drachenberg dnp2017 mastercopy ·...
TRANSCRIPT
Illuminating*QCD*and*Nucleon*Structure*Through*the*Study*of*Hadrons*Within*Jets*
at*STAR
OUTLINE• Introduction• RHIC.and.STAR• Data.and.Models• Near9term.Plans• Summary
James&L.&Drachenbergfor&the&STAR&Collaboration
Lamar&UniversityFall&Meeting&of&the&APS&Division&of&Nuclear&Physics
October&27,&2017
The&study&of&spin&in&particle&physics&has&unlocked&doors&to&a&deeper&understanding&of&nucleon&structure
1Spin.Results.at.STAR.9 Drachenberg
The&Fertile&Field&of&Spin&Physics
The&study&of&spin&in&particle&physics&has&unlocked&doors&to&a&deeper&understanding&of&nucleon&structure
• Helicity
Recent&results&enable&a&better&picture&of&gluon&and&sea3quark&helicitySTAR%data%have%played%a%key%role!
2Spin.Results.at.STAR.9 Drachenberg
The&Fertile&Field&of&Spin&Physics
NNPDF,&arXiv:1702.05077 DSSV,&PRL&113,&012001e.g.&Gluon&Helicity
5
P(↵), where ↵ is the factor by which the uncertainty onthe new data must be rescaled in order for both the priorand the reweighted sets to be consistent with each other(see Eq. (12) in Ref. [31]). If the modal value of ↵ is closeto unity, the new data is consistent with the old, and itsuncertainties have been correctly estimated.
I perform a simultaneous reweighting of the NNPDF-pol1.1 parton set with all the data sets listed in Tab. I. InTab. II, I show the values of the �
2 per data point afterreweighting, �2
rw
/N
dat
, the number of e↵ective replicas,N
e↵
, and the modal value of the P(↵) distribution, h↵i.The corresponding asymmetries, after reweighting withthe new data, are displayed in Figs. 1-2-3, on top of theircounterparts before reweighting.
The value of the �
2 per data point always decreasesafter reweighting. The improvement is marked for theW -boson production data, moderate for the di-jet pro-duction data and only slight for the neutral pion pro-duction data. This is expected, since the first data sethas the smallest uncertainties, among all, in compari-son to the PDF uncertainties on the theoretical predic-tions. This suggests that this data is bringing in a sig-nificant amount of new information. After reweighting,the �2 per data point is of order one for all the new datasets. However, in the case of W -boson and neutral pionproduction asymmetries, these numbers should be takenwith care, because a complete information on correlatedsystematics is not available. This is the reason why thereweighted �
2 is smaller than one for these sets, exceptfor the W
+ production data. In this case, the value ofthe �2 is raised by a sizable contribution coming from thepoint with the largest positron rapidity, which disagreesby about two sigma with the reweighted theoretical pre-diction and the previous STAR measurement from run2010-2011 (see also Fig. 4 in ref [19]).
The number of e↵ective replicas after reweighting de-pends significantly on the data set. The size of thereweighted parton set is about 90% of the originalNNPDFpol1.1 parton set (made of N
rep
= 100 replicas)for di-jet and pion production data, while it is only about30%-40% for W -boson production data. This result re-flects the di↵erent constraining power of the various datasets, which is maximized in the last case. In principle,a prior ensemble with a larger number of replicas shouldthen be needed for the reweighted ensemble to sample theprobability density in the space of PDFs with as much ac-curacy. However this is not relevant here, as reweightedresults only serve to assess the impact of the new data,and are not used to construct a new parton set.
The modal value of the P(↵) distribution is of orderone for all the new data sets. Values of h↵i slightly largerthan one are found for the W -boson production data: inthe case of W�, this is mostly determined by a sizablefluctuation of one data point (around ⌘
e
� ⇠ 0.25) withrespect to the shape of the corresponding asymmetry; inthe case of W+, this is mostly determined by the fourthdata point (around ⌘
e
+ ⇠ 0.75), which, as already noted,disagrees by about two sigma with the reweighted theo-
-0.1
0
0.1
0.2
0.3 x∆Σ(x,µ2)
µ2=10 GeV
0
0.03
0.06 σx∆Σ NNPDFpol1.1NNPDFpol1.1 (rw)
-0.6
-0.3
0
0.3
0.6x∆g(x,µ2)
NNPDFpol1.1NNPDFpol1.1 (rw)
0
0.1
0.2σx∆g
-0.01
0
0.01
0.02
0.03 x∆u-(x,µ2)
0
0.01
0.02
0.01 0.1 1
x
σx∆u-
-0.03
-0.02
-0.01
0
0.01x∆d-(x,µ2)
0.01 0.1 1 0
0.01
0.02
x
σx∆d-
FIG. 4. A comparison of polarized PDFs before and after thesimultaneous reweighting of NNPDFpol1.1 with the data setslisted in Tab. I. From left to right, top to bottom, the singlet,�⌃, the gluon, �g, and the up and down sea quarks, �uand �d, are displayed. Parton distributions are evaluated atµ2 = 10 GeV. Bands represent one-sigma uncertainties; theyare also shown in the lower inset of each panel.
retical prediction. Values of h↵i slightly smaller than oneare found for the neutral pion production data, thus sug-gesting that experimental uncertainties are likely to beoverestimated, possibly because of the lack of a completeinformation on correlations among systematics.In Fig. 4, I compare the polarized PDFs before and af-
ter the simultaneous reweighting of NNPDFpol1.1 withthe data sets listed in Tab. I. From left to right, topto bottom, I show the singlet (for n
f
active flavors),�⌃ =
Pnf
i=1
(�q
i
+�q
i
), the gluon, �g, and the up anddown sea quarks, �u and �d. Parton distributions areevaluated at µ
2 = 10 GeV2. Bands represent one-sigmauncertainties, which are also displayed separately in thelower inset of each panel.The impact of the wew data on the polarized PDFs of
the proton is twofold. On the one hand, it induces a shiftof the PDF central values, as a consequence of the ad-justment to the shape of the corresponding asymmetries.Specifically, the central value of �g increases by about30% of its original value in the region 0.1 . x . 0.2; thecentral value of �u increases by about 25% and that of�d decreases by about 10% approximately in the sameregion of x wher also �g is a↵ected. On the other hand,the new data induces a reduction of the PDF uncertain-ties, as a consequence of the improved precision of thecorresponding asymmetries. For �g and �d such a re-duction is moderate, and not larger than 5% of its orig-inal value; for �u it is fairly more pronounced, around
5
P(↵), where ↵ is the factor by which the uncertainty onthe new data must be rescaled in order for both the priorand the reweighted sets to be consistent with each other(see Eq. (12) in Ref. [31]). If the modal value of ↵ is closeto unity, the new data is consistent with the old, and itsuncertainties have been correctly estimated.
I perform a simultaneous reweighting of the NNPDF-pol1.1 parton set with all the data sets listed in Tab. I. InTab. II, I show the values of the �
2 per data point afterreweighting, �2
rw
/N
dat
, the number of e↵ective replicas,N
e↵
, and the modal value of the P(↵) distribution, h↵i.The corresponding asymmetries, after reweighting withthe new data, are displayed in Figs. 1-2-3, on top of theircounterparts before reweighting.
The value of the �
2 per data point always decreasesafter reweighting. The improvement is marked for theW -boson production data, moderate for the di-jet pro-duction data and only slight for the neutral pion pro-duction data. This is expected, since the first data sethas the smallest uncertainties, among all, in compari-son to the PDF uncertainties on the theoretical predic-tions. This suggests that this data is bringing in a sig-nificant amount of new information. After reweighting,the �2 per data point is of order one for all the new datasets. However, in the case of W -boson and neutral pionproduction asymmetries, these numbers should be takenwith care, because a complete information on correlatedsystematics is not available. This is the reason why thereweighted �
2 is smaller than one for these sets, exceptfor the W
+ production data. In this case, the value ofthe �2 is raised by a sizable contribution coming from thepoint with the largest positron rapidity, which disagreesby about two sigma with the reweighted theoretical pre-diction and the previous STAR measurement from run2010-2011 (see also Fig. 4 in ref [19]).
The number of e↵ective replicas after reweighting de-pends significantly on the data set. The size of thereweighted parton set is about 90% of the originalNNPDFpol1.1 parton set (made of N
rep
= 100 replicas)for di-jet and pion production data, while it is only about30%-40% for W -boson production data. This result re-flects the di↵erent constraining power of the various datasets, which is maximized in the last case. In principle,a prior ensemble with a larger number of replicas shouldthen be needed for the reweighted ensemble to sample theprobability density in the space of PDFs with as much ac-curacy. However this is not relevant here, as reweightedresults only serve to assess the impact of the new data,and are not used to construct a new parton set.
The modal value of the P(↵) distribution is of orderone for all the new data sets. Values of h↵i slightly largerthan one are found for the W -boson production data: inthe case of W�, this is mostly determined by a sizablefluctuation of one data point (around ⌘
e
� ⇠ 0.25) withrespect to the shape of the corresponding asymmetry; inthe case of W+, this is mostly determined by the fourthdata point (around ⌘
e
+ ⇠ 0.75), which, as already noted,disagrees by about two sigma with the reweighted theo-
-0.1
0
0.1
0.2
0.3 x∆Σ(x,µ2)
µ2=10 GeV
0
0.03
0.06 σx∆Σ NNPDFpol1.1NNPDFpol1.1 (rw)
-0.6
-0.3
0
0.3
0.6x∆g(x,µ2)
NNPDFpol1.1NNPDFpol1.1 (rw)
0
0.1
0.2σx∆g
-0.01
0
0.01
0.02
0.03 x∆u-(x,µ2)
0
0.01
0.02
0.01 0.1 1
x
σx∆u-
-0.03
-0.02
-0.01
0
0.01x∆d-(x,µ2)
0.01 0.1 1 0
0.01
0.02
x
σx∆d-
FIG. 4. A comparison of polarized PDFs before and after thesimultaneous reweighting of NNPDFpol1.1 with the data setslisted in Tab. I. From left to right, top to bottom, the singlet,�⌃, the gluon, �g, and the up and down sea quarks, �uand �d, are displayed. Parton distributions are evaluated atµ2 = 10 GeV. Bands represent one-sigma uncertainties; theyare also shown in the lower inset of each panel.
retical prediction. Values of h↵i slightly smaller than oneare found for the neutral pion production data, thus sug-gesting that experimental uncertainties are likely to beoverestimated, possibly because of the lack of a completeinformation on correlations among systematics.In Fig. 4, I compare the polarized PDFs before and af-
ter the simultaneous reweighting of NNPDFpol1.1 withthe data sets listed in Tab. I. From left to right, topto bottom, I show the singlet (for n
f
active flavors),�⌃ =
Pnf
i=1
(�q
i
+�q
i
), the gluon, �g, and the up anddown sea quarks, �u and �d. Parton distributions areevaluated at µ
2 = 10 GeV2. Bands represent one-sigmauncertainties, which are also displayed separately in thelower inset of each panel.The impact of the wew data on the polarized PDFs of
the proton is twofold. On the one hand, it induces a shiftof the PDF central values, as a consequence of the ad-justment to the shape of the corresponding asymmetries.Specifically, the central value of �g increases by about30% of its original value in the region 0.1 . x . 0.2; thecentral value of �u increases by about 25% and that of�d decreases by about 10% approximately in the sameregion of x wher also �g is a↵ected. On the other hand,the new data induces a reduction of the PDF uncertain-ties, as a consequence of the improved precision of thecorresponding asymmetries. For �g and �d such a re-duction is moderate, and not larger than 5% of its orig-inal value; for �u it is fairly more pronounced, around
5
P(↵), where ↵ is the factor by which the uncertainty onthe new data must be rescaled in order for both the priorand the reweighted sets to be consistent with each other(see Eq. (12) in Ref. [31]). If the modal value of ↵ is closeto unity, the new data is consistent with the old, and itsuncertainties have been correctly estimated.
I perform a simultaneous reweighting of the NNPDF-pol1.1 parton set with all the data sets listed in Tab. I. InTab. II, I show the values of the �
2 per data point afterreweighting, �2
rw
/N
dat
, the number of e↵ective replicas,N
e↵
, and the modal value of the P(↵) distribution, h↵i.The corresponding asymmetries, after reweighting withthe new data, are displayed in Figs. 1-2-3, on top of theircounterparts before reweighting.
The value of the �
2 per data point always decreasesafter reweighting. The improvement is marked for theW -boson production data, moderate for the di-jet pro-duction data and only slight for the neutral pion pro-duction data. This is expected, since the first data sethas the smallest uncertainties, among all, in compari-son to the PDF uncertainties on the theoretical predic-tions. This suggests that this data is bringing in a sig-nificant amount of new information. After reweighting,the �2 per data point is of order one for all the new datasets. However, in the case of W -boson and neutral pionproduction asymmetries, these numbers should be takenwith care, because a complete information on correlatedsystematics is not available. This is the reason why thereweighted �
2 is smaller than one for these sets, exceptfor the W
+ production data. In this case, the value ofthe �2 is raised by a sizable contribution coming from thepoint with the largest positron rapidity, which disagreesby about two sigma with the reweighted theoretical pre-diction and the previous STAR measurement from run2010-2011 (see also Fig. 4 in ref [19]).
The number of e↵ective replicas after reweighting de-pends significantly on the data set. The size of thereweighted parton set is about 90% of the originalNNPDFpol1.1 parton set (made of N
rep
= 100 replicas)for di-jet and pion production data, while it is only about30%-40% for W -boson production data. This result re-flects the di↵erent constraining power of the various datasets, which is maximized in the last case. In principle,a prior ensemble with a larger number of replicas shouldthen be needed for the reweighted ensemble to sample theprobability density in the space of PDFs with as much ac-curacy. However this is not relevant here, as reweightedresults only serve to assess the impact of the new data,and are not used to construct a new parton set.
The modal value of the P(↵) distribution is of orderone for all the new data sets. Values of h↵i slightly largerthan one are found for the W -boson production data: inthe case of W�, this is mostly determined by a sizablefluctuation of one data point (around ⌘
e
� ⇠ 0.25) withrespect to the shape of the corresponding asymmetry; inthe case of W+, this is mostly determined by the fourthdata point (around ⌘
e
+ ⇠ 0.75), which, as already noted,disagrees by about two sigma with the reweighted theo-
-0.1
0
0.1
0.2
0.3 x∆Σ(x,µ2)
µ2=10 GeV
0
0.03
0.06 σx∆Σ NNPDFpol1.1NNPDFpol1.1 (rw)
-0.6
-0.3
0
0.3
0.6x∆g(x,µ2)
NNPDFpol1.1NNPDFpol1.1 (rw)
0
0.1
0.2σx∆g
-0.01
0
0.01
0.02
0.03 x∆u-(x,µ2)
0
0.01
0.02
0.01 0.1 1
x
σx∆u-
-0.03
-0.02
-0.01
0
0.01x∆d-(x,µ2)
0.01 0.1 1 0
0.01
0.02
x
σx∆d-
FIG. 4. A comparison of polarized PDFs before and after thesimultaneous reweighting of NNPDFpol1.1 with the data setslisted in Tab. I. From left to right, top to bottom, the singlet,�⌃, the gluon, �g, and the up and down sea quarks, �uand �d, are displayed. Parton distributions are evaluated atµ2 = 10 GeV. Bands represent one-sigma uncertainties; theyare also shown in the lower inset of each panel.
retical prediction. Values of h↵i slightly smaller than oneare found for the neutral pion production data, thus sug-gesting that experimental uncertainties are likely to beoverestimated, possibly because of the lack of a completeinformation on correlations among systematics.In Fig. 4, I compare the polarized PDFs before and af-
ter the simultaneous reweighting of NNPDFpol1.1 withthe data sets listed in Tab. I. From left to right, topto bottom, I show the singlet (for n
f
active flavors),�⌃ =
Pnf
i=1
(�q
i
+�q
i
), the gluon, �g, and the up anddown sea quarks, �u and �d. Parton distributions areevaluated at µ
2 = 10 GeV2. Bands represent one-sigmauncertainties, which are also displayed separately in thelower inset of each panel.The impact of the wew data on the polarized PDFs of
the proton is twofold. On the one hand, it induces a shiftof the PDF central values, as a consequence of the ad-justment to the shape of the corresponding asymmetries.Specifically, the central value of �g increases by about30% of its original value in the region 0.1 . x . 0.2; thecentral value of �u increases by about 25% and that of�d decreases by about 10% approximately in the sameregion of x wher also �g is a↵ected. On the other hand,the new data induces a reduction of the PDF uncertain-ties, as a consequence of the improved precision of thecorresponding asymmetries. For �g and �d such a re-duction is moderate, and not larger than 5% of its orig-inal value; for �u it is fairly more pronounced, around
The&study&of&spin&in&particle&physics&has&unlocked&doors&to&a&deeper&understanding&of&nucleon&structure
• Helicity• Transversity
Multiple&mechanisms&in&play&to&constrain&transverse&spin3structure
3Spin.Results.at.STAR.9 Drachenberg
The&Fertile&Field&of&Spin&Physics
Opportunities%with%!↑ + !…!
other. One can see that the experimental data indeed showsome tension with the Soffer bound for the d quark in thehigh-x region as predicted in Ref. [94]. This saturationhappens in the region not explored by the current exper-imental data, so future data from Jefferson Lab 12 will bevery important to test the Soffer bound and to constrain thetransversity and tensor charge.The functions themselves are slightly different as can be
seen by comparing solid and dashes lines in Fig. 27(a). Infact Ref. [17] uses the tree-level TMD expression (no TMDevolution) for extraction, and we use the NLL TMDformalism. Results should be different even though inasymmetries, as we saw, at low energies results with NLLTMD are comparable with the tree level. At higher energiesand Q2, the situation changes, and extracted functionsmust be different. At the same time, one should rememberTMD evolution does not act as a universal Q2 suppressionfactor. A complicated Fourier transform should be per-formed that mixes Q2 and b dependence, and thus theresulting functions are different in shape but comparable inmagnitude. It is also very encouraging that tree-level TMDextractions yielded results very similar to our NLL extrac-tion. This makes the previous phenomenological resultsvalid even though the appropriate TMD evolution was nottaken into account. It also means that we need to haveexperimental data on unpolarized cross sections differentialin Ph⊥. As we have seen, the effects of evolution should beevident in the data, and those measurements will help toestablish the validity of the modern formulation of TMDevolution.We compare extracted Collins fragmentation functions
−zHð3ÞðzÞ in Fig. 28 at Q2 ¼ 2.4 GeV2 with the extractionof Torino-Cagliari-JLab 2013 [17]. The resulting CollinsFFs have the same signs, but shapes and sizes are slightlydifferent. Indeed one could expect it as far as Q2 of eþe− isdifferent, and the evolution effect must be more evident. Atthe same time, those functions for both tree-level and NLL
TMD give the same (or similar) theoretical asymmetriesthat are well compared to the experimental data of SIDISand eþe−. The favored Collins fragmentation function ismuch better determined by the existing data, as one cansee from Fig. 28 that the functions at Q2 ¼ 2.4 GeV2 arecompatible within error bands. The unfavored fragmenta-tion functions are different; however, those functions arenot determined very well by existing experimental data.We also compare the tensor change from our and other
extractions in Fig. 29. The contribution to the tensor chargeof Ref. [18] is found by extraction using the so-calleddihadron fragmentation function that couples to the col-linear transversity distribution. The corresponding func-tions have DGLAP-type evolution known at LO and wereused in Ref. [18]. The results plotted in Fig. 29 correspondto our estimates of the contribution to the u quark and dquark in the region of x½0.065; 0.35& at Q2 ¼ 10 GeV2 at68% C.L. (label 1) and the contribution to the u quark andd quark in the same region of x and the same Q2 using the
)2(x
,Q1
x h
ud
x
0
0.1
0.2
0.3
0 0.2 0.4 0.6 0.8 1
-0.15-0.1
-0.050
0.05
Kang et al (2015)
Anselmino et al (2013)
(a)
)2(x
,Q1
x h
ud
x
-0.2
0
0.2
0.4
0 0.2 0.4 0.6 0.8 1
-0.2
-0.1
0
0.1
Kang et al (2015)
Radici et al (2015)
(b)
FIG. 27. (a) Comparison of extracted transversity (solid lines and vertical-line hashed region) Q2 ¼ 2.4 GeV2 with the Torino-Cagliari-JLab 2013 extraction [17] (dashed lines and shaded region). (b) Comparison of extracted transversity (solid lines and shadedregion) at Q2 ¼ 2.4 GeV2 with Pavia 2015 extraction [18] (shaded region).
)2(z
,Q(3
)H
-zfa
vun
fav
z
0
0.02
0.04Kang et al (2015)Anselmino et al (2013)
0 0.2 0.4 0.6 0.8 1
-0.04
-0.02
0
FIG. 28. Comparison of extracted Collins fragmentation func-tions (solid lines and vertical-line hashed region) at Q2 ¼2.4 GeV2 with the Torino-Cagliari-JLab 2013 extraction [17](dashed lines and shaded region).
EXTRACTION OF QUARK TRANSVERSITY DISTRIBUTION … PHYSICAL REVIEW D 93, 014009 (2016)
014009-33
PRD.93,.014009.(2016)
Collins&from&SIDIS
IFF&from&SIDIS(Preliminary%update%atDIS2017)
Collins&from&SIDIS
Collins&Effect&in&Hadroproduction• Asymmetry.in.distribution%of%hadrons%within%jets%∼ sin () − (+
4Spin.Results.at.STAR.9 Drachenberg
The&Fertile&Field&of&Spin&Physics
Anselmino et.al.,.PRD.73,.014020.(2006)F..Yuan,.PRL.100,.032003.(2008)D’Alesio et.al.,.PRD.83,.034021.(2011)
φH–pbeam
pbeamS�
pπ
PJET
jT
φSA
Terms in&Numerator&of&TMD&SSA&for&qq Scattering English&Names Modulation
Δ-.//1↑ ⋅ .3/4 ⋅ 56/7 Sivers •.PDF •.FF sin ()8ℎ:/ ⋅ Δ-.3↑/4 ⋅ 56/7 Transversity •.Boer9Mulders.•.FF sin ()8ℎ:;</ ⋅ Δ-.3↑/4 ⋅ 56/7 Pretzelocity •.Boer9Mulders.•.FF sin ()8ℎ:/ ⋅ .3/4 ⋅ Δ-56/7↑ Transversity •.PDF.•.Collins sin ()8 − (+
Δ-.//1↑ ⋅ Δ-.3↑/4 ⋅ Δ
-56/7↑ Sivers •.Boer9Mulders.•.Collins sin ()8 − (+
ℎ:;</ ⋅ .3/4 ⋅ Δ-56/7↑ Pretzelocity •.PDF •.Collins sin ()8 + (+
Δ-.//1↑ ⋅ Δ-.3↑/4 ⋅ Δ
-56/7↑ Sivers •.Boer9Mulders.•.Collins sin ()8 + (+
Transverse&Momentum&Dependent&(TMD)&Approach
Collins&Effect&in&Hadroproduction• Asymmetry.in.distribution%of%hadrons%within%jets%∼ sin () − (+
5Spin.Results.at.STAR.9 Drachenberg
The&Fertile&Field&of&Spin&Physics
Anselmino et.al.,.PRD.73,.014020.(2006)F..Yuan,.PRL.100,.032003.(2008)D’Alesio et.al.,.PRD.83,.034021.(2011)
φH–pbeam
pbeamS�
pπ
PJET
jT
φSA
Terms in&Numerator&of&TMD&SSA&for&qq Scattering English&Names Modulation
Δ-.//1↑ ⋅ .3/4 ⋅ 56/7 Sivers •.PDF •.FF sin ()8ℎ:/ ⋅ Δ-.3↑/4 ⋅ 56/7 Transversity •.Boer9Mulders.•.FF sin ()8ℎ:;</ ⋅ Δ-.3↑/4 ⋅ 56/7 Pretzelocity •.Boer9Mulders.•.FF sin ()8ℎ:/ ⋅ .3/4 ⋅ Δ-56/7↑ Transversity •.PDF.•.Collins sin ()8 − (+
Δ-.//1↑ ⋅ Δ-.3↑/4 ⋅ Δ
-56/7↑ Sivers •.Boer9Mulders.•.Collins sin ()8 − (+
ℎ:;</ ⋅ .3/4 ⋅ Δ-56/7↑ Pretzelocity •.PDF •.Collins sin ()8 + (+
Δ-.//1↑ ⋅ Δ-.3↑/4 ⋅ Δ
-56/7↑ Sivers •.Boer9Mulders.•.Collins sin ()8 + (+
Transverse&Momentum&Dependent&(TMD)&Approach
Collins&Effect&in&Hadroproduction• Asymmetry.in.distribution%of%hadrons%within%jets%∼ sin () − (+• Do.TMD.jet(h).observables.factorize%in%p+p? [arXiv:1705.08443]• How.do.TMDs.evolve,.e.g..with. =, ?@ ?• Enhance.A9quark.sensitivity
6Spin.Results.at.STAR.9 Drachenberg
The&Fertile&Field&of&Spin&Physics
Anselmino et.al.,.PRD.73,.014020.(2006)F..Yuan,.PRL.100,.032003.(2008)D’Alesio et.al.,.PRD.83,.034021.(2011)
φH–pbeam
pbeamS�
pπ
PJET
jT
φSA
other. One can see that the experimental data indeed showsome tension with the Soffer bound for the d quark in thehigh-x region as predicted in Ref. [94]. This saturationhappens in the region not explored by the current exper-imental data, so future data from Jefferson Lab 12 will bevery important to test the Soffer bound and to constrain thetransversity and tensor charge.The functions themselves are slightly different as can be
seen by comparing solid and dashes lines in Fig. 27(a). Infact Ref. [17] uses the tree-level TMD expression (no TMDevolution) for extraction, and we use the NLL TMDformalism. Results should be different even though inasymmetries, as we saw, at low energies results with NLLTMD are comparable with the tree level. At higher energiesand Q2, the situation changes, and extracted functionsmust be different. At the same time, one should rememberTMD evolution does not act as a universal Q2 suppressionfactor. A complicated Fourier transform should be per-formed that mixes Q2 and b dependence, and thus theresulting functions are different in shape but comparable inmagnitude. It is also very encouraging that tree-level TMDextractions yielded results very similar to our NLL extrac-tion. This makes the previous phenomenological resultsvalid even though the appropriate TMD evolution was nottaken into account. It also means that we need to haveexperimental data on unpolarized cross sections differentialin Ph⊥. As we have seen, the effects of evolution should beevident in the data, and those measurements will help toestablish the validity of the modern formulation of TMDevolution.We compare extracted Collins fragmentation functions
−zHð3ÞðzÞ in Fig. 28 at Q2 ¼ 2.4 GeV2 with the extractionof Torino-Cagliari-JLab 2013 [17]. The resulting CollinsFFs have the same signs, but shapes and sizes are slightlydifferent. Indeed one could expect it as far as Q2 of eþe− isdifferent, and the evolution effect must be more evident. Atthe same time, those functions for both tree-level and NLL
TMD give the same (or similar) theoretical asymmetriesthat are well compared to the experimental data of SIDISand eþe−. The favored Collins fragmentation function ismuch better determined by the existing data, as one cansee from Fig. 28 that the functions at Q2 ¼ 2.4 GeV2 arecompatible within error bands. The unfavored fragmenta-tion functions are different; however, those functions arenot determined very well by existing experimental data.We also compare the tensor change from our and other
extractions in Fig. 29. The contribution to the tensor chargeof Ref. [18] is found by extraction using the so-calleddihadron fragmentation function that couples to the col-linear transversity distribution. The corresponding func-tions have DGLAP-type evolution known at LO and wereused in Ref. [18]. The results plotted in Fig. 29 correspondto our estimates of the contribution to the u quark and dquark in the region of x½0.065; 0.35& at Q2 ¼ 10 GeV2 at68% C.L. (label 1) and the contribution to the u quark andd quark in the same region of x and the same Q2 using the
)2(x
,Q1
x h
ud
x
0
0.1
0.2
0.3
0 0.2 0.4 0.6 0.8 1
-0.15-0.1
-0.050
0.05
Kang et al (2015)
Anselmino et al (2013)
(a)
)2(x
,Q1
x h
ud
x
-0.2
0
0.2
0.4
0 0.2 0.4 0.6 0.8 1
-0.2
-0.1
0
0.1
Kang et al (2015)
Radici et al (2015)
(b)
FIG. 27. (a) Comparison of extracted transversity (solid lines and vertical-line hashed region) Q2 ¼ 2.4 GeV2 with the Torino-Cagliari-JLab 2013 extraction [17] (dashed lines and shaded region). (b) Comparison of extracted transversity (solid lines and shadedregion) at Q2 ¼ 2.4 GeV2 with Pavia 2015 extraction [18] (shaded region).
)2(z
,Q(3
)H
-zfa
vun
fav
z
0
0.02
0.04Kang et al (2015)Anselmino et al (2013)
0 0.2 0.4 0.6 0.8 1
-0.04
-0.02
0
FIG. 28. Comparison of extracted Collins fragmentation func-tions (solid lines and vertical-line hashed region) at Q2 ¼2.4 GeV2 with the Torino-Cagliari-JLab 2013 extraction [17](dashed lines and shaded region).
EXTRACTION OF QUARK TRANSVERSITY DISTRIBUTION … PHYSICAL REVIEW D 93, 014009 (2016)
014009-33
PRD.93,.014009.(2016)
NIM.A499,.245.(2003)
Relativistic&Heavy&Ion&ColliderRHIC&as&PolarizedXproton&Collider• “Siberian.Snakes”.!mitigate.depolarization.resonances• Spin.rotators.provide.choice.of.spin.orientation.
independent&of&experiment• Spin.direction.varies.bucket9to9bucket.(9.4.MHz)• Spin.pattern.varies.fill9to9fill
7Spin.Results.at.STAR.9 Drachenberg
NIM.A499,.245.(2003)
Solenoidal Tracker&at&RHIC
Jets,&diXhadrons,&weak&bosonsTPC.+.Barrel.+.Endcap.EMC
RHIC&as&PolarizedXproton&Collider• “Siberian.Snakes”.!mitigate.depolarization.resonances• Spin.rotators.provide.choice.of.spin.orientation.
independent&of&experiment• Spin.direction.varies.bucket9to9bucket.(9.4.MHz)• Spin.pattern.varies.fill9to9fill
8Spin.Results.at.STAR.9 Drachenberg
Trigger on.calorimeter.energy
Year B� [GeV] STAR D [%]2006 200 8.5.pb91 57
2006 62.4 0.2.pb91 48
2008 200 7.8.pb91 45
2011 500 25.pb91 53/54
2012 200 22.pb91 61/58
2015 200 53.pb91 53/57
2015 200 pAu 0.42.pb91 60
2015 200.pAl 1.0.pb91 54
2017 510 320.pb91 56
PolarizedXproton&Datasets&at&RHIC
9Spin.Results.at.STAR.9 Drachenberg
Unique&opportunities&to&probe&nucleon&spin&structure!
Dramatically%increased%figure%of%merit%in%recent%years
Transverse%Luminosity%Recorded
Arbi
trary
Uni
ts
0
0.05
0.1
0.15
0.2 > 0η, JP2, QQ+x
< 0η, JP2, QQ+x
< 55 GeV/c
T22.7 < p
Log(x)3− 2.5− 2− 1.5− 1− 0.5− 00
0.05
0.1
0.15
0.2
0 > 0η, VPDMB, Gx
< 0η, VPDMB, Gx
points: unbiased; histogram: triggered
< 13.8 GeV/c
T6 < p
0
arXiv:1708.07080
B� = 500 GeV
Kinematic&Sensitivity&at&STAR
10Spin.Results.at.STAR.9 Drachenberg
Access&to&transversity in&interesting®ion!• Limited.constraints• Potentially.large.effects• Sensitivity.to.evolution• Insight%into%nature%of%Collins%mechanism!
other. One can see that the experimental data indeed showsome tension with the Soffer bound for the d quark in thehigh-x region as predicted in Ref. [94]. This saturationhappens in the region not explored by the current exper-imental data, so future data from Jefferson Lab 12 will bevery important to test the Soffer bound and to constrain thetransversity and tensor charge.The functions themselves are slightly different as can be
seen by comparing solid and dashes lines in Fig. 27(a). Infact Ref. [17] uses the tree-level TMD expression (no TMDevolution) for extraction, and we use the NLL TMDformalism. Results should be different even though inasymmetries, as we saw, at low energies results with NLLTMD are comparable with the tree level. At higher energiesand Q2, the situation changes, and extracted functionsmust be different. At the same time, one should rememberTMD evolution does not act as a universal Q2 suppressionfactor. A complicated Fourier transform should be per-formed that mixes Q2 and b dependence, and thus theresulting functions are different in shape but comparable inmagnitude. It is also very encouraging that tree-level TMDextractions yielded results very similar to our NLL extrac-tion. This makes the previous phenomenological resultsvalid even though the appropriate TMD evolution was nottaken into account. It also means that we need to haveexperimental data on unpolarized cross sections differentialin Ph⊥. As we have seen, the effects of evolution should beevident in the data, and those measurements will help toestablish the validity of the modern formulation of TMDevolution.We compare extracted Collins fragmentation functions
−zHð3ÞðzÞ in Fig. 28 at Q2 ¼ 2.4 GeV2 with the extractionof Torino-Cagliari-JLab 2013 [17]. The resulting CollinsFFs have the same signs, but shapes and sizes are slightlydifferent. Indeed one could expect it as far as Q2 of eþe− isdifferent, and the evolution effect must be more evident. Atthe same time, those functions for both tree-level and NLL
TMD give the same (or similar) theoretical asymmetriesthat are well compared to the experimental data of SIDISand eþe−. The favored Collins fragmentation function ismuch better determined by the existing data, as one cansee from Fig. 28 that the functions at Q2 ¼ 2.4 GeV2 arecompatible within error bands. The unfavored fragmenta-tion functions are different; however, those functions arenot determined very well by existing experimental data.We also compare the tensor change from our and other
extractions in Fig. 29. The contribution to the tensor chargeof Ref. [18] is found by extraction using the so-calleddihadron fragmentation function that couples to the col-linear transversity distribution. The corresponding func-tions have DGLAP-type evolution known at LO and wereused in Ref. [18]. The results plotted in Fig. 29 correspondto our estimates of the contribution to the u quark and dquark in the region of x½0.065; 0.35& at Q2 ¼ 10 GeV2 at68% C.L. (label 1) and the contribution to the u quark andd quark in the same region of x and the same Q2 using the
)2(x
,Q1
x h
ud
x
0
0.1
0.2
0.3
0 0.2 0.4 0.6 0.8 1
-0.15-0.1
-0.050
0.05
Kang et al (2015)
Anselmino et al (2013)
(a))2
(x,Q
1x
hu
d
x
-0.2
0
0.2
0.4
0 0.2 0.4 0.6 0.8 1
-0.2
-0.1
0
0.1
Kang et al (2015)
Radici et al (2015)
(b)
FIG. 27. (a) Comparison of extracted transversity (solid lines and vertical-line hashed region) Q2 ¼ 2.4 GeV2 with the Torino-Cagliari-JLab 2013 extraction [17] (dashed lines and shaded region). (b) Comparison of extracted transversity (solid lines and shadedregion) at Q2 ¼ 2.4 GeV2 with Pavia 2015 extraction [18] (shaded region).
)2(z
,Q(3
)H
-zfa
vun
fav
z
0
0.02
0.04Kang et al (2015)Anselmino et al (2013)
0 0.2 0.4 0.6 0.8 1
-0.04
-0.02
0
FIG. 28. Comparison of extracted Collins fragmentation func-tions (solid lines and vertical-line hashed region) at Q2 ¼2.4 GeV2 with the Torino-Cagliari-JLab 2013 extraction [17](dashed lines and shaded region).
EXTRACTION OF QUARK TRANSVERSITY DISTRIBUTION … PHYSICAL REVIEW D 93, 014009 (2016)
014009-33
PRD.93,.014009.(2016)
11Spin.Results.at.STAR.9 Drachenberg
First%Collins%asymmetry%observations%in%hadroproduction!New&500&GeV&Paper:&arXiv:1708.07080
STAR&Collins&Results&at& B� = 200 and 500 GeV
+ X±π jet + → + p ↑p
= 1
.3 G
eV
/c⟩
T,je
tp⟨
×m
inR
∆
z0.2 0.4 0.6
UT
)Hφ-
Sφsi
n(
A
0.05−
0
0.05
-π; Open points: +πClosed points:
STAR = 12.9 GeV/c (Preliminary)⟩
T,jetp⟨ = 200 GeV, s
= 31.0 GeV/c⟩T,jet
p⟨ = 500 GeV, s
At&the¤t&precision,results&from&! + ! exhibit&=; scaling
200&GeV:&Int.&J.&Mod.&Phys.&Conf.&Ser.&40,&1660040
Comparison&to&Models
12Spin.Results.at.STAR.9 Drachenberg
z0.1 0.2 0.3 0.4 0.5 0.6
UT
)Hφ-
Sφsi
n(
A
0.05−
0
0.05
+ X±π jet + → + p ↑p = 500 GeVs < 1
jetη0 < = 31.0 GeV/c⟩
T,jetp⟨
+πSTAR 2011
-πSTAR 2011
+πModel Curves Positive: -πModel Curves Negative:
DMP+2013 KPRY KPRY-NLL
arXiv:1708.07080
Models%based%on%SIDIS/IJIK• Assume universality%and.
robust%factorization• DMP&KPRY:&no&TMD&evol.• KPRYXNLL:&TMD&evolution&up&
to&NLLDMP:&PLB&773,&300&(2017)KPRY:&arXiv:1707.00913
Comparison&to&Models
13Spin.Results.at.STAR.9 Drachenberg
Consistency&between&models&and&STAR&data&at&95%&confidence&level! Suggests&robust&factorization&and&universality
z0.1 0.2 0.3 0.4 0.5 0.6
UT
)Hφ-
Sφsi
n(
A
0.05−
0
0.05
+ X±π jet + → + p ↑p = 500 GeVs < 1
jetη0 < = 31.0 GeV/c⟩
T,jetp⟨
+πSTAR 2011
-πSTAR 2011
+πModel Curves Positive: -πModel Curves Negative:
DMP+2013 KPRY KPRY-NLL
arXiv:1708.07080
Models%based%on%SIDIS/IJIK• Assume universality%and.
robust%factorization• DMP&KPRY:&no&TMD&evol.• KPRYXNLL:&TMD&evolution&up&
to&NLLDMP:&PLB&773,&300&(2017)KPRY:&arXiv:1707.00913
Comparison&to&Models
14Spin.Results.at.STAR.9 Drachenberg
Consistency&between&models&and&STAR&data&at&95%&confidence&level! Suggests&robust&factorization&and&universality
Slight&preference&for&no&evolution?
z0.1 0.2 0.3 0.4 0.5 0.6
UT
)Hφ-
Sφsi
n(
A
0.05−
0
0.05
+ X±π jet + → + p ↑p = 500 GeVs < 1
jetη0 < = 31.0 GeV/c⟩
T,jetp⟨
+πSTAR 2011
-πSTAR 2011
+πModel Curves Positive: -πModel Curves Negative:
DMP+2013 KPRY KPRY-NLL
arXiv:1708.07080
Models%based%on%SIDIS/IJIK• Assume universality%and.
robust%factorization• DMP&KPRY:&no&TMD&evol.• KPRYXNLL:&TMD&evolution&up&
to&NLLDMP:&PLB&773,&300&(2017)KPRY:&arXiv:1707.00913
Comparison&to&Models
15Spin.Results.at.STAR.9 Drachenberg
Consistency&between&models&and&STAR&data&at&95%&confidence&level! Suggests&robust&factorization&and&universality
Slight&preference&for&no&evolution?L@/M = 14/10 (w/o).vs..17.6/10 (with)
For&now,&“Beauty&is&in&the&eye&of&the&beholder!”(a.k.a..need.more.data!)
z0.1 0.2 0.3 0.4 0.5 0.6
UT
)Hφ-
Sφsi
n(
A
0.05−
0
0.05
+ X±π jet + → + p ↑p = 500 GeVs < 1
jetη0 < = 31.0 GeV/c⟩
T,jetp⟨
+πSTAR 2011
-πSTAR 2011
+πModel Curves Positive: -πModel Curves Negative:
DMP+2013 KPRY KPRY-NLL
arXiv:1708.07080
Models%based%on%SIDIS/IJIK• Assume universality%and.
robust%factorization• DMP&KPRY:&no&TMD&evol.• KPRYXNLL:&TMD&evolution&up&
to&NLLDMP:&PLB&773,&300&(2017)KPRY:&arXiv:1707.00913
16Spin.Results.at.STAR.9 Drachenberg
Asymmetries&appear&to&decrease&with&S;Consistent&between&energies?
STAR&Collins&Results&at& B� = 200 and 500 GeV
< 1jet
η0 <
UT
) Hφ- Sφsi
n(A
0.05−
0
0.05
< 0jet
η-1 <
= 0.13⟩z
⟨
+ X±π jet + → + p ↑p = 500 GeVs = 31.0⟩
T,jetp⟨
0.05−
0
0.05
= 0.23⟩z
⟨
1−10 1
0.05−
0
0.05STAR 2011
+π-π
= 0.37⟩z
⟨
[GeV/c]Tj
1−10 1
) Hφ - Sφ
sin(
UT
A
+π-π
STAR Preliminary > 0Fx
= 200 GeVs + X at ±π jet + → + p ↑p5.6% Scale Uncertainty Not Shown
0.04
0
-0.04
)Hφ
- Sφ
sin(
UT
A
+π-π < 0Fx0.04
0
-0.04
TJe
t p
< 31.6 GeV/cT
10 < Jet p16
12
8
[GeV/c]Tj
-110 1Z
0.1 < z < 0.60.750.5
0.25
Dependence&on&S; (momentum&transverse&to&jet)
arXiv:1708.07080
Int..J..Mod..Phys..Conf..Ser..40,.1660040
17Spin.Results.at.STAR.9 Drachenberg
Further&investigation&of&low&TU region&needede.g.&unpolarized TMD&data,&model¶meterization,&etc.
STAR&Collins&Results&at& B� = 200 and 500 GeV
) Hφ - Sφ
sin(
UT
A
+π-π
STAR Preliminary > 0Fx
= 200 GeVs + X at ±π jet + → + p ↑p5.6% Scale Uncertainty Not Shown
0.04
0
-0.04
)Hφ
- Sφ
sin(
UT
A
+π-π < 0Fx0.04
0
-0.04
TJe
t p
< 31.6 GeV/cT
10 < Jet p16
12
8
[GeV/c]Tj
-110 1Z
0.1 < z < 0.60.750.5
0.25
Dependence&on&S; (momentum&transverse&to&jet)
Compare&to&DMP&modelsPLB.773,.300.(2017)
Int..J..Mod..Phys..Conf..Ser..40,.1660040
The&NearXterm&Future:&Collins&Evolution&
18Spin.Results.at.STAR.9 Drachenberg
+ X±π jet + → + p(Au) ↑p
z0.2 0.4 0.6
UT) Hφ- Sφ
sin(
A
0.05−
0
0.05
STAR 0.13≈ TJet x
-π; Open points: +πClosed points:
= 200 GeV (Preliminary 2012)sp+p, = 200 GeV (proj. stat. 2012+2015)sp+p,
= 200 GeV (proj. stat. 2015)sp+Au, = 500 GeV (STAR 2011)sp+p, = 500 GeV (proj. stat. 2017)sp+p,
2011.and.Preliminary.2012.Collins.asymmetries.suggest.=; scalingImplications%for%TMD%evolution?
+ X±π jet + → + p(Au) ↑p
z0.2 0.4 0.6
UT) Hφ- Sφ
sin(
A
0.05−
0
0.05
STAR 0.13≈ TJet x
-π; Open points: +πClosed points:
= 200 GeV (Preliminary 2012)sp+p, = 200 GeV (proj. stat. 2012+2015)sp+p,
= 200 GeV (proj. stat. 2015)sp+Au, = 500 GeV (STAR 2011)sp+p, = 500 GeV (proj. stat. 2017)sp+p,
The&NearXterm&Future:&Collins&Evolution&
19Spin.Results.at.STAR.9 Drachenberg
Higher&precision&in&2015&and&2017&will&allow&more&
precise&comparison!
2011.and.Preliminary.2012.Collins.asymmetries.suggest.=; scalingImplications%for%TMD%evolution?
+ X±π jet + → + p(Au) ↑p
z0.2 0.4 0.6
UT) Hφ- Sφ
sin(
A
0.05−
0
0.05
STAR 0.13≈ TJet x
-π; Open points: +πClosed points:
= 200 GeV (Preliminary 2012)sp+p, = 200 GeV (proj. stat. 2012+2015)sp+p,
= 200 GeV (proj. stat. 2015)sp+Au, = 500 GeV (STAR 2011)sp+p, = 500 GeV (proj. stat. 2017)sp+p,
The&NearXterm&Future:&! + V Collins
20Spin.Results.at.STAR.9 Drachenberg
Higher&precision&in&2015&and&2017&will&allow&more&
precise&comparison!
First&!↑ + VW run!Should&allow&for&first&
glimpse&of&Collins&in&! + V! Explore&hadronization
2011.and.Preliminary.2012.Collins.asymmetries.suggest.=; scalingImplications%for%TMD%evolution?
Summary
21Recent.Transverse.Spin.Developments.at.RHIC.9 Drachenberg
• Spin&physics&is&a&fertile&field&and&! + ! plays&a&critical&role
Summary
22Recent.Transverse.Spin.Developments.at.RHIC.9 Drachenberg
• Spin&physics&is&a&fertile&field&and&! + ! plays&a&critical&role• First.observations.of.Collins.effect.in.polarized.! + !
9 Possible.=; scaling9 Consistency.with.models.suggests.robust.factorization.and.
universality9 Evolution.effects.slow.(more.precise.data.needed.to.quantify)
Summary
23Recent.Transverse.Spin.Developments.at.RHIC.9 Drachenberg
• Spin&physics&is&a&fertile&field&and&! + ! plays&a&critical&role• First.observations.of.Collins.effect.in.polarized.! + !
9 Possible.=; scaling9 Consistency.with.models.suggests.robust.factorization.and.
universality9 Evolution.effects.slow.(more.precise.data.needed.to.quantify)
• Recent&and&nearXfuture&runs&offer&even&more&potential9 Substantially.increased.precision.for.Collins.at.200.and.510.GeV9 First.investigation.of.Collins.in.! + V
Summary
• Spin&physics&is&a&fertile&field&and&! + ! plays&a&critical&role• First.observations.of.Collins.effect.in.polarized.! + !
9 Possible.=; scaling9 Consistency.with.models.suggests.robust.factorization.and.
universality9 Evolution.effects.slow.(more.precise.data.needed.to.quantify)
• Recent&and&nearXfuture&runs&offer&even&more&potential9 Substantially.increased.precision.for.Collins.at.200.and.510.GeV9 First.investigation.of.Collins.in.! + V
Stay%tuned%for%more%new%results%from%STAR!
24Recent.Transverse.Spin.Developments.at.RHIC.9 Drachenberg
BackXup&Slides
25Spin.Results.at.STAR.9 Drachenberg
STAR&Results&at& B� = 500 GeV
26Spin.Results.at.STAR.9 Drachenberg
< 1jet
η0 <
UT
) Hφ- Sφsi
n(A
0.05−
0
0.05
< 0jet
η-1 <
= 10.6 GeV/c
⟩T,jet
p⟨
0.05−
0
0.05STAR 2011
+π-π = 20.6 G
eV/c⟩
T,jetp⟨
+ X±π jet + → + p ↑p = 500 GeVs
0.2 0.4 0.6
0.05−
0
0.05
= 31.0 GeV/c
⟩T,jet
p⟨
z0.2 0.4 0.6
< 1jet
η0 <
UT
) Hφ- Sφsi
n(A
0.05−
0
0.05STAR 2011
+π-π
< 0jet
η-1 <
= 0.14⟩z
⟨
+ X±π jet + → + p ↑p = 500 GeVs
0.05−
0
0.05
= 0.24⟩z
⟨
10 20 30 40 50
0.05−
0
0.05
= 0.38⟩z
⟨
[GeV/c]T
Particle-jet p10 20 30 40 50
) Hφ - Sφ
sin(
UT
A
+π-π
STAR Preliminary > 0Fx
= 200 GeVs + X at ±π jet + → + p ↑p5.6% Scale Uncertainty Not Shown
0.04
0
-0.04
)Hφ
- Sφ
sin(
UT
A
+π-π < 0Fx0.04
0
-0.04
TJe
t p
< 31.6 GeV/cT
10 < Jet p16
12
8
[GeV/c]Tj
-110 1
Z
0.1 < z < 0.60.750.5
0.25) Hφ
- Sφsi
n(U
TA
+π-π
STAR Preliminary > 0Fx
= 200 GeVs + X at ±π jet + → + p ↑p5.6% Scale Uncertainty Not Shown
0.04
0
-0.04
)Hφ
- Sφ
sin(
UT
A
+π-π < 0Fx0.04
0
-0.04
Z
0.1 < z < 0.60.75
0.5
0.25
[GeV/c]T
Particle Jet p5 10 15 20 25 30
Tj
< 4.5 GeV/cT
0.125 < j1.51
0.5
STAR&Results&at& B� = 200 GeV
27Spin.Results.at.STAR.9 Drachenberg
Transverse&Asymmetries&for&Gluon&Jets
φH
–pbeam
pbeamS�
pπ
PJET
jT
φSA
Transverse&Momentum&Dependent&(TMD)&Approach
28Spin.Results.at.STAR.9 Drachenberg
Terms in&Numerator&of&TMD&SSA&for&gg Scattering English&Names Modulation
Sivers •.PDF •.FF
Transversity9like•.Boer9Mulders9like.•.FF
Pretzelocity9like•.Boer9Mulders9like.•.FFTransversity9like.•.PDF
•.Collins9likeSivers •.Boer9Mulders9like
• Collins9likePretzelocity9like.•.PDF
•.Collins9likeSivers •.Boer9Mulders9like.•.
Collins9like
ΔN fa/A↑
• fb/B •Dπ /g sin φSA( )sin φSA( )sin φSA( )
sin φSA − 2φH( )sin φSA − 2φH( )sin φSA + 2φH( )sin φSA + 2φH( )
ImF+−a+− •Δ
N fTb1/B
•Dπ /g
ΔN fa/A↑
•ΔN fTb1/B
•ΔNDπ /T g1
ImF−+a+− • fb/B •Δ
NDπ /T g1
ΔN fa/A↑
•ΔN fTb1/B
•ΔNDπ /T g1
ImF−+a+− •Δ
N fTb1/B
•Dπ /g
ImF+−a+− • fb/B •Δ
NDπ /T g1
Anselmino et.al.,.PRD.73,.014020.(2006)F..Yuan,.PRL.100,.032003.(2008)
D’Alesio et.al.,.PRD.83,.034021.(2011)
UNCONSTRAINED!
Asymmetry&modulations&sensitive&to&various&contributions
(often.involving.transversely&polarized&quarks or.linearly&
polarized&gluons)
VX; – Transverse.single9spin.asymmetry.(also.written.V-)
STAR&Results&at& B� = 500 GeV
29Spin.Results.at.STAR.9 Drachenberg
[GeV/c]T
Particle-jet p5 6 7 8 9 10 11 12 13 14 15
UT
) Hφ-2 Sφ
sin(
A
0.02−
0
0.02
+ X±π jet + → + p ↑p = 500 GeVs| < 1
jetη| = 0.18⟩z⟨
STAR 2011DMP+KretzerDMP+DSS
z0.1 0.2 0.3 0.4 0.5 0.6
UT
) Hφ-2 Sφ
sin(
A
0.02−
0
0.02
+ X±π jet + → + p ↑p = 500 GeVs| < 1
jetη| = 10.6 GeV/c⟩
T,jetp⟨
STAR 2011DMP+KretzerDMP+DSS
UT
) Sφsi
n(A
0.02−
0
0.02
< -0.5jet
η-1 <
10 20 30 40 50
0.02−
0
0.02
< 0.5jet
η0 <
STAR 2011
jet + X→ + p ↑p = 500 GeVs
< 0jet
η-0.5 <
[GeV/c]T
Particle-jet p10 20 30 40 50
< 1jet
η0.5 <
Jet&Reconstruction&at&RHIC
e.g..Anti9Y; algorithmJHEP.0804,.063.(2008)
Radius.parameter.Z = 0.5 or.0.6
Use.PYTHIA +.GEANT to.quantify.detector.response
STAR&DiXjet&event&at&detector&level
!"
#0
!0!+
eC
e+
Detector
GEANT
PYTHIA
etcpe
,,,π
γν ,,
gq,
Data&jets&&&&&&&&&&&&&&&&&&&&&&&&&&&&& MC&jets
Particle
Parton
[± Kinematic&Variables] – [ momentum./.jet.momentumS; – [ !; relative.to.jet.axis
30Spin.Results.at.STAR.9 Drachenberg
Non9zero.asymmetry.from.multi9partoncorrelation.functions
e.g..Qiu and.Sterman,.PRL.67,.2264.(1991); PRD.59,.014004.(1998)
Correlators closely%related%to%kTmoments%of%TMDs
Boer,.Mulders,.Pijlman,.NPB.667,.201.(2003)
Formalisms&for&Transverse&SingleXspin&Asymmetries
Sivers mechanism: asymmetry.in.e.g..jet.or.γ productionD..Sivers,.PRD.41,.83.(1990);.43,.261.(1991)
Collins*mechanism: asymmetry.in.the.jet.fragmentationJ..Collins,.NP.B396,.161.(1993)
Sensitive.to proton&spin–parton transverse&motioncorrelations.(needs.Lz)
SP kT,parton
p
p
⟨SP � (p × kT,parton) ⟩ ≠.0 SP
p
p
Sq kT,π
Sensitive.to.transversity (h1)
⟨Sq� (p × kT,π) ⟩ ≠.0
π± Kinematic&Variablesz = pπ / pjet
jT (kT, π) = π pT relative.to.jet.axis
!p !p !p
p p p
! ! !
Y..Koike,.RSC.Discussion.(2004)
Transverse&Momentum&Dependent&(TMD)&PDFs&and&FFs
Collinear&TwistX3&Correlators
31Spin.Results.at.STAR.9 Drachenberg