASE5 +0.000 +1.69 0.563 5 & HECA2CE +0.000 +1.29 0.563 6 & HECE3 +0.000 -5.10 2.293 7 NPAE +0.00 8 CA -0.000 -15.62 0.736 9 AN +0.000 -9.48 0.353 10 AW +0.000 +10.40 0.310 11 TF -0.000 +116.34 6.365 12 TX +0.000 +18.52 3.405 Sky RMS =3D 3.29 Popn SD =3D 3.42 Observation #24 is a weak outlier candidate. The new fitting process results slightly better fitting rms with less empirical terms. I think it=E2=80=99s better. So I have implemented my new model.=20 I plan to run the way for a couple of more models, if it works well, I will update the relevant documents. > On Mar 20, 2019, at 10:20 AM, Mark Klaene wrote: >=20 >=20 > On Mar 20, 2019, at 6:29 AM, dmbiz@apo.nmsu.edu = wrote: >=20 >>=20 >> APACHE POINT OBSERVATORY SDSS 2.5M >>=20 >> OBSERVING LOG >> Tuesday March 19, 2019 >>=20 >> (SJD 58562) >=20 >=20 >=20 > could you please explain the pointing model data=20 >=20 > 1. is one of these and if so which one is new data but old fit ? >=20 > 2. Which is new data with new fit and resulting rms errors? >=20 > 3. why do none of these show real change in value ? =20 >=20 > mark=20 >=20 >=20 >=20 >=20 >=20 >>=20 > We got clouds from ~6:40 to 7:30 UT, which affected the signal for = some stars > (~20% of all). Seeing flared to 3.5" during the last several stars.=20 >=20 > * indat ptdata_2019-03-20T03:28:38.5.dat > * fix > * fit >=20 > coeff change value sigma >=20 > 1 IA +88.01 > 2 IE -140.39 > 3 & HASA2 +0.40 > 4 & ACES -2.40 > 5 & HESE +172.72 > 6 & HECE2 +40.67 > 7 NPAE -13.83 > 8 CA -0.67 > 9 & HSCE3 -1.27 > 10 AN -9.56 > 11 AW +10.44 > 12 & HWCA2 +0.27 > 13 TF +112.37 > 14 PZZ1 +10.67 >=20 > Sky RMS =3D 3.42 > Popn SD =3D 3.42 >=20 > * use ie ia pzz1 an aw ca > * fit >=20 > coeff change value sigma >=20 > 1 IA +0.000 +88.01 1.125 > 2 IE -0.000 -140.39 0.688 > 3 & HASA2 +0.40 > 4 & ACES -2.40 > 5 & HESE +172.72 > 6 & HECE2 +40.67 > 7 NPAE -13.83 > 8 CA -0.000 -0.67 0.763 > 9 & HSCE3 -1.27 > 10 AN -0.000 -9.56 0.321 > 11 AW -0.000 +10.44 0.321 > 12 & HWCA2 +0.27 > 13 TF +112.37 > 14 PZZ1 -0.000 +10.67 0.867 >=20 > Sky RMS =3D 3.42 > Popn SD =3D 3.49 >=20 > Observation #66 is a weak outlier candidate. >=20 > * use > * fit >=20 > coeff change value sigma >=20 > 1 IA +0.000 +88.01 30.967 > 2 IE +0.000 -140.39 4.549 > 3 & HASA2 -0.000 +0.40 0.593 > 4 & ACES +0.000 -2.40 0.671 > 5 & HESE -0.000 +172.72 12.495 > 6 & HECE2 -0.000 +40.67 10.499 > 7 NPAE +0.001 -13.83 52.958 > 8 CA -0.001 -0.67 53.062 > 9 & HSCE3 +0.000 -1.27 8.655 > 10 AN -0.000 -9.56 0.365 > 11 AW +0.000 +10.44 0.330 > 12 & HWCA2 -0.000 +0.27 0.466 > 13 TF +0.001 +112.37 5.564 > 14 PZZ1 -0.000 +10.67 12.111 >=20 > Sky RMS =3D 3.42 > Popn SD =3D 3.59 >=20 > Mark Klaene > Site Operations Manager=20 > Apache Point Observatory =20 > Work : 575 437 6822 > Direct : 575-437-0436 > PO Box 59 fax: 575 434 5555 > 2001 Apache Point Rd cell: 575 430 0210 > Sunspot NM 88349-0059 web: www.apo.nmsu.edu = =20 >=20 >=20 >=20 >=20 >=20 > ____________________________________________________________ > Logs of Nightly APO Activities > = https://trac.sdss.org/tracmailman/browser/private/sloan-25mlog/index.html --Apple-Mail=_D040F9B9-76DB-461D-B018-3AE474083FC1 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=utf-8 We ran some tests with 20 point data sets and found that the = fittingrms will be larger if we allow rotator move = during the data collection.However, with the = rotator axis, we had pointing models with fitting rmsat ~ 2 arc seconds, but the telescope could be off set by = more than 15arc seconds when goto field (before = last month's twist on the instrumentblock, we saw = 30+ arc-seond pointing occasionally). We know that star images will fall on = slightly different spots on CCD fordifferent = rotator angle (due to the rotator center doesn=E2=80=99t exactly overlap = withthe instrument center) for the same AZ/Alt = coordinates. Because the actual point errors = are much large than the point model fitting rms, I feel that = partof the inconsistency might be due to leaving = the rotator axis out in slews ofthe data = collection. Therefore, I decided to let rotator move (slews will = belonger and more out-close-loop to deal with) = last night. The hope is thateven the fitting rms = is worse, the actual pointing may be better. Last night, we observed two fields after the new point model, both has = guiding stars on gotoField. It is a good = sign, but only the time will tell.I think that all fitting are on new = data. The fitting in the first step depends onthe = input model, but the final model should not. In the past, we always = fit pointing model data with the same = empirical terms. By the Tpoint documents,some of = these terms are not significant, should not be there (they may = reducefitting rms, but could actually increase = pointing error). Last month, we usea different = fitting approach in aiming to pick up =E2=80=9Cright=E2=80=9D empirical = terms (except formust have physical terms), which = seems worked well. With the point model fromthe new fitting process, we see the results in the = log.If I fit = the last night data with the =E2=80=9Cold=E2=80=9D fitting process and = use the pointing modelfrom the =E2=80=9Cold=E2=80=9D= fitting process, we do see changes in different steps as show = below:* = fix* fit coeff change = value sigma 1 IE = = -2.64 2 IA = +96.44 3 PZZ1 = +123.40 = 4 AN = -9.75 5 AW = +14.40 = 6 CA = -21.26 7 HWCA2 = -0.81 = 8 HWSA2 = +0.29 9 HNCA1 = -0.82 10 HNSA1 = +0.53 11 = ACEC = +0.29 12 ACES = -1.39 13 TX = = +3.46Sky RMS = =3D 8.62Popn SD =3D 8.62* use ie ia pzz1 an aw = ca* fit coeff change = value sigma 1 IE = -3.786 -6.43 = 0.744 2 IA = +1.008 +97.45 1.216 = 3 PZZ1 -0.156 +123.25 = 0.937 4 AN -0.118 = -9.87 0.347 5 AW = -4.030 +10.37 = 0.347 6 CA = +5.230 -16.03 0.825 = 7 HWCA2 = -0.81 8 HWSA2 = +0.29 9 HNCA1 = -0.82 10 = HNSA1 = +0.53 11 ACEC = +0.29 12 ACES = -1.39 13 = TX = +3.46Sky RMS = =3D 3.69Popn SD =3D 3.77Observation #54 is a weak outlier = candidate.* = use* fit coeff change = value sigma 1 IE = -2.563 -8.99 = 0.963 2 IA = -0.529 +96.92 1.154 = 3 PZZ1 -9.502 +113.75 = 3.391 4 AN +0.374 = -9.49 0.408 5 AW = +0.056 +10.43 = 0.410 6 CA = +0.653 -15.38 0.856 = 7 HWCA2 +1.071 = +0.26 0.542 8 HWSA2 = -0.483 -0.19 = 0.540 9 HNCA1 +1.617 = +0.80 0.642 10 HNSA1 = -0.842 -0.31 = 0.643 11 ACEC -0.316 = -0.02 0.773 12 ACES = -1.142 -2.54 = 0.778 13 TX +4.777 = +8.24 1.664Sky RMS =3D 3.42Popn SD = =3D 3.58Observation #41 is a weak outlier = candidate.If I fit the new data with auto fitting process by using = thepointing = model from the =E2=80=9Cold=E2=80=9D process, I got: coeff change = value sigma = 1 IA +0.000 = +97.12 1.085 = 2 IE -0.000 = -8.65 0.873 = 3 & ACES -0.000 = -2.46 0.647 = 4 & HECASE5 +0.000 +1.67 = 0.563 = 5 & HECA2CE +0.000 +1.29 = 0.563 = 6 NPAE = +0.00 = 7 CA -0.000 = -15.62 0.736 = 8 AN +0.000 = -9.49 0.352 = 9 AW -0.000 = +10.41 0.310 10 = & PZZ1 -0.000 +113.13 = 3.271 11 = TX +0.000 = +8.60 1.606Sky RMS = =3D 3.29Popn SD = =3D 3.41Observation #24 is a weak outlier = candidate.If I fit the new data = set using the new fitting process and use the point model that = weused last month (from the new fitting process), = I got a very similar model. = coeff change = value sigma 1 IA = +0.000 +97.12 = 1.085 2 IE +0.000 = -9.02 1.328 3 & ACES = -0.000 -2.46 0.647 = 4 & HECASE5 +0.000 +1.69 = 0.563 5 & HECA2CE +0.000 = +1.29 0.563 6 & HECE3 = +0.000 -5.10 2.293 = 7 NPAE = +0.00 8 CA = -0.000 -15.62 0.736 = 9 AN +0.000 = -9.48 0.353 10 AW = +0.000 +10.40 = 0.310 11 TF -0.000 = +116.34 6.365 12 TX = +0.000 +18.52 = 3.405Sky RMS = =3D 3.29Popn SD =3D 3.42Observation #24 is a weak outlier = candidate.The new fitting process results slightly better = fitting rms withless= empirical terms. I think it=E2=80=99s better. So I have = implementedmy = new model. I = plan to run the way for a couple of more models, if it works = well,I will update = the relevant documents.On Mar 20, 2019, at 10:20 AM, = Mark Klaene wrote:On Mar 20, 2019, at 6:29 AM, = dmbiz@apo.nmsu.edu = wrote: = &n= bsp; APACHE POINT = OBSERVATORY SDSS 2.5M = &n= bsp; &nbs= p; OBSERVING LOG = &n= bsp; &nbs= p;Tuesday March 19, 2019 = &n= bsp; &nbs= p; (SJD 58562)could you please explain the pointing model = data 1. = is one of these and if so which one is new data but old = fit ?2. Which = is new data with new fit and resulting rms errors?3. why do none of these = show real change in value ? mark We got clouds from ~6:40 to 7:30 UT, which = affected the signal for some stars(~20% of all). Seeing = flared to 3.5" during the last several stars. * indat ptdata_2019-03-20T03:28:38.5.dat* = fix* fit coeff = change value = sigma 1 = IA = &n= bsp; +88.01 2 = IE = &n= bsp; -140.39 3 = & HASA2 = &n= bsp; +0.40 4 & ACES = &n= bsp; -2.40 5 = & HESE = &n= bsp; +172.72 6 & HECE2 = &n= bsp; +40.67 7 = NPAE = &n= bsp; -13.83 8 = CA = &n= bsp; -0.67 9 = & HSCE3 = &n= bsp; -1.2710 AN = &n= bsp; -9.5611 = AW = &n= bsp; +10.4412 = & HWCA2 = &n= bsp; +0.2713 TF = &n= bsp; +112.3714 = PZZ1 = &n= bsp; +10.67Sky RMS =3D = 3.42Popn SD =3D 3.42* use ie ia pzz1 an aw ca* = fit coeff= change = value sigma 1 IA = +0.000 = +88.01 1.125 2 = IE = -0.000 = -140.39 0.688 3 = & HASA2 = &n= bsp; +0.40 4 & ACES = &n= bsp; -2.40 5 = & HESE = &n= bsp; +172.72 6 & HECE2 = &n= bsp; +40.67 7 = NPAE = &n= bsp; -13.83 8 = CA = -0.000 = -0.67 0.763 9 & HSCE3 = &n= bsp; -1.2710 AN = -0.000 = -9.56 0.32111 = AW = -0.000 = +10.44 0.32112 = & HWCA2 = &n= bsp; +0.2713 TF = &n= bsp; +112.3714 = PZZ1 -0.000 = +10.67 0.867Sky RMS =3D 3.42Popn SD =3D = 3.49Observation #66 is a weak = outlier candidate.* use* = fit coeff= change = value sigma 1 IA = +0.000 = +88.01 30.967 2 = IE = +0.000 = -140.39 4.549 3 = & HASA2 -0.000 = +0.40 0.593 4 & ACES = +0.000 -2.40 = 0.671 5 & HESE = -0.000 +172.72 = 12.495 6 & HECE2 = -0.000 +40.67 = 10.499 7 NPAE = +0.001 -13.83 = 52.958 8 CA = -0.001 = -0.67 53.062 9 = & HSCE3 +0.000 = -1.27 8.65510 = AN = -0.000 = -9.56 0.36511 = AW = +0.000 = +10.44 0.33012 = & HWCA2 -0.000 = +0.27 0.46613 = TF = +0.001 = +112.37 5.56414 = PZZ1 -0.000 = +10.67 12.111Sky RMS =3D 3.42Popn SD =3D = 3.59 Mark KlaeneSite Operations = Manager Apache Point Observatory = Work : 575 437 = 6822Direct : 575-437-0436PO Box 59 = = = fax: 575 434 55552001 Apache Point Rd = cell: 575 430 = 0210Sunspot NM 88349-0059 = web: www.apo.nmsu.edu _________________________________________________________= ___Logs of Nightly APO Activities https://trac.sdss.org/tracmailman/browser/private/sloan-25mlog/= index.html= --Apple-Mail=_D040F9B9-76DB-461D-B018-3AE474083FC1--