Difference: Minutes (1 vs. 51)

(1) Reports statistics

Tiara: Showing different delta m distribution attachment. If cutting on dm < 0 then things like iPTF15af would be eliminated. The overall RB score is high but not significantly different for the bogus group.

Suvi: We should leave RB flexible for now.

Sjoert: We need to double check our strategies for saving candidates.

Suvi: I can draft up a scanning guideline and put it on the twiki.

Sjoert: The only things I throw away are things close to bright stars or in crowded stellar fields.

Suvi: Any plan on putting up PS1 cutout on scanning page? Otherwise people are biased toward things within SDSS.

Nadia: They will not put the PS1 cutout on the scanning page because the loading the image is slow. But they could put a link there.

Suvi: Is there a way that we can skip scanning?

Sjoert: We can save everything on the scanning page with code. Or use AMPEL to save things.

(2) AMPEL updates

Rob: On ZTFBH slack there is a new channel called AMPEL live. We are posting summary for Sjoert and Charlotte’s filter on there.

Rob: Contact me to see the SWG filters on github. However, modification won’t be activated until someone here make a git pull.

Rob: Catalog matching is working. There is an example on github.

Suvi: Have you been using the AMPEL output for ranking?

Sjoert: Not possible at this moment, since data are missing from the summary. For example, light curve, ra and decnr...etc.

Rob: It’s not easy to fetch avro packets. There isn’t any plan for this currently. However, in T2 there are different modules and maybe you can integrate your ranking script in there.

*(3) Ranking of interesting transients *

ZTF18aaizyqk (blue rising) ZTF18aaysxzv (blue rising) ZTF18aauaayg (likely nuclear Ia since it has reddened over time.)

Nadia: The limits along detections are likely due to cross-match. Frank will fix the code.

Sources from Sjoert’s list: ZTF18aaxzhzf (rising), ZTF18aaysxzv (rising), ZTF18aaxvsip (red), ZTF18aayxvsn (rising but offnuclear), ZTF18aamnrdd (cool AGN), ZTF18aaufprm (peak now?), ZTF18aajupnt (weird agn), ZTF18aazcuno (AGN spectrum but light curve is unAGN-like.)

Brad: Can we color code the offset plot by filter in case there is any systematic error?

Suvi: Please share your ranking and criteria to the group. We have a DCT run tonight.

(4) SEDm update

Nadia: Monday night lenslet array was replaced, there was no down time. We have been taking data yesterday and today.

Suvi: Need to put together a 45min presentation for Sweden. Will start coordinating soon.

Attendee: Suvi, Brad, Nadia, David, Nathan, Robert, Sjoert, Sara, Tess, Tingting, Tiara

(1) SEDm updates

Nadia: SEDm is doing well. Will test and install the new MLA soon. Testing the throughput in the lab on Monday night. Hopefully the throughput will increase by 2x. Engineering test will tell us if we could go for fainter targets.

Nadia: Finder chart is also on the SEDm page now.

(2) AMPEL updates

Robert: Ampel started running - it is listening to the alert stream, filters are running, candidates are being saved to the database. There are plans to post nightly summary to slack in T3 development. Currently there’s no way to interact directly with the AMPEL db. Integration with the marshal will be the focus for the next few weeks.

(3) RB updates

Charlotte: No obvious improvement in the statistics on the RB training dataset. The values change for different versions of RB, the plan is to publish the total RB distribution for each version so people can change the threshold accordingly.

Nadia: Now you can transfer things to garbage.

(4) Scanning experience

Suvi: Definitely seeing fainter transients now.

Sara: Default scanning query only goes back by an hour instead of a day.

Sjoert: We need to investigate why ZTF18aaxvpsw is only just saved yet has a ~10 day history.

Candidates with two sets of offsets: ZTF18aaogibq, ZTF18aaxwjmp. Need to look into this.

(5) CLAGN progress

Sara: Filter criterion for CLAGN. Sy type 1.8 or greater or LINER in archival spectra. 8 bogus, 2 SN in Sy2. Found 6 CLAGN so far.

(6) SEDm ranking and reports page

Tiara: SEDm targets mostly overlap with RCF.

Minutes for May 30, 2018

Attendee: Suvi, Sara, Tiara, Tingting, Charlotte, Nathan, Rob, Scott, Brad, Ginny, Nadia

New RB score in place since last night.

Charlotte: It has a large training set but statistics are not wildly different from last version.

(1) Public alerts

Suvi: Public alert starts on Monday

Brad: Three brokers signed an agreement with the project (Antares/Belfast/Chile). They will deliver two main capabilities: to deliver the full stream and apply basic cut on bright SN.

(2) AMPEL

Suvi: How should we proceed in the next few weeks with AMPEL.

Rob: Post your filters on AMPEL channel (Slack) or email to me.

(3) SED machine updates Suvi: What is the difference between RCF and BCS?

Nadia: RCF is based on Caltech time. BCS is partnership (proposed by cosmology group). But the targets intersect.

(4) Scanning experience Brad: Got an error message saying that I cannot save a candidate.

Suvi: sgscore = 0.5 for some sources.

Nadia: Focus has been an issue since a few days ago.

Nadia: Upper limits are hidden. It messes up with real-bogus training because everything saved is real.

Suvi: Need a function to clean our reports page by deleting bogus. Scanning page should display non-detections.

Suvi: The person scanning should also do vetting: mention if there is anything notable (e.g. change from SDSS, light curve properties...etc) .

(5) Interesting transients

ZTF18aakxlbf

Nadia: Looks like two stars blended together.

Brad: M-dwarf or CV flares.

ZTF18aardcky

Brad: Broad H alpha emerging. Is this a changing-look AGN? Will keep monitoring.

(6) Ned Stark

Brad: Another epoch of HST is being observed.

(7) Jon Snow

Brad: Too late to trigger HST now. Will wait and see after three months.

(1) Survey status updates

Matthew: Lost most of last week to bad weather. Frank is putting in some upstream filters at IPAC. Nothing will be put out with FWHM > 7. Will get a number of completeness i the next few days. Will likely to have a lot more real than bogus after this. Shri is not happy with how SEDm is performing. We are having a review this afternoon and see what the plan is. There may be a change in utility and availability in SEDm.

Matthew: A lot of hardware problems. We have run out of money to support it. Suvi: Are you going to talk about what other instruments to put into P60.

Matthew: There are a few options.

(2) ZTF18aahqkbt (Jon Snow?)

Tiara: The light curve is quite flat and has been on for ~ a month. Spectrum shows a blue continuum and broad Halpha, Hbeta, HeII features. SDSS spectrum is a starforming galaxy. Swift ToO was triggered and approved so we should get UV data in the next few days. Also triggered P60 to get the g-r color.

Tiara: Showing comparison plots for AGN, SN Ia, SN II, bogus, TDE on the reports page.

Suvi: Change offset cut on the marshal filter to 0.5”.

(3) Ned Stark

Sjoert: Most updates are on Slack. Constant color in Swift. Almost no temperature evolution. Measured galaxy and black hole mass (8e6 Msun). Cannot measure decay rate with good accuracy. Thinking about how to use the rise time. X-ray is fading a bit. X-ray to optical ratio is small, unlike 15oi, which re-brightened in the X-ray after some time.

Nathan: Will you fit the light curve with Mosfit?

Sjoert: Not confident with the results presented in the paper. We should stay empirical or compare with various models. But if we compare models we will have to endorse one of them.

Suvi: If you measure the blackbody radius you can get an estimate on the black hole mass. (assuming it’s outside ISCO)

(4) Interesting transients

ZTF18aatycgd - blue and nuclear on a faint galaxy, but only one detection so far. Matthew thinks this could be a CV. ZTF18aatzygk - blue nuclear transient on a nice galaxy. Not bright enough for SEDm so will be assigned to Brad’s DCT run.

Suvi: Need to send the targets to Christoffer.

Minutes for May 2, 2018

Attendee: Suvi, Sara, Tiara, Sjoert, Charlotte, Brad, Nathan, David, Sara, Scott, Tom, Daniela, Nadia, Matthew

Suvi: supernova team plans to put alerts brighter than 19 mag on TNS. We should propose to publish the full alert stream. Sjoert should draft a proposal.

Suvi: Found another CLAGN. From scanning, there are many sources that have a long history. Matthew suggests this is due to the recent update of RB version.

Sjoert: I found the same on Monday. Before RB update.

Suvi: ZTF18aaqehoc. Why does the upper limits almost look like a light curve?

Brad: Could have something to do with the moon phase.

Brad: I don’t think SNID does galaxy subtraction very well. Should try to run superfit on this.

Brad: Filter exchanger is working.

Suvi: We can do better than 0.8”.

Sjoert: Now we can check the offset from PS1 and from IPAC PSF catalog.

Suvi: Hopefully this will help us get rid of more blue Ia’s.

SEDm strategy

Suvi: How to deal with the overlap for SEDm targets. Is it possible to be spectroscopically complete down to 18.5 mag?

Nadia: Shri has some Caltech time to be used on the bright targets imaged by MSIP.

Suvi: Is it possible to keep SEDm spectra in our program proprietary?

Nadia: Right now it should be.

Suvi: Should we not be looking at MSIP alerts now?

Ned Stark

Suvi: XMM took another epoch.

Brad: Next visit is scheduled for 5/21. 3 other orbits scheduled before it goes behind the sun.

New data policy

Brad: General consensus is that anything found in the commissioning data, no probelm.

Suvi: What about the MSIP points?

Brad: Not sure but anything through 3/15 is proprietary.

Brad: A few P60 imaging on Marshal now.

Brad: We can look at that data. But if we want to trigger follow up, we have to announce it on TNS before we do.

Suvi: Shri told me that we are not allowed to look at the MSIP alerts and Matthew and Brad know this very well. This is surprising to me.

Matthew: The bottom line is we are at the end of the commissioning of the alert stream. Possibly 6/1 is the timescale. If the alerts are coming from partnership time.

Brad: The restriction on MSIP is apparent when started. It is not apparent that we are not allowed to follow up things discovered on MSIP fields.

Suvi: What is the policy for the work we’ve worked on the past two month? Do we need to make the changing-look AGNs we found public?

Matthew: Those are still fine. But going forward we need to be more careful. If we detect something on MSIP alerts, we need to make it public before triggering follow-up.

Suvi: Will we be able to get the full light curve that the public doesn’t have?

Matthew: The ZTF light curve you have is what’s reported in the alert.

Suvi: Why don’t we make the alert stream public.

Matthew: It’s not ready. I don’t know how TNS is going to work with subsequent observations coming with MSIP.

Suvi: We can put alerts on Amazon cloud. During commissioning nobody can look at the alerts.

Matthew: Need to talk to Shri.

Suvi: What do we do with SEDm triggers overlapping.

Peter: The inital concept was 18.3 mag. I look at the sections 18.5 - 19 mag. Say you triggered something at m=19. But if it becomes brighter than 18.5 mag, it goes to the public.

Peter: If you trigger SEDm and classify you don’t have to make it public. But if the bright transient program triggered SEDm, it will become public.

Sjoert: What is stopping us from releasing the stream?

Matthew: We are working with the brokers. RB is not ready. Alert stream is not ready.

Minutes for May 2, 2018

Attendee: Suvi, Sara, Tiara, Sjoert, Charlotte, Brad, Nathan, David, Nadia, Buson

Agenda: (1) General ZTF operation (2) Ned Stark

(1) General ZTF operation

Nadia: Cloudy weather in the past week.

Suvi: How is SED machine doing?

Nadia: SED machine is undersubscribed now. Most targets come from the RCF program. Sometimes observing TNS targets. People should be more generous on assigning targets.

Suvi: Tiara needs to make a page where people can track status of SEDm triggers in our program.

Sjoert: Using the same ranking for DCT follow up.

Sjoert: Interesting source to keep an eye on ZTF18aaltrjw

Nadia: There are planning to deploy the new RB next week. Hopefully it will do better on bright stars.

Suvi: Will we use RB score to judge the quality of subtraction?

Charlotte: The new distribution might show some distinction.

Suvi: Many people really want access to the alert packets. How should we ask for it?

Nadia: Need to talk to Matthew.

Brad: At some point IPAC might archive all the alerts. We can ask about it.

Sjoert: They are already archiving but making it accessible could happen at any time.

Suvi: When will UW be distributing alerts? How the public is getting the alerts?

Nadia: The plan is that we will send things through TNS. Probably only the highest confidence ones. TNS will not have the whole alert packets.

Sjoert: Kowalski is the Caltech database.

Nadia: They don’t want to give people access because the machine could only handle a limited amount of users.

Ned Stark

Brad: We got another epoch of HST. Spectrum has not changed by much. Some low significance absorption (Ly-alpha). We can get a reasonable redshift from this. We should be getting another Gemini spectrum any day now.

Sjoert: Triggered another VLT observation. Have not reduced the data yet. Lack of evolution seems promising for a TDE origin.

Suvi: Excited news from XMM. Detection at 2x10^-14 erg/cm^2/s. Temperature is 0.15keV. No photons above 1keV. ASASSN-14li is ~50 eV.

Suvi: UV-optical temperature is cooler than other TDEs.

Brad: UV-optical spectrum is flat, not well-described by a single temperature blackbody.

Suvi: But X-ray looks thermal.

Brad: That is unsupernova like.

Suvi: This object is seen by other surveys. What do we think on the timing of publishing?

Brad: It will be nice if we could reach a similar timescale with the instrumental papers at the end of May.

Brad: HST data is proprietary for 6 month.

Suvi: X-ray fits in the global properties of TDE. Should be in the discovery paper.

Sjoert: Are the Swift Quick look images and data in the archive different?

Brad: Quality becomes stable when you wait until reprocessing number >= 3

Attendee: Sara, Tiara, Sjoert, Charlotte, Brad, Scott, Robert, Nathan, Nadia, Tom

Scanning dm_by_transient_class

• New annotations are for alerts that also pass the Validation filter. This filter adds more info.

• We made a change to be within 2" of a PS1 source, this removes alerts from artifacts in the IPAC catalog close to the bright stars.

• We also tested, but not implemented, a change to have at least two detection within one night (or 24h). This removes many of the bright galaxies with suspicious detections. Peter suggested to instead require no significant non-detections between two consecutive detection, which would be slightly more conservative. Tiara will do a few more tests and update the filter in next few days.

• We also need to update the filter to account for cases where the PS1 photometry is bad and the nearest sgscore is wrong (see other email / Slack ZTF #general). This is not urgent, but should happen this week.

• Reports page - Indeed the reports page is not efficient. Besides search problems, also the the annotations from the original filter are not repeated here, so it's impossible to retrieve dm.
• Solution part 1: I wrote a script that keep listening to the reports page, push annotations, and then pull all the info. This info is stored it locally (will move it to a dropbox soon). The functions for pulling stuff from the Marshal were written by Taira and me. They are now part of Ampel. This step still doesn't give us dm, because the host magnitude is not listed anywhere on the Marshal.
• Solution part 2: I wrote a script to get the .avro data for all alerts in our Reports page (using somewhat of a hack, but nothing illegal, because I'm removing any Caltech-only data). These are stored in a dropbox, along with the prospectus (cf. #faux-marshal) and the FITS images. I'll add you to this dropbox so you can have a look.

Transients of interest & spectroscopy results

• No updates here. There are some sources on the rise that can soon be triggered in SEMD.

SEDM TDE triggers and results

• No interesting SEDM spectra yet. Given the limitations of the Marshal, Tiara has been selecting sources by hand. She got a few SEDM spectra for dm<-0.3 sources, but there were galaxies. These spectra motivated us to work harder at removing these galaxies.

• We are getting close to automated selection. I only finished part 2 this morning. But was able to write a simple ranking algorithm using Ampel. Two missing ingredient are: - information about current spectra/triggers for this sources (we need to pull this from the Marshal as wel).

- better AGN classifications. The current catalogs are not good enough to identify Seyfert type and should not be used for blind rejection. Sara is working on better identification.

Real/bogus clean up project

• Some second-hand info (via Brad via SN telecon) that ML group is working on updating rb with current low SNR data. However, it seems our stream is already very clean at rb<0.2 (except the bright galaxies, but not clear if these will be handled well by the standard rb).

Who is going to Sweden?

• Peter (likely), Sjoert (very likely), Most of team Ampel (but not Robert)

• Peter gave an update about this Tractor plans. They want to do a hands-on session in June.
• Peter also suggested to look in more detail at the PSF of those bright galaxies with likely bogus subtractions. Charlotte will interact with the ML group to look at this.

Minutes for April 18, 2018 Attendee: Suvi, Sara, Tiara, Sjoert, Charlotte, Brad, Scott, Robert, Nathan, Nadia, Lin Agenda: (1) Scanning experience (2)Transient of interest

(1) Scanning experience

Suvi: Only 5 alerts passed by our filter last night.

Brad: There might have been a problem with the shutter. Not sure how many alerts are generated.

Suvi: Left two of the candidates on the scanning page because I am not sure what to think of them.

Brad: No obvious residual. Plausible. Faint thing on bright galaxy.

Suvi: We are saving both of them now. But need to keep track of them on the saved report.

Sjoert: Need a class for these likely artefacts.

Suvi: Let set a rough timeline on this, say a month.

Brad: Both of them have Sloan spectra with broad H-alpha lines. So likely there is some variability.

Brad: A bug was found in some of the PSF photometry code. Hopefully it should be reduced.

Lin: Observed one event. Photometry seemed consistent until after the spec observation. A big discrepancy suddenly occurred. Let’s keep an eye on this and email Frank if finding anything alike.

*ZTF18aaimsgg*, *ZTF18aabtahy* Sjoert & Lin: Huge scatter in the light curve.

Suvi: Is this a SN?

Sjoert: If this is a subtraction problem in the center of the galaxy. It might be worrisome.

Suvi: Send this to Frank.

Real interesting transient

Suvi: 18aahgakg. Brought up by Steve.

LRIS

Nadia: Mirror broke. Bad weather. Nothing was observed except a noisy spectrum of Ned Stark.

Suvi: Reassign the targets to P200 in two days.

Sara: changing look candidate: ZTF18aahkrpr.

Suvi: Code is “bogus” for bright galaxy subtraction artefact.

Ned Stark

Brad: Spectrum is on slack channel. Continnum is flat. Bright Ly-alpha. A bunch of wiggles that are not identified.

Lin: Overplotted gaia16apd. Broad absorption lines look similar to Ned Stark. Late time behaviour will be more clear.

Sjoert: What is special about the late time?

Lin: Recent paper SN2013hx, 15br look like this. Spectra evolved into typical IIn. Broad + narrow Balmer indicating ejecta and CSM interaction. Conclusion in the paper is that power source for these IIn’s are similar to type I. magnetar ...etc.

Brad: Next epoch is scheduled on 4/25.

Lin: We should see some changes.

Suvi: What’s confusing is that the optical spectra look a lot like 14ae. Could 14ae be a SN?

Nadia: 12500 K is the best fit after correcting for Galactic extinction. Flat.

Sjoert: 14ae has a plateau in the UV at the late time. That’s different from a SN. Mean temperature is 19000K.

Suvi: XMM data is taken. Might provide some more clue to this source.

Sjoert: If there is X-ray could this still be an SN?

Lin: Interaction powered SN is expected to have X-ray. Detection in both X-ray and radio can support the interaction scenario. Late time spectroscopy will tell more.

Lin: Just triggered Spitzer today. First window is in July.

Brad: 4 sigma detection after stacking all Swift epochs. Should have detection in XMM.

Suvi: Second epoch scheduled a month from now.

Lin: 5 epochs are scheduled with Spitzer depending on visibility window.

Sjoert: Should we get NIR spectrum?

Lin: Waiting for good weather.

Sjoert: One more epoch after the first non-detection.

Suvi: What is our imaging program.

Nadia: P60 is our only choice now.

Lin: Suvi should talk to Stuart (board rep) about access to MSIP data.

Attendee: Suvi, Sara, Tiara, Sjoert, Brad, Scott, Robert, Tom, Matthew, Nathan

Action item: (1) Suvi will send out a scanning guideline and email Nadia to get P60 photometry on Ned Stark (2) Tiara will create a SEDm targets ranking/monitoring page. (3) Scanners please send good and bad examples to Ashish (4) Sara will make notes on spectra that turn out to be bogus.

(1) Scanning and follow up plans

Suvi: Should we start saving candidates with automated scripts? Brad: At this point eye scanning is still helpful as real-bogus is still being ironed out. Suvi: How do you tell if something is real especially when the transient is faint? Sjoert: I have been generous with saving a candidate. Suvi: Is our filter cutting on dm now? Tiara: No. It is only being annotated. Suvi: How do we access an individual alert now? Tiara & Sjoert: Haven’t heard back about Kowalski yet. Suvi: Will email scanners to let them know dm is a good discriminator. Suvi: Does something marked as bogus come back later? Tiara: Yes if there is a new alert on the said source. Sjoert: Maybe we shouldn’t contaminate bogus with low dm sources by marking them. Suvi: But Low dm is likely an artifact. Suvi: In the age of AMPEL there is no scanning. There will just be more specific cuts to push things in the marshal.

Sjoert: For SED machine project. Should there be some kind of ranking per night? Brad: MSIP along with the SEDm experiment started on the last day in Pasadena. Tiara: SEDm targets from this week - ZTF18aahzmqo and ZTF18aahqmyq. Sjoert: the subtraction of ZTF18aahzmqo looks a bit problematic. Could be a image subtraction. Tiara: Will create a monitoring page so everyone can comment. Suvi: Sara please make notes when we are taking spectra of junks. There is gonna be bogus subtractions. We should always keep track of follow up to help train our eyes and real-bogus.

Suvi: We should agree that we want to be generous when scanning now. We want all AGN variability, nuclear TDE and SN. We don’t want image artifacts but it’s a bit difficult to tell by the post stamp. Suvi: Please record names of good and bad examples when scanning to Ashish. This is for the zooniverse tutorial.

Matthew: DBSP last Wednesday night. Observed 3 ZTF sources but they are not on the marshal yet. One is Ned Stark. One is the changing-look. Suvi: We had two DCT runs last week. We will coordinate targets for future observing runs (DCT 5/5 DBSP 5/6). Suvi: We have to figure out how to get the RB score more sensible. Matthew: New RB algorithm is in place but was trained on S/N of 15. Performance should be better than before.

(2) Interesting transients

Sjoert: Steve is interested in this source because it’s in a starforming. Could be a SLSN. Not an AGN based on the line ratio. Tiara: Peak mag is only ~-17.7 mag. Not a SLSN. Could be a core-collapse?

(3) Ned Stark

Suvi: Ned Stark is being observed by HST and XMM today. Exciting because there is a X-ray detection with Swift. Brad: Swift monitoring will be back on again on Friday. Suvi: We are in the conundrum that Ned Stark is caught in the new reference images so there won’t be any new alerts. MSIP images are not accessible so we cannot get Ned Stark from this point. We should ask Nadia to continue imaging with P60. But we should have access to the commissioning data before ~3/20.

Matthew: This is an unfortunate event. Suvi: Brad did his own difference imaging to get the g band. What is the status of ATel? Matthew: Just sent them to the board yesterday. Gathering comments now. Suvi: Please show it to us one more time so we can add Avishay in there. Matthew: We are now in alert infrastructure commissioning. If all goes according to plan in 3 weeks time. The plan is to make bright alerts (SN?) public. Suvi: If we find John Snow, can we publish that even if the public cannot see the alerts. Matthew: Need to discuss that. It will depend on a number of other things but we would probably be okay. Matthew: Does Lin want to trigger ToO with Spitzer? Sjoert: Unclear. It’s not observable for Spitzer in a while. Suvi: Is the Pan-Starrs group following this up with LCOGT? (no one knows) Brad: ASASSN group detected it around peak. It will be good for our group to write a ZTF discovery paper. Sjoert: If we have the ASASSN bandpass we can do some rough photometry. Suvi: Will send an email with scanning guideline.

Minutes for April 4, 2018 Attendee: Suvi, Charlotte, Sara, Tiara, Sjoert, Brad, Nathan, Scott, Robert, Tom

Action items: (1) For Growth Marshal, we need a way to remove mistakenly saved sources from the report. (2) Auto-annotation should not be commented by a “robot”. (3) Change the offset plot limits to +/-2.

Brad: Scanned on Tuesday. Most bogus look like ghost but not quite the same as the ones Ginny looked at. People in the supernovae group are inspecting further.

TDE (Ned Stark)

We will continue getting spectra with a cadence of ~ 1 week. Sjoert will check if this coordinate has been observed in the past 3 weeks. We should ask Matthew if we can schedule ZTF on this field. Can request P60 for R and i band if needed. What is the g-r right now? Sjoert is modelling the host galaxy and will add Sloan-u band later.

Rob: Testing AMPEL on dedicated hardware last week. Will As soon as possible. Suvi: will you ingest older alerts? Rob: We have been saving the alerts so yes! Suvi: If we have question about a certain alert, can we query from you? Rob: Not sure how easy it is to access individual alerts. In the future we should be able to rerun things.

Sjoert: We have scripts that can automatically save and comment on the Marshal. We need a linux box to run save/comment script everyday.

Minutes for March 28, 2018

Attendee: Suvi, Charlotte, Sara, Tiara, Sjoert, Matthew, Robert, Scott, Nadia, Dan, Nathan, Brad Agenda: (1) Project update. (2) TDE candidate (3) AMPEL & SEDm update

(1) Project update

Matthew: Switched to g filter last Thursday. Lowered S/N threshold to 5 down to 22.5 mag. Building up the reference in g now. 16000 alerts were generated last night. Will spend another week or so on the g filter. Still waiting for the filter exchange review to find out the timeline. Sorting out DIQ at high airmass. Another engineering night next week. Weather looks clear for the next few nights. SG separator is not in place yet. Time line is end of this week ~ start of next week.

(2) TDE candidate

Brad: Found in the center of a galaxy without star-formation. Discovered by PS1 at the same time of ZTF discovery. Re-run subtraction on images since October. Low significance of detection in February. Swift will continue monitoring in the next couple of weeks. Triggered/ triggering VLA, HST, and XMM. Spectrum looks like ASASSN-14ae. Color evolved redder than the other TDEs. HST orbits will run out in September so we are triggering on this event. This target is too faint in the UV to meet the triggering criteria for the ASASSN HST program. Will update P48 photometry on marshal in the next couple of days. Target is in field 663 (MSIP)/1706 (partnership).

Matthew: Will check with Shri about advertising on ATel.

Sjoert: Reporter from New scientist contacted to find out about the new TDE. Highlighted ZTF: finding it in the commissioning data, things are working...etc. Sjoert: Lin worries that this could be a supernova with Spitzer. There is penalty for false triggers.

Optical spectra plans: Next week - DCT, P200 (Triple spec + DBSP). Gemini time with HST (April 11th or 12th).

Brad: Can try to extrapolate IR brightness from UVOT. Lin: Will wait until seeing the HST result to decide whether triggering Spitzer. Sjoert: By the end of the day we will know if AMI (radio telescope) saw anything. If there is detection, we can trigger VLA right away since the sensitivity is much higher.

Suvi: Will we see the photometry updated on the Growth marshal in the morning if the weather is clear tonight? Nadia: Yes. Matthew: Will find out if this is in MSIP or not.

Sjoert: It will be useful being able to retrieve the alert packets per night. Matthew: Need to talk to Mansi. UW is sending the full stream to the broker. They do not support arbitrary users to download partial packets.

(3) AMPEL & SEDm update

Suvi: Timeline for AMPEL to be listening to UW? Robert: Not sure. Suvi: Sara made some SEDm resolution quasar spectra. Nadia: SEDm spectra from iPTF are quite useful. Nadia: Yesterday’s packet is in Kolowski.

Attendee: Suvi, Charlotte, Sara, Tiara, Sjoert, Matthew, Robert, Scott, Nadia Agenda: (1) Project status update (Matthew) (2) “Nuclear Transients” filter on the Marshal (3) Tests with AMPEL filter (Sjoert) (4) SED machine update (Nadia)

Action items before Caltech meeting: (1) Please volunteer for scanning. (2) Science group dinner on Monday (3/19) 6pm. (3) Keep clicking zooniverse real/bogus page. (4) Please talk on our slack channel if you see anything interesting.

Details

(1) Project status update

Matthew: First proper test on filter exchanger -- 300 exchanges in one go, which is good progress. We are a couple of weeks away from actually using it. Mini survey in the past couple of nights results in ~50k alerts coming through each night. We will be switching the operation cadence in the next few weeks in r and g filters (depending on lunation). We’ve noticed some problem with the CCDs. Thermal component may be contributing to a seeing of 3.5”. You will have all data from the partnership time and alerts from MSIP time. New ATel policy will be sent in the next day or two.

Suvi: Is there a statistics page for nightly P48 operation?

Matthew: IPAC mailing list has that information (how many alerts per night). How many got through filters in marshal is not available at this point.

(2) “Nuclear Transients” filter on the Marshal

Suvi: Our nuclear transient filter is working. We put up a scanning schedule for this week. This is a temporary thing. In the future when we are using the AMPEL filter we won’t rely on visual scanning. Please sign up for one day in a week.

Matthew: RB score is going to be updated. The sgscore is still the old version. Adam Miller is putting up the new version and that should be in place next week.

Suvi: The marshal should also be automatically updating classification from other surveys.

(3) Tests with AMPEL filter

Sjoert: Things look great with astrometry, we may be able to do better than 0.8”. Cross-matching is happening in the (AMPEL) filter except not the full PS1 catalog (which is too big). Should we use the mean properties of previous alerts to make decision? We should test Growth and AMPEL filter with new data.

Sjoert: In the mini survey, the reference images are updated nightly because there are supernova in there.

Matthew: Going forward, more images will be added and we will freeze the reference images at some point so they are updated every 3-6 months. The reference images should improve (with more good quality images) and individual transients shouldn’t dominate in the future.

Sjoert: Will isdiffpos parameter be there later in the survey?

Matthew: Yes.

Sjoert: When the marshal is plotting the light curve, it doesn’t care if the flux is negative. This should be fixed (added to wishlist). Aperture photometry won’t be available? PSF mag will always be for the difference image.

Matthew: Not until public data release.

Sjoert: Will force photometry be performed on MSIP data?

Matthew: No, only for the partnership.

Sjoert: How soon AMPEL is ready to ingest UW alerts?

Robert: We are!

Suvi: We still want to use Growth marshal for visualization and annotating even when filtering with AMPEL.

Suvi: Tiara will rewrite her offset plot code to adapt to the avro packets.

(4) SED Machine

Suvi: Should we only use Growth marshal for SED machine?

Nadia: It’s important to keep everything in the same place.

Nadia: The SEDm stats website is (pharos.caltech.edu). Please email Nadia for password.

Suvi: Is the reduction automated? SNID classification is automated?

Nadia: Reduction sometimes involves human clicking. Yes, As long as SNID works.

Suvi: Is adding templates feasible? Nadia: As long as its in SNID format.

Suvi: Will you need to assign priority?

Nadia: No.

Suvi: Can you kill a queued observation?

Nadia: Yes.

Minutes for February 22, 2018

Attendee: Suvi, Brad, Charlotte, Sara, Rob, Scott, Sjoert, Matthew, Nadia, Tiara Agenda: (1) update on survey operations (Matthew) (2) update on AMPEL (Rob) (3) plans for “debugging” and modifying the GROWTH Marshal filter for the “nuclear transients” science program (4) plans for scanning (5) plans for AMPEL (Sjoert) (6) communication through our new ZTFbh slack channel

Details

(1) update on survey operations

Matthew: g filter went on on Friday. Attempted to get reference. Weathered out over the weekend + forest fire. Back on with g filter tonight. Commissioning the i filter tomorrow. Need to verify the IPAC pipeline with i band filter before the alerts can come through. Some more engineering next week (CCD + filter exchanger). Will know what the Caltech component is next week. We are switching to the new operation (MSIP+partnership+Caltech) the week after. Will be alternating between g and r each night before filter exchanger is on. All alerts (including MSIP) will go into the marshal.

(2) update on AMPEL

Robert: T1 is now able to filter the alert stream. T2 - cross matching with catalogues. Github repo for cross matching software will be made available next couple of days. Not ready to ingest UW alerts yet. T3 response module in place - can push to slack channel. Will integrate with growth marshal.

Suvi: status with cross matching to PS1? Robert: Not ready yet.

Suvi: Status with Growth marshal. Robert: In communication.

Sjoert: Example of a basic filter. Can we make filtering code private? Robert: Yes!

Matthew: NOAO is planning to make Antares available for MSIP alerts in 2018b. Suvi: It seems like brokers are going to be more and more important for interacting with survey products. Are you trying to be the leading broker community?

Matthew: Strictly speaking AMPEL is not a broker because you can not subscribe to it.

Matthew: IPAC stream is the UW stream. There are some server issue now.

Matthew: Alerts were being issue in the past few nights, they just don’t pass the filters (Mansi).

Matthew: Filters only act on new alerts. You cannot go back unless you reingest the alerts.

Suvi: Confusion with testing new filters. Don’t know when that was updated.

Matthew: Transient group is putting together a list of wish list for the marshal. We should do the same and send it to Mansi.

(3) plans for “debugging” and modifying the GROWTH Marshal filter for the “nuclear transients” science program

Suvi: Should we have one filter master to update the filter? (Tiara)

(4) plans for scanning

Sjoert: How many growth channels do we want?

Suvi: What is the purpose of scanning? Real bogus? Science decisions? We should be able to turn off scanning in the future.

Suvi: Need to find out how to integrate with AMPEL? Do we need Growth filters?

Nadia: We should try to reingest the alerts.

(5) plans for AMPEL

Sjoert: Sjoert is the AMPEL master. We will be testing both AMPEL and GROWTH. If our nuclear transient filter finds a Ia. Do we transfer? Or do we just notify the other group?

(6) communication through our new ZTFbh slack channel

Suvi: Are there ways to organize the chats (about a certain transient)? Hashtagging? Matthew: you can!

Nadia: where can we see all the features that we can use in AMPEL to design our filters? Robert: Documentation not public yet. But will put on the twiki.

Suvi: Coming DCT night. No new ZTF nuclear transients.

Matthew: Took 10 spectra last week on a Keck night but nothing exciting. Rg score were 0.5 to 0.7. Some of them may easily match the nuclear transient criteria.

Nadia: SEDm is still down. Dec cable still broken. Fixed marshal bug on SEDM.

Brad: Yes, we can add spectra to SNID library to improve fitting to AGNs.

Suvi: Nadia can be the SEDM master.

Sjoert: How are we going to deal with things turn out to be SNe. Should we be saving the photometric properties somewhere for later use?

Robert and Matthew: AMPEL and Caltech are storing all the alerts somewhere but marshal has a 5 day rule.

Brad: NOAO deadline is end of march. Chandra 3/15.

Suvi: NOAO idea - getting u band with LCOGT.

Attending: Suvi, Tiara, Sjoert, Scott, Buson, Umaa, Matthew, Tom Barlow, Erin, Nathan, Scott Anderson, Brad, Sara

Agenda: (1) Big push to exercise transient object astronomy (2) Report on PSF photometry in the SV images of Stripe 82 (3) Report on difference imaging in the SV images of Stripe 82 (4) Machine Learning Working Group update

Details

(1) Big push to exercise transient object astronomy

Matthew: Will issue the first ATel in the next couple of days. Camera will be back on sky tomorrow night. 2-3 nights of engineering work is expected. Will be obtaining reference images on Sunday/ Monday. Starting from Wednesday will do a 5 to 7 nights run of the survey area. We plan to do 3-4 blocks of 1000 squared deg (overlap with Sloan) and will be carrying out a “mini” survey of 4 x two hour each night in the r band. Alerts will be generated, going through the filtering in the transient Marshal. We have a decent amount of DBSP time starting next week so we can start classifying transients.

Suvi: Are the alerts in the form of alert packets?

Matthew: Yes. in the form of avro alert packets issued by IPAC. Might tentatively use the UW alerts and push results into marshal.

Suvi: We will be eager to getting access to the marshal to help vetting transients.

Sjoert: Can we use previous reference images?

Matthew: We don’t know yet. Don’t know if we have enough reference images with good quality to do subtraction. Might end up with many artifacts if we use the old reference images and that’s why we need to take more reference after making changes to the optics.

Sjoert: So we won’t see many AGNs as the baseline is short.

(2) Report on PSF photometry in the SV images of Stripe 82

Matthew: From the email Frank sent earlier, there is very little scatter around zero while calibrating against PS1 down to 18-18.5 mag. On the faint end, Frank has tracked down the problem, reducing the systematic scatter from 0.05 mag to 0.02 mag. Reprocessing catalogues in S82 at IPAC now.

Sjoert: Is this also expected in aperture photometry.

Matthew: Yes. But it is affected by how PSF is constructed, so improvement might also show up in aperture photometry. Frank also pushed the PSF limit to a fainter limit so we will get more sources (detections).

Suvi: Regarding science validation, do we need to demonstrate image quality/depth of the survey?

Matthew: Ideally yes. But right now all the changes need to go through the change control board.

(3) Report on difference imaging in the SV images of Stripe 82

Sjoert: Image subtraction is reasonably well, except for the bright star. But on a lot of nights, no difference images were created. I did aperture photometry on the center of host galaxy for the difference images. The residuals are centered around 0 but in some cases it deviates by more than 10 sigma. Brad: that is not the right way to quantify significance.

Sjoert: Why only ⅓ of the observations have difference images? Suvi: where is the 5-sigma detection file? Matthew: Email Frank to find out why the pipeline did not run on these images. There are certain quality flags need to pass.

(4) Machine Learning Working Group update

Umaa: In ZTF, more filtering happen before RB applied. So the real-to-bogus ratio is reduced compared to iPTF. RB was deployed on 1/12 for three nights. Performance evaluation is limited right now by the amount of labelled data. We have a high false-negative rate at false positive rate of 1% that is ~36%. Can do better with more labelled data for training set to ensure data is not biased.

Suvi: Are labelled data and training set the same?

Umaa: Some labelled data are reserved as an independent test set. A random grab of the database should be bogus. Right now the real score is on average higher than the bogus score.

Suvi: what is the threshold RB score?

Umaa: Not decided yet. We want a full set of objects to be labelled to do further tests.

Suvi: What does applying filtering mean?

Umaa: A set of filters (cosmic ray...etc) were applied before RB. The majority of the sources are rejected and are not saved.

Suvi: Are you using the same alert packet?

Umaa: We use input from the pipeline.

Sjoert: The feature set seems to cut on the distance to the nearest source. Could they get a lower score?

Umaa: Will do more investigation.

Suvi: Do you have any variable AGNs?

Umaa: No we don’t have classification of the objects.

Umaa: Hopefully will be retraining the classifier biweekly so its improving each time.

Sjoert: In iPTF we found many AGNs. It may be useful to use them as training set for nuclear transients.

Umaa: Many difficulties in doing that. The subtraction were not saved during iPTF. Since the image subtraction algorithm is different, IPAC will need to run ZOGY on iPTF images. Besides, features coming out of ZTF have changed.

Matthew: still some work need to be done (guider CCD, filter exchange, i band...etc). But the operation should start any time in Feb.

Minutes for January 17, 2018

Agenda: (1) Presentation from AMPEL

Next meeting will be in two weeks (1/31)

Details

(1) Alert Management Photometry and Extraction of Light curves (AMPEL)

• AMPEL can be used to perform science calculations including catalogue matching, fitting with template light curves, Gaussian process light curve fitting, photo-z estimation, host galaxy identification...etc. However, most of these won’t be available at the time the survey starts.
• To use AMPEL, one will need to define an AMPEL channel by setting up a selection filter (choice of catalogue(s) to be checked against). Need to identify the calculations needed and how to be notified. On the other hand, one can simply push the calculated values to the Marshal. The team will publish the package on github soon.

Updates from Matthew

• New image subtraction algorithm does a better job than the PTF days, seems optimistic.
• Star-galaxy separator is ready at IPAC.
• End of February UW will be ready to push alert stream.

Attending: Suvi, Sjoert, Charlotte, Tiara, Rob, Brad, Chris, Matthew, Sara, David, Scott

Agenda: (1) SDSS Stripe 82 Science Validation experiment (2) ZTF Engineering Update (3) SEDm White Paper Ideas (4) AGN catalogs (5) Transient alert filtering implementation: Plan of action

Next meeting will be at AAS (1/8-1/12), will send around an email to schedule a time soon.

Details

(1) SDSS Stripe 82 Science Validation experiment (analysis from Sjoert)

Sjoert: Downloaded sextractor catalog and calibrated each epoch using SDSS star catalog.

Matthew: PSF catalog has zero-point in the header. It is calibrated against PS1 instead of SDSS.

Sjoert: tested photometry MAG_APER[3]

Matthew: Compiling limiting magnitude per field correlated to airmass and seeing. Will put results in a database.

Matthew: PSF catalog data are kept in a directory called PSFcat. The sextractor catalog goes fainter than the PSF catalog. When calibrating data there is also color term correction that needs to be accounted for.

Sjoert: Checked astrometry for stars. The rms is ~0.15" at 19th mag.

Sjoert: Should we start doing experiments with the current data or wait until the instrumental specs are close to the real survey?

Matthew: We need to start now.

(2) ZTF Engineering Update (Matthew)

Suvi: What is the schedule for S82 next week?

Matthew: We will keep monitoring S82 as much as possible. Will likely change to g band. The filter exchanger is expected to go up on Jan 9th. There will be a ~one week down time in mid-January and we will be fixing cryostat contamination and re-shimming CCD.

Matthew: Linearity is well characterised now. non-linearity correction. Sara has been looking at the out of band leakage for r band and will be looking at g band when the data is available.

(3) SEDm White Paper Ideas (draft by Tiara)

Tiara: Simulated the SED time needed by applying criteria to last year's g+r experiment. By selecting nuclear transients that have an variability amplitude > 0.5 mag (not caused by known AGN), we end up with ~10 sources per week (no interruption). If we apply color selection then the number can be halved.

Matthew: Do you exclude TDEs in AGN hosts?

Suvi: Yes, we want to be systematic. Tiara should send start putting in text and send the draft around.

(4) AGN catalogs (see Twiki)

Suvi: Accumulated various catalogs with wide areal coverage on the twiki.

Matthew: We should include the half million quasars catalog?

Brad: Does cross-matching happen locally?

Matthew: Caltech purchased a new database machine. All the catalogs could go there and we can perform tasks on that machine.

Suvi: When will it arrive

Matthew: It's been shipped but a couple of weeks will be needed to get it up and running.

(5) Transient alert filtering implementation: Plan of action

Suvi: We need to think about what science experiments, filters, and decision trees are needed.

Suvi: Is there a catalogue for all variable sources in CRTS?

M: There is a catalogue for things that are variable but without classification.

Suvi: So we can query variability history in CRTS. Sjoert is checking the possibility of querying variability history in PS1.

Suvi: Will the SN Ia group report the probability of something being a SN Ia?

Matthew: SN Ia probability might be available in downstream/broker service.

Suvi: How would we communicate with them?

Rob: The information will appear in both the broker and growth marshal. The broker is flexible so everyone can put in different filters.

Suvi: When will we learn more about the broker.

Rob: We could arrange a demo in the ZTFBH telecon.

Suvi: Okay. We can talk more about this at AAS!

Minutes for November 29, 2017

Agenda: (1) Updates on ZTF engineering and science validation (2) Data products from ZTF (3) TDE selection strategies (4) SEDM white paper ideas (5) Proposal updates

Action items:

• Now can start looking at the science validation images (S82) and examine the photometry.
• What catalogs does this SWG need? Where will we do the cross-matching?
• Tiara will work with Sjoert to draft the SEDM white paper.
• Please update introduction and proposals on the twiki.

Next meeting is in two weeks (12/13)

Details

(1) science validation data

Suvi: Working group at UMd proposed for observing the Stripe 82 fields. We want to use it as a reference to calibrate how well the difference imaging is in galaxy nuclei.

Matthew: We are at the stage of engineering commissioning. Currently there are 3 major concerns with the instrument: 1) linearity (good to 1st order) 2) contamination (largely dependent on observing conditions) 3) mechanical effects from gravity vector causing flexure issues and causing PSF to change.

Matthew: We don't have difference imaging yet but has single epoch imaging pipeline running. Now have ~20 visits to Stripe 82 in g and r filters. Limiting magnitude is close to 21 mag in g (depending on seeing and airmass). High cadence Galactic fields (M31) are now in IPAC archive as well.

Suvi: is the photometry in agreement with SDSS? Matthew: It looked reasonably good across the magnitude range now and can get better.

Nadia: I proposed to look at the background around bright stars in science validation.

Matthew: No mandate forced photometry service at this stage but might be available in mid-2018.

(2) Data products for this SWG

Matthew: IPAC will provide the light curves. We will keep monitoring old AGN during ZTF partnership time and a student is working on comparing that to predictive models. The light curve will be stored in a DB on a Caltech machine and will be available to the partnership.

Suvi: Where does cross-matching happen between the catalog and alert stream?

Matthew: Some (e.g. PS1) will be done at IPAC and will come with the alert packets. It is possible that additional cross matching will be done at UW and be available with the UW alert distribution system. But UW does not plan to distribute alerts before May.

Suvi: We should work together on how to further filter UW alerts. Need to come up with our catalogs and criteria and decide where the catalogs will sit. The results should be put back into the marshal.

(3) TDE selection strategies

Sjoert: Systematic selection is easy to implement and can be tweaked to get low background rate. But is subject to biases when applying the cuts. One solution is to assign tags to each event based on host and flare properties (similar to SDSS QSO selection) and trigger follow-up observations when the tags accumulate to a certain number. The follow up results can also become a tag (e.g. UV bright). This way we don’t always find the same kind of TDEs. Please send comments.

Lin: what is the scientific drive for selecting r<2" transients in E+A?

Sjoert: recoiled BHs.

(4) SEDM white paper

Suvi: We also need flow diagrams for SEDM white paper. Should we use SEDM for u band imaging?

Nadia: u band is too shallow (ZP~20mag). We should only use it for spectroscopy.

Suvi: Is there any way to get u band at Caltech side?

Nadia: Can do it with P200 on assigned nights.

Matthew: No particular need for AGN science with SEDM.

Nadia: We should prioritize sources with existing SDSS spectra to find changing-look quasars?

Suvi: SEDM white paper deadline is Jan 1st. Tiara needs to sketch the numbers to the white paper in the following week.

(5) Proposal updates

Suvi: Proposals can be found on twiki. Some are already accepted (HST spectroscopy (PI Cenko); VLA (PI van Velzen); 2 XMM programs (PI Gezari and PI Kara). Lin is working on Spitzer DDT proposal for ZTF TDEs.

-- SuviGezari - 27 Nov 2017