Difference between revisions of "Double Regge Exchange Meeting"
(→March 1 2021) |
|||
(44 intermediate revisions by 8 users not shown) | |||
Line 5: | Line 5: | ||
* 10:30am: F326/327 | * 10:30am: F326/327 | ||
− | + | Zoom (physics meeting link) - https://jlab-org.zoomgov.com/j/1619984443?pwd=RVVnWFdYTEFVZ3RXT3E3K1BqeHRnZz09 | |
+ | |||
+ | |||
+ | =Ambiguities= | ||
+ | Vincent - [https://www.overleaf.com/5817438755qhhdwnrttdqz Ambiguities overleaf] | ||
+ | |||
+ | Wyatt - [https://www.overleaf.com/read/nxxkscbchpsf Ambiguities overleaf pt2] | ||
=Google Group= | =Google Group= | ||
gluex-jpac-double-regge-discussion@googlegroups.com | gluex-jpac-double-regge-discussion@googlegroups.com | ||
+ | |||
+ | = May 12 2022 = | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: Derek Glazier was also looking into ambiguities. Vincent previously showed proof that ambiguities exist for a very small unpolarized waveset of like D-waves with M=0,1. Derek - with polarization these ambiguities disappear. Seems like mathematical ambiguities basically do not exist when polarization is included. Mariana is working on analyzing more statistics for the etaprimepi0 channel. Seems like fit-related ambiguities are cropping up which may be due to low data statistics. Mariana is also working on the event selections for etaprimepi-delta++ and will send the data over to Malte so he can analyze the partial waves with an extra factor to incorporate the delta++ lineshape. | ||
+ | |||
+ | |||
+ | = April 11 2022 = | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | No presentations. Discussion about the scale factor difference we see between data and theory ( theory needs to be scaled down by 10x ). The original theory predictions did not include any dynamical energy dependence. When you include some energy dependence you get slightly closer to the scale of the data, maybe within 2x. Malte and Lawrence will meet and try to make sure the measured cross sections are calculated correctly. For ambiguities, not much has changed. They believe that non-trivial ambiguities only exist for small wavesets. For "larger" waveset (maybe greater than 2 waves) the ambiguities dissappear. | ||
+ | |||
+ | = March 28 2022 = | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Lawrence showed fit comparisons to Vincent's theory curves that include the a2(1700) mass region (up to 1.8 GeV). The general message did not change, the t-dependence of the waves are similar. The cross sections for the a2(1320) is still about an order of magnitude smaller than vincent's predictions. The extraced a2(1700) cross sections are about an order of magnitude larger than the a2(1320) cross sections putting it on a similar scale as the theory. Zach showed an update for the q-factors where he calculated two q-factors (one for eta(') and another for the omega), using the same set of neighbors, and then finally multiplying the two together. This is in contrast to the conditional q-factor people where the second q-factor has to be conditioned on the first q-factor. Zach is going to JLab (next week I think) and will gather a group of people to try and sort this out. | ||
+ | |||
+ | = March 16 2022 = | ||
+ | * Experimental Update | ||
+ | ** Lawrence [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5527 [Slides]] | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Lawrence showed comparisons between mass independent (MI) fits, hybrid fits and theory predictions ( from Vincent's "TMD" model ). Comparison of MI fits and hybrids suggest that the inclusion of an a2(1700) Breit-wigner greatly improves the agreement between the two fit procedures. Comparing the hybrid fits to theory show significant differences. In particular, the evolution of D-waves (M-projection and reflectivity) as a function of t show differences. One example, is that the D2+ wave appear to completely vanish in the middle t-bin whereas the theory curves do not. Vincent remarks that the model was only fit on the total cross sections of CLAS data so an estimate of the D-waves is probably not so correct. The current measured differential cross section is on the order of ~10 nb but the cross sections predicted in theory is about an order of magnitude larger. Vincent's predictions for the CLAS data also needed to be scaled by ~60% but scaling down to ~10% is quite unexpected. Extracted partiy asymmetries suggest natural exchange dominant, which favors the TMD model predictions as opposed to the minimal model. For the a2(1700) it was suggested that its probably ok to fit over a wider range (maybe up to 2GeV) since Vincent does not believe the double Regge process to be strong until >2GeV. Very preliminary estimates of the differential cross section for the a2(1700) is ~100nb. It is expected that the a2(1320) and a2(1700) cross sections should be at the same scale or smaller, not larger. Suggests the current fits might be picking up garbage and over-estimating the contributions of the a2(1700). | ||
+ | |||
+ | Vincent mentioned that Wyatt, et. al is seeing that there is probably no ambiguities in the moments for "larger" wavesets. For wavesets with 2 waves with the same M-projection ambiguities might appear. For our waveset it is currently not expected to show ambiguous solutions. I think, this was the reasoning. | ||
+ | |||
+ | |||
+ | = Feb 14 2022 = | ||
+ | * Experimental Update | ||
+ | ** Zach | ||
+ | ** Rebecca [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5453 [Slides]] | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Zach gave an update on the q-factor analysis of eta(')pi0 channels, where the eta(') goes through the charged pions decay. The subtraction of the M(eta) looks good, but was potentially using too many neighbors (900). Looking at the lambda distributions looks ok'ish. The omega backgrounds should have a linearly increasing distribution in lambda, whereas etas should be flat. The q-value signal still looks linearly increasing, but does appear to remove a lot of omegas. This discrepancy could be due to the amount of neighbors. Maybe selecting on M(eta) after calculation of the q-factors can help. For the eta'pi channel, the subtraction for M(eta') looks like it has a bump underneath the peak. Not sure if it is real or not, but there were some suggestions about modifying the phase space. Zach also talked about some hybrid fits he performed which were showing significant execution time differences but still have an error with -nan likelihoods | ||
+ | |||
+ | Rebecca showed some plots for the double regge fits that were performed with Vincent's model. There were some unexpected features in phi_GJ which seems to have been caused by baryons. Making a selection on the M(pi0p) and M(etap) causes a significant asymmetry in the cosTheta_GJ distribution. This asymmetry could be caused by the change in acceptance due to these extra selections, so a suggestion was made to plot the acceptance corrected cosTheta_GJ distribution. A strong asymmetry might appear with a strong odd wave which we do not expect; the pi1 in etapi0 is expected to be small. | ||
+ | |||
+ | = Jan 31 2022 = | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Discussion about the double Regge fits that Rebecca and Vincent are working on. There was some discrepancy between the data and fits that the fit was not describing. One of them were showing dips in an angle whereas the other did not. They are considering adding a S-wave to include some more intensity towards 0 cosTheta. Double regge are strongly peaking and does not produce significant intensity towards the center whereas baryon production could. Would be interesting to look at how baryons populate the spectrum, maybe by looking at simulations (which have been looked at with the Delta++). The double Regge process is invariant mass dependent and is stronger at higher masses. | ||
+ | |||
+ | |||
+ | = Jan 17 2022 = | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Overview of the progress and direction for the eta(')pi studies. An updated wiki can be found at [https://halldweb.jlab.org/wiki-private/index.php/Eta(%27)_Pi_Working_Group EtaPi studies] where you can find yields of different decay modes / channels for the the etapi, the double regge meeting page, various fit studies, and a method of obtaining matching t-slopes (between data and mc). A detailed plan has been developed to understand fit systematics (also linked in the wiki) and work has been divided between various two-pseudoscalar analyzers. There is quite a lot of work to do! On the JPAC side, they (Wyatt) are working out the ambiguities and are going to make an attempt at getting something concrete this week. Future biweekly meetings will start at 12PM Eastern time (Adam will be teaching until 12:30) | ||
+ | |||
+ | = Dec 6 2021= | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | No updates from the experimental side. JPAC has been busy with a review paper, Wyatt has been making some progress on the ambiguities finder which is basically contructing some polynomials with a given waveset and finding the zeros. The polynomials are waveset dependent. There was a discussion about the theory model fits of the the double Regge data which was showing unexpected zeros in the fit. Adam requested to see Rebeccas fits so that they could think about how to modify the data. The zeros are due to the Gamma function in the model. Lawrence/Colin is working on the double Regge analysis note to get the data out with or without a theory model. Would be nice to have a theory model though, but we should get this moving. | ||
+ | |||
+ | |||
+ | = Nov 8 2021= | ||
+ | * Experimental Update | ||
+ | ** Lawrence - [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5356 Working on ambiguities example] | ||
+ | ** Gabriel - [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5357 KsKs update] | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Productive discussions. Lawrence is looking into mathematical ambiguities for a worked out example {S,D0,D1} waveset. We can do many randomized fits and see there are two easily separable continuous set of solutions. If we pick one set of solutions we can reliably calculate the other set. Will continue working with JPAC in understanding these ambiguities, Wyatt is doing some good work in developing the algebra/algorithm. Malte talked about a hybrid/piece-wise fits (basically mass independent fits + breit-wigner for some waves) which seem to be pulling out the correct contributions when looking at his k-matrix MC. When looking at data, the fits were also producing expected results in the a2 mass region which acts as a strong anchor/indicator. There was some discussion about an upper limit on the number of ambiguities which should follow something like 2^(n-1) or 2^n-n or something like that. It's complicated by polarization and the target. | ||
+ | |||
+ | Gabriel talked about the KsKs analysis and what he wants to present at SESAPS. He showed an input/output study where he simualted some breit-wigners and reconstructed and fitted to partial waves. Got the go ahead to present the result. Some discussion about comparing the relative yields/cross sections for the f2 and a2 between the KsKs and etapi channels. Malte mentioned he wrote up the flatte distribution in amptools so we can generate and fit to that now which allows a better handling of the KsKs threshold behavior. Vincent had some predictions for f2/a2 also and suggested it would be interesting to look at the t-dependence of the f2. In etapi->a2 there seemed to be a dip in the cross section which is not expected to happen in the f2 | ||
+ | |||
+ | |||
+ | = October 25 2021= | ||
+ | * Experimental Update | ||
+ | ** Rebecca - Double regge fits | ||
+ | * Theory Update | ||
+ | ** Vincent - [https://www.overleaf.com/5817438755qhhdwnrttdqz Ambiguities overleaf] | ||
+ | |||
+ | Minutes: | ||
+ | Mostly verbal updates. Vincent talked over the ambiguities overleaf (linked above) and how the conditions for ambiguities can be derived for these two pseudoscalar Zlm amplitudes. A simple example for the m=0,1 case is already done. Wyatt, a theory student working with JPAC, is looking into finding the conditions for ambiguities in the full waveset (S + all 5 D-waves), but anyone who wants to work on it is welcome. Lawrence is working on generating a larger flat/acceptance (with proper production kinematics) MC sample which can be used for the PWA fits and the double Regge studies. Malte is working on hybrid fits in amptools where one can define a breit wigner for the production of the a2 and keep other waves free of any mass dependence inputs. Rebecca has a working version of fitting Vincent's double Regge model using GPUs. There are 5 parameters in the fit but it was taking a very long time. The converged parameters were scattered suggesting the need for more constraints. Vincent suggested to restrict the (s?) parameter to around 1 GeV. There was some discussion about some differences between the data and the fit results around 2.4 GeV, both Vincent and Rebecca are looking into this. | ||
+ | |||
+ | = September 27 2021= | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Mainly revolved around a discussion on what will go into the future a2 publication. We will focus on the cross sections for the a2 for both the charged and neutral channels. To get these cross sections we need to do a PWA. Matt/Malte is thinking about constructing some piece-wise amptools fits to include a breit-wigner lineshape for the a2 for the different m-projections. One thing to keep in mind is there appears to be a bowing the GJ vs Mpi0eta distribution. There doesn't seem like there is much we can do to extract the SDMEs since there is a huge S-wave contribution. The SDMEs written in the Exclusive Tensor Meson Production paper is only valid for isolated resonances. The moments are equivalent to the SDMEs in the case of a single wave. It seems like the best we can do is to extract the moments then which we get for free at this point. Vincent is going to look into making some predictions for the charged channel but he is not so enthusiastic about its performance. Turns out including a pion exchange is very hard and require a lot of tweaking as the pion is rarely a real pion (I think thats what he said). | ||
+ | |||
+ | = September 13 2021= | ||
+ | * Experimental Update | ||
+ | ** Lawrence - [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5266] | ||
+ | ** Malte | ||
+ | ** Zach | ||
+ | ** Rupesh | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Lawrence showed pwa fits of the phase 1 data in the etapi0->4g channel in 5 tbins. Agrees with what we saw before but with a bit more resolution in t. Malte showed the summed D-waves (fitted with S and D waves and showed non-acceptance corrected plots) and saw that the summed D-wave yield at the a2(1320) region followed the expected trend: large, smaller, large. These expectations were from the CLAS study. Lawrence will look at the acceptance corrected summed D-wave yields so we can compare with what Vincent will show. Vincent will look at making plots and values for comparison of the different D-waves in the 5 t bins. Apparently negative reflectivity falls off faster with beam energy. This would have a bigger effect on the etapi- channel as pion exchange / negative reflectivity dominates in the a2 region. | ||
+ | |||
+ | Zach showed progress in extracting some partial waves in the phase 1 data. To convince us the software stack sort of works, he did I/O tests and they do look pretty good. Zach also talked about a cut he has looked into, this cosine theta in the center of mass frame. He saw that it had good separation for the baryons and rest of the signal. He will look into how this compares with the van hove selection which he has done before and see how it shapes the phase space. He will also look into binning in 3 t bins to try and make a comparison to the 4g channel. | ||
+ | |||
+ | Rupesh talked about a multivariate analysis he performed on the large bggen sample we have at GlueX. He showed some plots on how the signal and background distributions looks like. He performed some feature selection by varying the inputs to see how the outputs changed and ended up choosing around 5 ( I think? ). Trained a neural network to separate signal and background and the network does appear to learn, the loss keeps going down. It was mentioned that bggen does not contain the correct angular distributions and does not model the data correctly. So, the study is interesting but it is not clear how to apply this study to data. Suggestion to use another variable to separate signal and background as thrown_topology will not separate alternate combinations. Also, one can look at Shapley values to determine feature importance. | ||
+ | |||
+ | = August 30 2021= | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: Lawrence saw a bug. Somehow in setting up the amptools config file the likelihood was nan or not nan depending on how one sets up the parameters to the breit-wigner function. If one uses the define keyword things works, the likelihood is not nan. If one uses parameters keyword to define the breit-wigner parameters, the likelihood is nan. Matt suggested to go into the breit-wigner code and output the values to see if we are getting any zeros (as zeros can produce likelihood as nan given the log). DataReader can also be modified to only look at the first N elements but need to modify the return code. <b> Turns out it was not a bug. There is a subtlety of amptools requiring you to put square brackets around parameters like [bw_mass] [bw_width] instead of simply bw_mass bw_width </b> | ||
+ | |||
+ | Malte started a discussion on the TMD model suggestions. In the data we were seeing as a function of the 3 t-bins the D2+ wave's intensity goes from large to smaller to medium. So there is this dip which is also seen in the TMD suggestions which can be argued for by looking at the right hand side of Fig 3. of https://journals.aps.org/prd/pdf/10.1103/PhysRevD.102.014003 (where we also see a dip). The D1+ wave is a bit trickier as there doesnt seem to be a dip rather the D1+ becomes apparently larger as t increases. In the middle t-bin (0.3-0.6) the D1+ wave is largest where the TMD model suggests the D1- wave should be largest. There was some discussion about ambiguities again and from the algebraic side, Adam suggested someone to look into that. | ||
+ | |||
+ | = Aug 16 2021= | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Short meeting with verbal updates. Rupesh is looking into using PyPWA and seeding with the Amptools fit results. He is using Q-factors, the original method, with the same selections as before. He is running into some exception error when using PyPWA. Rebecca is working on moving the Double Regge fits to GPU and that is in progress. The results between CPU and GPU differ... Zach fixed a bug in the amptools config file, where he was using the TwoPSAngles instead of the Zlm amplitudes. He is also looking into binning the data in t so he can compare the results to MCL's results (Malte, Colin, Lawrence). Yes, I am going to try and start the trend and give our current results a new name. | ||
+ | |||
+ | = Aug 2 2021= | ||
+ | * Experimental Update | ||
+ | ** Colin | ||
+ | ** Zach | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Colin presented some slides showing what he wants to show for the practice talk for HADRON conference. The big question was to understand the reliability of the fits since there is a very large overall P-wave that is seen over the entire M(pi0eta) range. 3 t bins were shown where PWA fits were performed. The D-waves seem to move between the t bins and so does the S-wave. There is also a large D-wave contribution (summed over all M projections) over the entire mass range. | ||
+ | |||
+ | Zach showed some plots in the same t bins also for both the etapi and eta'pi. He should be able to get some PWA results soon which will be very helpful in understanding the acceptance. In the lowest t-bin Zach's plot showed a bowing of the a2 mass peak into the left. Lawrence sees the same thing in the lowest t-bin also. In the higher t-bins the comparison is a bit hard but for Lawrence he sees a bowing into the right. There were some discussion about what might be source of this. Could be interference with baryon resonances? Could also be a mass dependent interference between different m-projections. A similar thing is seen in dalitz plot analyses. | ||
+ | |||
+ | Some regrouping is needed also. In the past few months we have made a lot of progress in the etapi and omegapi pwa studies. Need to get together to draw up some goals for the near term so we can get the theorists and the experimentals on the same page. Matt came up with some document. | ||
+ | |||
+ | = June 21 2021= | ||
+ | * Experimental Update | ||
+ | ** Colin [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5174] | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Colin gave a nice update on the etapi-Delta++ channel PWA. Looked at the 0.1 to 0.3 t bin with the "nominal/standard" partial wave set and with the TMD suggestions. D-waves are pretty consistent between the two wave sets but the D2 seems to go away with the TMD suggestions. This was for the eta->2g. Colin also looked at eta->3pi and there is good agreement with the 2g channel. The TMD suggestions were made for the positive reflectivity waves only (I think); Colins results in the positive reflectivity waves does show a strong D2 wave compared to the intensity of the other D waves. | ||
+ | |||
+ | GJ frame vs Helicity frame. GJ frame can be seen as in the rest frame of the exchange. In Colin's case we expect and see pion exchange to dominate at low t. If it is a real pion exchange with spin 0 then in the GJ frame only the D1 wave should populate. GJ and Helicity frame are rotations of one another and would shuffle the M projections so if we fit in the helicity frame and there was a real pion exchange then different M projections might populate the partial waves. Lawrence showed some previous fits to generated MC and does indeed see a shuffling of the different M projections when fitting in the GJ frame while simulating in the Helicity frame. | ||
+ | |||
+ | = June 7 2021= | ||
+ | * Experimental Update | ||
+ | ** Lawrence [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5155] | ||
+ | ** Zach [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5161] | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | |||
+ | Lawrence showed some preliminary results on the t-dependence of the partial waves in the etapi0 channel. 0.1 - 0.3 looks similar to what was presented before at the APS meeting / collaboration meeting. 0.3-0.6 shows a shift from the D2 wave to the D1 wave. There is also some strong peak structure in the S-wave here, lower than the a0(1450) but other experiments also have seen something shifted lower. 0.6-1.0 t bin shows a bit of a mixture of D waves but D1 was the strongest still. | ||
+ | |||
+ | Zach presented a bggen study. Here he applys various cuts one after another and checks the final event topologies that are left over. Even though the relative yields of the bggen sample are not the same as it is in data it is still promising to see how the feed-through from other final states are reduced. Again, Zach and Curtis are looking into this Delta Radial cut as an alternative to a pi0pi0 cut (https://halldweb.jlab.org/DocDB/0051/005125/003/ToyMonteCarlo.pdf). | ||
+ | |||
+ | = May 10 2021= | ||
+ | * Experimental Update | ||
+ | ** Zach [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5090] | ||
+ | ** Rupesh [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=5092] | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | |||
+ | Zach and Rupesh showed updates on eta' pi0. Zach looked at mass differences in order to understand the enhancement he sees at threshold. He sees that he can place cuts on the mass difference of the eta and pi0 to enhance his eta' signal without cutting any true eta' events. These cuts need to be optimized and will be studied more. He also sees leakage from the 2pi0 and eta pi0 channels into this threshold region. He will study what cuts he can place on these distributions in an attempt to clean up his distributions. Rupesh recently switched to using eta' pi0 trees with the eta and pi0 masses constrained. He presented his updated plots. Adam has spent some time recently connecting the Double Regge and resonance regions. He has some small details to work out in the calculation but believes the formalism is there. Vincent will work on generating theory curves for the eta pi0 and eta pi- beam asymmetries for different exchanges. | ||
+ | |||
+ | =Apr 26 2021= | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes:<br> | ||
+ | [https://arxiv.org/pdf/2104.10646.pdf JPAC analysis of the double Regge region in COMPASS pwa results]<br> | ||
+ | Adam suggested it would be helpful to have the t2 dependence, the t at bottom vertex, when doing the analysis of the double Regge region in COMPASS data. There was something about the s12 variable and the higher COMPASS energies that makes things difficult also. GlueX energies should help alleviate this problem. Looking at the phi_GJ is also helpful. | ||
+ | |||
+ | Some discussion about baryon resonances. Malte is making a mock model of the etapi0 data and will look into including baryon resonances. Lukasz has a model of baryon resonances which can be made into a generator but I am not sure how would one integrate that into the GlueX software. Would deinitely be helpful to look at though since we know there are baryon production events that remain but just not sure how it populates various distributions. | ||
+ | |||
+ | |||
+ | |||
+ | =Apr 12 2021= | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Lots of random discussions. Lawrence showed preliminary pwa fits to show for APS. a2 produced dominantly in M=2 which is also preferred by models to the tensor mesons as seen in the CLAS data. Non-resonant production should also be in the positive reflectivity since it has the same quantum numbers as the a0 (need to verfiy this). Baryon production, of say the Delta+, might be evenly split between positive and negative reflectivity, probably should not have a bias towards one reflectivity. The reflectivity of a double regge exchange should depend on the naturality of the bottom exchange where the top exchange + pion + eta is lumped into one system. | ||
+ | |||
+ | The current path for double Regge model: Develop the model at high mass where we can see the effects better. Determine some observables to look at. Calculate those observables at high mass and into the resonance region and impose some continuity which we can then restrict/constrain the waveset at lower masses. Maybe comparing the integrated intensity of the waves at low mass to high mass. Rebecca is working on pushing the fits onto a GPU for faster computation. These double regge fits take awhile. | ||
+ | |||
+ | Some discussion about ambiguities and leakages which there is no prescription for to handle just yet. Here are a few things that are important: {combinatorics, resolution, acceptance}. Using the generated data removes all these effects but we can break them in turn to study leakage/ambiguities. Malte is working on generating a more realistic model of the data which we can then break. Mariana has seen a problem with the acceptance correction where S-waves can leak in where there was no generated S-wave. Getting an understanding of the magnitude of these effects are probably the next big hurdle to overcome. | ||
+ | |||
+ | |||
+ | =March 29 2021= | ||
+ | * Experimental Update | ||
+ | ** Zach - | ||
+ | *** gave an update on the strong dip around the b1 mass in his M(etapi) distribution. It was mainly caused by his omega selection which was from the the other pi0 in his reaction. Will look into how to better select away these omegas | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | |||
+ | There was a long discussion about PWA results for the a2, tensor, production. Experimentally we are seeing M=2 projection with positive reflectivity for the neutral case whereas in the charged channel we are seeing dominance of M=1 in the negative reflectivity. It is suggested that this is due to pion exchange causing a favoring of the negative reflectivity in the charged case. Reflectivity=naturality in this case. Also for a real pion exchange, not the reggeized version, the spin is 0 so the helicity of the photon must be passed directly to the resonance. This results in M=1 projection dominating for the charged channel. Pion exchange not possible in neutral case. | ||
+ | |||
+ | As for the different M projections we can look in table 1 of Vincent's paper of the 3 models. The Tensor Meson Dominance model (which was the better fit to f2 data from CLAS, I think) suggests a stronger M=2 coupling for the tensor. TMD also suggests negative M projections are zero as well as the Minimal model. We might also be able to safely fix all the phases of the projections of a given wave to be the same. These two constraints can help lower fit complexity a bit. | ||
+ | |||
+ | The models also suggest some structure in the transfer momentum, t. Figure 2 of the paper shows a dip in the differential cross section of the a2. This dip is caused by omega exchange. omega exchange, I think, is not possible in the charged case (https://halldweb.jlab.org/DocDB/0047/004788/012/exotic.pdf) so we do not expect a dip in this channel but it is possible in the neutral case. | ||
+ | |||
+ | (Vincent's paper) https://journals.aps.org/prd/pdf/10.1103/PhysRevD.102.014003 | ||
+ | |||
+ | |||
+ | =March 15 2021= | ||
+ | * Experimental Update | ||
+ | * Theory Update | ||
+ | |||
+ | Minutes: | ||
+ | Verbal updates from Zach and Rebecca. Zach is trying to understand his acceptance by tracking where particles end up in the detector. Rebecca is working with the fits. Fits seems to take a long time to run so there is a need to cache the results, Rebecca is working on getting that to happen. There was a discussion about fitting the moments directly and using some relations defined in Vincent's paper to restrict some of the amplitudes. In particular, the negative M projections might not be important if a given relation between moments is satisfied. | ||
+ | |||
+ | Sean mentioned that there is a spreadsheet that we can enter the observed yields in different channels by different analyzers. From here we can make some estimates to relate the expected yields of the channels and to understand where we might be losing some events. | ||
+ | https://docs.google.com/spreadsheets/d/1LiiByy4JIxySZIbT32Fbn9nC_xZsvh5F7omKbP8NuRg/edit#gid=0 | ||
=March 1 2021= | =March 1 2021= |
Latest revision as of 10:14, 12 May 2022
Contents
- 1 Location
- 2 Ambiguities
- 3 Google Group
- 4 May 12 2022
- 5 April 11 2022
- 6 March 28 2022
- 7 March 16 2022
- 8 Feb 14 2022
- 9 Jan 31 2022
- 10 Jan 17 2022
- 11 Dec 6 2021
- 12 Nov 8 2021
- 13 October 25 2021
- 14 September 27 2021
- 15 September 13 2021
- 16 August 30 2021
- 17 Aug 16 2021
- 18 Aug 2 2021
- 19 June 21 2021
- 20 June 7 2021
- 21 May 10 2021
- 22 Apr 26 2021
- 23 Apr 12 2021
- 24 March 29 2021
- 25 March 15 2021
- 26 March 1 2021
- 27 February 15 2021
- 28 February 1 2021
- 29 January 18 2021
- 30 December 7 2020
- 31 November 23
- 32 November 9
- 33 October 26
- 34 October 12
- 35 September 27
- 36 September 14
- 37 August 31
- 38 August 17
- 39 August 03
- 40 July 20
- 41 July 6
- 42 June 29
- 43 June 8
- 44 April 27
- 45 April 13
- 46 March 30
- 47 March 16
- 48 March 2020 Agenda
- 49 December 2019 Agenda
- 50 Action Items
Location
Jefferson Lab
- 10:30am: F326/327
Zoom (physics meeting link) - https://jlab-org.zoomgov.com/j/1619984443?pwd=RVVnWFdYTEFVZ3RXT3E3K1BqeHRnZz09
Ambiguities
Vincent - Ambiguities overleaf
Wyatt - Ambiguities overleaf pt2
Google Group
gluex-jpac-double-regge-discussion@googlegroups.com
May 12 2022
- Experimental Update
- Theory Update
Minutes: Derek Glazier was also looking into ambiguities. Vincent previously showed proof that ambiguities exist for a very small unpolarized waveset of like D-waves with M=0,1. Derek - with polarization these ambiguities disappear. Seems like mathematical ambiguities basically do not exist when polarization is included. Mariana is working on analyzing more statistics for the etaprimepi0 channel. Seems like fit-related ambiguities are cropping up which may be due to low data statistics. Mariana is also working on the event selections for etaprimepi-delta++ and will send the data over to Malte so he can analyze the partial waves with an extra factor to incorporate the delta++ lineshape.
April 11 2022
- Experimental Update
- Theory Update
Minutes: No presentations. Discussion about the scale factor difference we see between data and theory ( theory needs to be scaled down by 10x ). The original theory predictions did not include any dynamical energy dependence. When you include some energy dependence you get slightly closer to the scale of the data, maybe within 2x. Malte and Lawrence will meet and try to make sure the measured cross sections are calculated correctly. For ambiguities, not much has changed. They believe that non-trivial ambiguities only exist for small wavesets. For "larger" waveset (maybe greater than 2 waves) the ambiguities dissappear.
March 28 2022
- Experimental Update
- Theory Update
Minutes: Lawrence showed fit comparisons to Vincent's theory curves that include the a2(1700) mass region (up to 1.8 GeV). The general message did not change, the t-dependence of the waves are similar. The cross sections for the a2(1320) is still about an order of magnitude smaller than vincent's predictions. The extraced a2(1700) cross sections are about an order of magnitude larger than the a2(1320) cross sections putting it on a similar scale as the theory. Zach showed an update for the q-factors where he calculated two q-factors (one for eta(') and another for the omega), using the same set of neighbors, and then finally multiplying the two together. This is in contrast to the conditional q-factor people where the second q-factor has to be conditioned on the first q-factor. Zach is going to JLab (next week I think) and will gather a group of people to try and sort this out.
March 16 2022
- Experimental Update
- Lawrence [Slides]
- Theory Update
Minutes: Lawrence showed comparisons between mass independent (MI) fits, hybrid fits and theory predictions ( from Vincent's "TMD" model ). Comparison of MI fits and hybrids suggest that the inclusion of an a2(1700) Breit-wigner greatly improves the agreement between the two fit procedures. Comparing the hybrid fits to theory show significant differences. In particular, the evolution of D-waves (M-projection and reflectivity) as a function of t show differences. One example, is that the D2+ wave appear to completely vanish in the middle t-bin whereas the theory curves do not. Vincent remarks that the model was only fit on the total cross sections of CLAS data so an estimate of the D-waves is probably not so correct. The current measured differential cross section is on the order of ~10 nb but the cross sections predicted in theory is about an order of magnitude larger. Vincent's predictions for the CLAS data also needed to be scaled by ~60% but scaling down to ~10% is quite unexpected. Extracted partiy asymmetries suggest natural exchange dominant, which favors the TMD model predictions as opposed to the minimal model. For the a2(1700) it was suggested that its probably ok to fit over a wider range (maybe up to 2GeV) since Vincent does not believe the double Regge process to be strong until >2GeV. Very preliminary estimates of the differential cross section for the a2(1700) is ~100nb. It is expected that the a2(1320) and a2(1700) cross sections should be at the same scale or smaller, not larger. Suggests the current fits might be picking up garbage and over-estimating the contributions of the a2(1700).
Vincent mentioned that Wyatt, et. al is seeing that there is probably no ambiguities in the moments for "larger" wavesets. For wavesets with 2 waves with the same M-projection ambiguities might appear. For our waveset it is currently not expected to show ambiguous solutions. I think, this was the reasoning.
Feb 14 2022
- Experimental Update
- Zach
- Rebecca [Slides]
- Theory Update
Minutes: Zach gave an update on the q-factor analysis of eta(')pi0 channels, where the eta(') goes through the charged pions decay. The subtraction of the M(eta) looks good, but was potentially using too many neighbors (900). Looking at the lambda distributions looks ok'ish. The omega backgrounds should have a linearly increasing distribution in lambda, whereas etas should be flat. The q-value signal still looks linearly increasing, but does appear to remove a lot of omegas. This discrepancy could be due to the amount of neighbors. Maybe selecting on M(eta) after calculation of the q-factors can help. For the eta'pi channel, the subtraction for M(eta') looks like it has a bump underneath the peak. Not sure if it is real or not, but there were some suggestions about modifying the phase space. Zach also talked about some hybrid fits he performed which were showing significant execution time differences but still have an error with -nan likelihoods
Rebecca showed some plots for the double regge fits that were performed with Vincent's model. There were some unexpected features in phi_GJ which seems to have been caused by baryons. Making a selection on the M(pi0p) and M(etap) causes a significant asymmetry in the cosTheta_GJ distribution. This asymmetry could be caused by the change in acceptance due to these extra selections, so a suggestion was made to plot the acceptance corrected cosTheta_GJ distribution. A strong asymmetry might appear with a strong odd wave which we do not expect; the pi1 in etapi0 is expected to be small.
Jan 31 2022
- Experimental Update
- Theory Update
Minutes: Discussion about the double Regge fits that Rebecca and Vincent are working on. There was some discrepancy between the data and fits that the fit was not describing. One of them were showing dips in an angle whereas the other did not. They are considering adding a S-wave to include some more intensity towards 0 cosTheta. Double regge are strongly peaking and does not produce significant intensity towards the center whereas baryon production could. Would be interesting to look at how baryons populate the spectrum, maybe by looking at simulations (which have been looked at with the Delta++). The double Regge process is invariant mass dependent and is stronger at higher masses.
Jan 17 2022
- Experimental Update
- Theory Update
Minutes: Overview of the progress and direction for the eta(')pi studies. An updated wiki can be found at EtaPi studies where you can find yields of different decay modes / channels for the the etapi, the double regge meeting page, various fit studies, and a method of obtaining matching t-slopes (between data and mc). A detailed plan has been developed to understand fit systematics (also linked in the wiki) and work has been divided between various two-pseudoscalar analyzers. There is quite a lot of work to do! On the JPAC side, they (Wyatt) are working out the ambiguities and are going to make an attempt at getting something concrete this week. Future biweekly meetings will start at 12PM Eastern time (Adam will be teaching until 12:30)
Dec 6 2021
- Experimental Update
- Theory Update
Minutes: No updates from the experimental side. JPAC has been busy with a review paper, Wyatt has been making some progress on the ambiguities finder which is basically contructing some polynomials with a given waveset and finding the zeros. The polynomials are waveset dependent. There was a discussion about the theory model fits of the the double Regge data which was showing unexpected zeros in the fit. Adam requested to see Rebeccas fits so that they could think about how to modify the data. The zeros are due to the Gamma function in the model. Lawrence/Colin is working on the double Regge analysis note to get the data out with or without a theory model. Would be nice to have a theory model though, but we should get this moving.
Nov 8 2021
- Experimental Update
- Lawrence - Working on ambiguities example
- Gabriel - KsKs update
- Theory Update
Minutes: Productive discussions. Lawrence is looking into mathematical ambiguities for a worked out example {S,D0,D1} waveset. We can do many randomized fits and see there are two easily separable continuous set of solutions. If we pick one set of solutions we can reliably calculate the other set. Will continue working with JPAC in understanding these ambiguities, Wyatt is doing some good work in developing the algebra/algorithm. Malte talked about a hybrid/piece-wise fits (basically mass independent fits + breit-wigner for some waves) which seem to be pulling out the correct contributions when looking at his k-matrix MC. When looking at data, the fits were also producing expected results in the a2 mass region which acts as a strong anchor/indicator. There was some discussion about an upper limit on the number of ambiguities which should follow something like 2^(n-1) or 2^n-n or something like that. It's complicated by polarization and the target.
Gabriel talked about the KsKs analysis and what he wants to present at SESAPS. He showed an input/output study where he simualted some breit-wigners and reconstructed and fitted to partial waves. Got the go ahead to present the result. Some discussion about comparing the relative yields/cross sections for the f2 and a2 between the KsKs and etapi channels. Malte mentioned he wrote up the flatte distribution in amptools so we can generate and fit to that now which allows a better handling of the KsKs threshold behavior. Vincent had some predictions for f2/a2 also and suggested it would be interesting to look at the t-dependence of the f2. In etapi->a2 there seemed to be a dip in the cross section which is not expected to happen in the f2
October 25 2021
- Experimental Update
- Rebecca - Double regge fits
- Theory Update
- Vincent - Ambiguities overleaf
Minutes: Mostly verbal updates. Vincent talked over the ambiguities overleaf (linked above) and how the conditions for ambiguities can be derived for these two pseudoscalar Zlm amplitudes. A simple example for the m=0,1 case is already done. Wyatt, a theory student working with JPAC, is looking into finding the conditions for ambiguities in the full waveset (S + all 5 D-waves), but anyone who wants to work on it is welcome. Lawrence is working on generating a larger flat/acceptance (with proper production kinematics) MC sample which can be used for the PWA fits and the double Regge studies. Malte is working on hybrid fits in amptools where one can define a breit wigner for the production of the a2 and keep other waves free of any mass dependence inputs. Rebecca has a working version of fitting Vincent's double Regge model using GPUs. There are 5 parameters in the fit but it was taking a very long time. The converged parameters were scattered suggesting the need for more constraints. Vincent suggested to restrict the (s?) parameter to around 1 GeV. There was some discussion about some differences between the data and the fit results around 2.4 GeV, both Vincent and Rebecca are looking into this.
September 27 2021
- Experimental Update
- Theory Update
Minutes: Mainly revolved around a discussion on what will go into the future a2 publication. We will focus on the cross sections for the a2 for both the charged and neutral channels. To get these cross sections we need to do a PWA. Matt/Malte is thinking about constructing some piece-wise amptools fits to include a breit-wigner lineshape for the a2 for the different m-projections. One thing to keep in mind is there appears to be a bowing the GJ vs Mpi0eta distribution. There doesn't seem like there is much we can do to extract the SDMEs since there is a huge S-wave contribution. The SDMEs written in the Exclusive Tensor Meson Production paper is only valid for isolated resonances. The moments are equivalent to the SDMEs in the case of a single wave. It seems like the best we can do is to extract the moments then which we get for free at this point. Vincent is going to look into making some predictions for the charged channel but he is not so enthusiastic about its performance. Turns out including a pion exchange is very hard and require a lot of tweaking as the pion is rarely a real pion (I think thats what he said).
September 13 2021
- Experimental Update
- Lawrence - [1]
- Malte
- Zach
- Rupesh
- Theory Update
Minutes: Lawrence showed pwa fits of the phase 1 data in the etapi0->4g channel in 5 tbins. Agrees with what we saw before but with a bit more resolution in t. Malte showed the summed D-waves (fitted with S and D waves and showed non-acceptance corrected plots) and saw that the summed D-wave yield at the a2(1320) region followed the expected trend: large, smaller, large. These expectations were from the CLAS study. Lawrence will look at the acceptance corrected summed D-wave yields so we can compare with what Vincent will show. Vincent will look at making plots and values for comparison of the different D-waves in the 5 t bins. Apparently negative reflectivity falls off faster with beam energy. This would have a bigger effect on the etapi- channel as pion exchange / negative reflectivity dominates in the a2 region.
Zach showed progress in extracting some partial waves in the phase 1 data. To convince us the software stack sort of works, he did I/O tests and they do look pretty good. Zach also talked about a cut he has looked into, this cosine theta in the center of mass frame. He saw that it had good separation for the baryons and rest of the signal. He will look into how this compares with the van hove selection which he has done before and see how it shapes the phase space. He will also look into binning in 3 t bins to try and make a comparison to the 4g channel.
Rupesh talked about a multivariate analysis he performed on the large bggen sample we have at GlueX. He showed some plots on how the signal and background distributions looks like. He performed some feature selection by varying the inputs to see how the outputs changed and ended up choosing around 5 ( I think? ). Trained a neural network to separate signal and background and the network does appear to learn, the loss keeps going down. It was mentioned that bggen does not contain the correct angular distributions and does not model the data correctly. So, the study is interesting but it is not clear how to apply this study to data. Suggestion to use another variable to separate signal and background as thrown_topology will not separate alternate combinations. Also, one can look at Shapley values to determine feature importance.
August 30 2021
- Experimental Update
- Theory Update
Minutes: Lawrence saw a bug. Somehow in setting up the amptools config file the likelihood was nan or not nan depending on how one sets up the parameters to the breit-wigner function. If one uses the define keyword things works, the likelihood is not nan. If one uses parameters keyword to define the breit-wigner parameters, the likelihood is nan. Matt suggested to go into the breit-wigner code and output the values to see if we are getting any zeros (as zeros can produce likelihood as nan given the log). DataReader can also be modified to only look at the first N elements but need to modify the return code. Turns out it was not a bug. There is a subtlety of amptools requiring you to put square brackets around parameters like [bw_mass] [bw_width] instead of simply bw_mass bw_width
Malte started a discussion on the TMD model suggestions. In the data we were seeing as a function of the 3 t-bins the D2+ wave's intensity goes from large to smaller to medium. So there is this dip which is also seen in the TMD suggestions which can be argued for by looking at the right hand side of Fig 3. of https://journals.aps.org/prd/pdf/10.1103/PhysRevD.102.014003 (where we also see a dip). The D1+ wave is a bit trickier as there doesnt seem to be a dip rather the D1+ becomes apparently larger as t increases. In the middle t-bin (0.3-0.6) the D1+ wave is largest where the TMD model suggests the D1- wave should be largest. There was some discussion about ambiguities again and from the algebraic side, Adam suggested someone to look into that.
Aug 16 2021
- Experimental Update
- Theory Update
Minutes: Short meeting with verbal updates. Rupesh is looking into using PyPWA and seeding with the Amptools fit results. He is using Q-factors, the original method, with the same selections as before. He is running into some exception error when using PyPWA. Rebecca is working on moving the Double Regge fits to GPU and that is in progress. The results between CPU and GPU differ... Zach fixed a bug in the amptools config file, where he was using the TwoPSAngles instead of the Zlm amplitudes. He is also looking into binning the data in t so he can compare the results to MCL's results (Malte, Colin, Lawrence). Yes, I am going to try and start the trend and give our current results a new name.
Aug 2 2021
- Experimental Update
- Colin
- Zach
- Theory Update
Minutes: Colin presented some slides showing what he wants to show for the practice talk for HADRON conference. The big question was to understand the reliability of the fits since there is a very large overall P-wave that is seen over the entire M(pi0eta) range. 3 t bins were shown where PWA fits were performed. The D-waves seem to move between the t bins and so does the S-wave. There is also a large D-wave contribution (summed over all M projections) over the entire mass range.
Zach showed some plots in the same t bins also for both the etapi and eta'pi. He should be able to get some PWA results soon which will be very helpful in understanding the acceptance. In the lowest t-bin Zach's plot showed a bowing of the a2 mass peak into the left. Lawrence sees the same thing in the lowest t-bin also. In the higher t-bins the comparison is a bit hard but for Lawrence he sees a bowing into the right. There were some discussion about what might be source of this. Could be interference with baryon resonances? Could also be a mass dependent interference between different m-projections. A similar thing is seen in dalitz plot analyses.
Some regrouping is needed also. In the past few months we have made a lot of progress in the etapi and omegapi pwa studies. Need to get together to draw up some goals for the near term so we can get the theorists and the experimentals on the same page. Matt came up with some document.
June 21 2021
- Experimental Update
- Colin [2]
- Theory Update
Minutes: Colin gave a nice update on the etapi-Delta++ channel PWA. Looked at the 0.1 to 0.3 t bin with the "nominal/standard" partial wave set and with the TMD suggestions. D-waves are pretty consistent between the two wave sets but the D2 seems to go away with the TMD suggestions. This was for the eta->2g. Colin also looked at eta->3pi and there is good agreement with the 2g channel. The TMD suggestions were made for the positive reflectivity waves only (I think); Colins results in the positive reflectivity waves does show a strong D2 wave compared to the intensity of the other D waves.
GJ frame vs Helicity frame. GJ frame can be seen as in the rest frame of the exchange. In Colin's case we expect and see pion exchange to dominate at low t. If it is a real pion exchange with spin 0 then in the GJ frame only the D1 wave should populate. GJ and Helicity frame are rotations of one another and would shuffle the M projections so if we fit in the helicity frame and there was a real pion exchange then different M projections might populate the partial waves. Lawrence showed some previous fits to generated MC and does indeed see a shuffling of the different M projections when fitting in the GJ frame while simulating in the Helicity frame.
June 7 2021
Minutes:
Lawrence showed some preliminary results on the t-dependence of the partial waves in the etapi0 channel. 0.1 - 0.3 looks similar to what was presented before at the APS meeting / collaboration meeting. 0.3-0.6 shows a shift from the D2 wave to the D1 wave. There is also some strong peak structure in the S-wave here, lower than the a0(1450) but other experiments also have seen something shifted lower. 0.6-1.0 t bin shows a bit of a mixture of D waves but D1 was the strongest still.
Zach presented a bggen study. Here he applys various cuts one after another and checks the final event topologies that are left over. Even though the relative yields of the bggen sample are not the same as it is in data it is still promising to see how the feed-through from other final states are reduced. Again, Zach and Curtis are looking into this Delta Radial cut as an alternative to a pi0pi0 cut (https://halldweb.jlab.org/DocDB/0051/005125/003/ToyMonteCarlo.pdf).
May 10 2021
Minutes:
Zach and Rupesh showed updates on eta' pi0. Zach looked at mass differences in order to understand the enhancement he sees at threshold. He sees that he can place cuts on the mass difference of the eta and pi0 to enhance his eta' signal without cutting any true eta' events. These cuts need to be optimized and will be studied more. He also sees leakage from the 2pi0 and eta pi0 channels into this threshold region. He will study what cuts he can place on these distributions in an attempt to clean up his distributions. Rupesh recently switched to using eta' pi0 trees with the eta and pi0 masses constrained. He presented his updated plots. Adam has spent some time recently connecting the Double Regge and resonance regions. He has some small details to work out in the calculation but believes the formalism is there. Vincent will work on generating theory curves for the eta pi0 and eta pi- beam asymmetries for different exchanges.
Apr 26 2021
- Experimental Update
- Theory Update
Minutes:
JPAC analysis of the double Regge region in COMPASS pwa results
Adam suggested it would be helpful to have the t2 dependence, the t at bottom vertex, when doing the analysis of the double Regge region in COMPASS data. There was something about the s12 variable and the higher COMPASS energies that makes things difficult also. GlueX energies should help alleviate this problem. Looking at the phi_GJ is also helpful.
Some discussion about baryon resonances. Malte is making a mock model of the etapi0 data and will look into including baryon resonances. Lukasz has a model of baryon resonances which can be made into a generator but I am not sure how would one integrate that into the GlueX software. Would deinitely be helpful to look at though since we know there are baryon production events that remain but just not sure how it populates various distributions.
Apr 12 2021
- Experimental Update
- Theory Update
Minutes: Lots of random discussions. Lawrence showed preliminary pwa fits to show for APS. a2 produced dominantly in M=2 which is also preferred by models to the tensor mesons as seen in the CLAS data. Non-resonant production should also be in the positive reflectivity since it has the same quantum numbers as the a0 (need to verfiy this). Baryon production, of say the Delta+, might be evenly split between positive and negative reflectivity, probably should not have a bias towards one reflectivity. The reflectivity of a double regge exchange should depend on the naturality of the bottom exchange where the top exchange + pion + eta is lumped into one system.
The current path for double Regge model: Develop the model at high mass where we can see the effects better. Determine some observables to look at. Calculate those observables at high mass and into the resonance region and impose some continuity which we can then restrict/constrain the waveset at lower masses. Maybe comparing the integrated intensity of the waves at low mass to high mass. Rebecca is working on pushing the fits onto a GPU for faster computation. These double regge fits take awhile.
Some discussion about ambiguities and leakages which there is no prescription for to handle just yet. Here are a few things that are important: {combinatorics, resolution, acceptance}. Using the generated data removes all these effects but we can break them in turn to study leakage/ambiguities. Malte is working on generating a more realistic model of the data which we can then break. Mariana has seen a problem with the acceptance correction where S-waves can leak in where there was no generated S-wave. Getting an understanding of the magnitude of these effects are probably the next big hurdle to overcome.
March 29 2021
- Experimental Update
- Zach -
- gave an update on the strong dip around the b1 mass in his M(etapi) distribution. It was mainly caused by his omega selection which was from the the other pi0 in his reaction. Will look into how to better select away these omegas
- Zach -
- Theory Update
Minutes:
There was a long discussion about PWA results for the a2, tensor, production. Experimentally we are seeing M=2 projection with positive reflectivity for the neutral case whereas in the charged channel we are seeing dominance of M=1 in the negative reflectivity. It is suggested that this is due to pion exchange causing a favoring of the negative reflectivity in the charged case. Reflectivity=naturality in this case. Also for a real pion exchange, not the reggeized version, the spin is 0 so the helicity of the photon must be passed directly to the resonance. This results in M=1 projection dominating for the charged channel. Pion exchange not possible in neutral case.
As for the different M projections we can look in table 1 of Vincent's paper of the 3 models. The Tensor Meson Dominance model (which was the better fit to f2 data from CLAS, I think) suggests a stronger M=2 coupling for the tensor. TMD also suggests negative M projections are zero as well as the Minimal model. We might also be able to safely fix all the phases of the projections of a given wave to be the same. These two constraints can help lower fit complexity a bit.
The models also suggest some structure in the transfer momentum, t. Figure 2 of the paper shows a dip in the differential cross section of the a2. This dip is caused by omega exchange. omega exchange, I think, is not possible in the charged case (https://halldweb.jlab.org/DocDB/0047/004788/012/exotic.pdf) so we do not expect a dip in this channel but it is possible in the neutral case.
(Vincent's paper) https://journals.aps.org/prd/pdf/10.1103/PhysRevD.102.014003
March 15 2021
- Experimental Update
- Theory Update
Minutes: Verbal updates from Zach and Rebecca. Zach is trying to understand his acceptance by tracking where particles end up in the detector. Rebecca is working with the fits. Fits seems to take a long time to run so there is a need to cache the results, Rebecca is working on getting that to happen. There was a discussion about fitting the moments directly and using some relations defined in Vincent's paper to restrict some of the amplitudes. In particular, the negative M projections might not be important if a given relation between moments is satisfied.
Sean mentioned that there is a spreadsheet that we can enter the observed yields in different channels by different analyzers. From here we can make some estimates to relate the expected yields of the channels and to understand where we might be losing some events. https://docs.google.com/spreadsheets/d/1LiiByy4JIxySZIbT32Fbn9nC_xZsvh5F7omKbP8NuRg/edit#gid=0
March 1 2021
Minutes: A review is happening late march where there will be some discussion of the EtaPi systems. Malte will organize some slides and some plots that should be presented. There is some discussion of constraining the waveset (reflectivity and m-projections) since even for the D-wave there will be 5 m-projections and 2 reflectivites, lots of degrees of freedom. In the EtaPi0 system seems like m=2 is dominant projection in the D-wave, might be expected comparing with previous results from (Belle?). m=1 wave is dominant in EtaPi- probably due to pion exchange. Had a small discussion about ambiguities we are seeing (Malte/Lawrence) given a specific waveset. Malte looked at phases and Lawrence looked at the wave intensities. Lawrence also talked about using the b1 photoproduction cross section to determine an expected yield to gain some insight into what physical processes might leak into the data.
February 15 2021
- Experimental Update
- Theory Update
Meeting canceled due to time conflicts
February 1 2021
Minutes: Lawrence talked about multiphoton backgrounds that can leak into etapi final state. b1(1232) is probably pretty important and also potentially pi0pi0. Next steps are to use photoproduction cross sections to actually obtain an estimate.
Zach showed some comparison plots between the etapi0 where eta->3pi and eta->2g. M(pi0p) and M(etapi) looks very similar. Zach did a more in depth study of the distributions of the baryon resonances by selecting a forward going eta (with GJ cut) within the delta region.
January 18 2021
- Experimental Update
- Zach - [11]
- Theory Update
Minutes: Zach had an update on the asymmetries (fixed the negative asymmetries he was seeing before) for the charged decay of the eta. Things look relatively consistent between his asymmetries and Lawrence's (neutral decay of the eta). The asymmetry vs teta distributions starts of small and saturates around 0.6 whereas in Lawrence's case it is closer to 0.4. Binning in u3 (t proton) the asymmetries increase in both cases. The Mpi0eta distibution looks similar and overlaying the two channels can give a hint at the separation between the a0 and a2 where a b1 bkg can live in the neutral eta decay channel.
Interesting that Zach's acceptance is higher in the forward whereas in the neutral case the opposite is true.
December 7 2020
- Experimental Update
- Theory Update
Minutes: Zach still seeing a negative beam asymmetry (will talk with Lawrence+Colin). Colin showed some acceptance/mass plots with an updated gen_amp generator. The generator now considers Delta++ as the recoiling particle instead of a resonance at the bottom vertex (proton+pi0). The t-slope is now calculated with respect to the Delta++ instead of the recoiling proton of the breit-wigner shaped resonance. One year worth of meetings are done!
November 23
- Experimental Update
- Zach - [12]
- Theory Update
Minutes:
Zach gave an update of the beam asymmetries in the eta(')pi channels for just the 2017 data. Looking to include GlueX phase 1 soon. Saw mostly negative asymmetries.
Vincent gave a nice presentation on the COMPASS double Regge study they have been working on. They generated pseudodata according to the published COMPASS amplitudes+errors which extend up to M(etapi)=3 GeV (consistency checks were performed but would be interesting to hear how they generated the pseudodata). At high energies in this channel natural parity exchange is dominating. 6 double Regge diagrams can contribute (a2/P, a2/f2, f2/P, P/f2, P/P, f2/f2}. COMPASS did their analysis with L<=6 and with this number of waves the fits did not reconstruct the sharp peaks at forward/backward directions - needed more waves (~10). a2 at the top vertex would produce a peak in the forward direction whereas f2/P would produce a peak in the backward direction. The regge trajectories for a2/f2 are the same whereas the pomeron has a flatter slope + larger intercept. By looking at the relative contributions to each of these diagrams extracted from fits to the psuedodata it appears that pomeron exchange is more prevalent in eta'pi channel than it is in the etapi channel. This is sort of expected since eta'pi should have more gluonic content in the wavefunction.
Seems like it would be interesting to look at the cross sections as a function of t which is split into the forward and backward directions. From the slope of the cross section we might be able to determine which diagrams might contribute more in the channels we are looking at in GlueX
November 9
- Experimental Update
- Theory Update
Minutes: Rebecca working on fits. Vincent talked about their progress with the COMPASS Deck analysis and is close to finalizing their results, maybe a week or so to do the bootstrapping for the etaprime channel. In the coming weeks they will give a talk about the physics/interpretations. Vincent has also been working on the theory with a photon beam - spin structure. Takes a little more work to get beam asymmetry predictions since they need normalizations. Zach gave an update on the eta(')pi -> 4gamma pi+ pi reactions. Measured some asymmetries, saw a weird effect in the 2018 data where the yields were ~doubled. Colin was thinking about how baryon decays might change the acceptance but it might be a small order effect anyways.
October 26
No meeting, collaboration meeting and DNP preparations
October 12
- Experimental Update
- Theory Update
Minutes: Lawrence, Mariana, Rebecca, and Zach gave updates on the near term goals. Lawrence's near term goals are focused on backgrounds/MC. Mariana talked about PWA and working on reviving/updating moments code. Rebecca showed a fit and a simulation using Vincent's model. Fits need more data (only 2017 right now, Lawrence is working on that). Simulation was probably not made properly since there was a Gamma function factor that was missing in the parameterization. Zach gave a first look at the etapi0 where eta->3pi channel. Some trouble with GetAsymmetry not returning an asymmetry that is centered at 0. Vincent gave an update on the model he is developing.
September 27
- Experimental Update
- Theory Update
Minutes: Short meeting, no updates. Fixing bugs. Classes/Teaching.
September 14
- Experimental Update
- Theory Update
- Vincent - Discussion of the model being developed
Minutes: Rebecca is currently working on updating and cross checking Vincent's models. Vincent talked about the Double Regge model he is developing. Zach is working on getting beam asymmetry measurements for that are similar to what Colin and Lawrence have shown before. The channel Zach is looking at is gamma p -> eta pi0 p -> pi+ pi- 4gamma p.
August 31
- Experimental Update
- Rebecca - update
- Theory Update
Minutes: Rebecca showed some initial moment fits in the a0 and a2 regions. Vincent suggested some studies to debug the fits better. Sean pointed out that backgrounds from other processes could still exist in these regions. Rebecca is coding up the new Double Regge model from Vincent. Vincent agreed to present more model details at the next meeting.
August 17
- Experimental Update
- Theory Update
- Rebecca - [13]
- Zach - update
Minutes: Rebecca showed the results of her current fits to the 2-3 GeV region with the JPAC model. It was suggested that with a 3-parameter fit, she can try fitting the data in smaller M(pi0eta) bins. Also, it would be worth generating events only in that mass range, then trying to reconstruct them to see what the distribution looks like - the attempt to include lower masses did not work, but it might be an affect of the accept/reject method with the amplitude. Possible improvements to the model will be discussed in some detail in tomorrow's JPAC meeting. Zach showed some initial results from his etapi data. Some yield asymmetries were shown, also mass distributions showing differing contributions of various baryon excitations. Among other studies, he will try to understand why the efficiency for backward eta's is so low by using MC samples.
August 03
- Experimental Update
- Theory Update
July 20
- Experimental Update
- Theory Update
July 6
- Theory Update
- Lawrence - [14]
- Zach
June 29
- Theory Update
- Rebecca - [15]
Minutes: Most of the discussion was centered around JPAC's double Regge model fit to the COMPASS PWA. One limitation in the fit is that the model allows for infinite partial waves but the data is truncated at L=6. A first look at fitting the partial waves is okay, but due to the reason above they are not ideal. Vincent has begun looking at modeling the double Regge process with a photon beam. The hope is we can use the beam asymmetry or other observables to tune this that can not be done by the partial waves. Him and Rebecca have been working on modeling this. Rebecca has been generating data using these amplitudes provided by Vincent. They both see a buildup of events at low eta pi masses. These are not physical events as the model is only valid at large eta pi masses. Eventually, a connection will need to be made between low and high eta pi masses.
We did not get around to updates from Lawrence and Zach due to time constraints. We will meet again Monday July 6th and continue biweekly from there.
June 8
- Colin
- Theory Update
We spent a lot of time discussing our ideas for publication. We are still trying to finalize what observables are useful for the JPAC model. JPAC is currently finishing double Regge things with Compass and will soon begin (this week or next) working on photoproduction. We will then have an idea of how sensitive the model is to different observables and kinematics. Adam suggested we look at ratios (or asymmetries) of fast pi to fast eta events as this is an easy prediction for the model to make. We also discussed making plots for eta' pi and further restricting the eta pi kinematics.
April 27
- Colin -Σ for γ p -> η π- Δ++ Moment Method
- Lawrence - γ p -> η π0 p
- Theory Update
Colin showed results for the beam asymmetry when one uses the cos nphi moment method from Jon Z. The results agree with the nominal extraction method. Lawrence showed some background distributions and provided an update on monte carlo studies. Rebecca has coded up Vincent's amplitudes and has begun generating events. Adam mentioned that JPAC is finishing up the Compass double Regge analysis and hopes to show the results soon. Once they finish Compass, it will be easy to implement for photoproduction.
April 13
- Lawrence - γ p -> η π0 p
- Theory Update
March 30
- Colin - Σ for γ p -> η π- Δ++
- Lawrence - γ p -> η π0 p
- Theory Update
Colin presented updated beam asymmetries that corrected a sign issue. He also showed beam asymmetries in bins of recoil particles (eta Delta++ or pi- Delta++ mass). The results show the the beam asymmetry is independent of these variables. Lawrence presented beam asymmetries for uncertainties that were calculated from bootstrapping. For both analyses, we had a long discussion on why the beam asymmetry is ~0.5 and not 1. For the experimenters, we were tasked with looking at potential backgrounds and playing around with some cuts, such as the eta pi mass. The theorists were tasked with thinking about why the asymmetry was saturating at 0.5, and if there is a good physics reason behind it.
March 16
- Colin - Σ for γ p -> η π- Δ++
- Lawrence - γ p -> η π0 p
- Vincent
Minutes:
- Colin showed asymmetries as a function of t_eta and t_pi-. There appears to be an issue with the sign based off the expectations. Eg t_eta is negative, but should be positive. Colin check his fit function and extraction of the asymmetry with Lawrence and they are using the same function and the phi0 offsets are nearly identical. Therefore, the extraction method does not appear to cause the sign issue. In discussion with Adam and Vincent, the sign may need to be flipped if one has a Delta++ as the recoiling particle. Adam and Vincent will look into this more offline.
- Lawrence showed asymmetries as a function of t_eta and t_pi0. They agree with what is to be expected.
- Vincent showed some slides going over the theory aspects of having 3 particle final states. This needs to be studied more for the kinematics where double Regge production is dominant (high eta pi masses). Hopefully this will give a reason for the sign difference between having a recoil proton and Delta++. It may be that the sign of the asymmetry depends on the product of naturality between the two exchange particles. If this is the case, then having a Delta++ at the bottom vertex would cause a sign flip. This has not been accounted for in Colin's asymmetries.
March 2020 Agenda
- 9:30: Colin - γ p -> η π- Δ++
- 10:00: Lawrence - γ p -> η π0 p
- 10:30: Vincent - [16]
- 11:00: Discussion
- Study secondary exchange vertex. If η is on the upper vetex, look at yields as a function of s_{πp} and t_{p}. Do we see the features of Reggeon exchange at this vertex?
- Calculate beam asymmetries as a function of low t_{π} and t_{η}. Can make some general cuts on the kinematics to reduce potential backgrounds. For example, with a fast π, we can cut to make sure the remaining particles are slow/backwards.
- Vincent and Lukasz will provide the amplitudes for us to fit to out data. Can we extract reasonable values for α?
- Can a prediction be made for ηπ+n? Yes, can use this channel as a check.
- Work on setting up a bi weekly (or monthly) meeting to keep things moving forward.
December 2019 Agenda
- 3:00: Colin -γ p -> η π- Δ++
- 3:20: Lawrence - γ p -> η π0 p : γ p -> η π0 p
- 3:40: Lukasz - Theory Update
- 4:00: Discussion