Stefano Bertini, Sergio L. Cacciatori, and Bianca L. Cerchiai
Citation: Journal of Mathematical Physics 47, 043510 (2006); doi: 10.1063/1.2190898 View online: http://dx.doi.org/10.1063/1.2190898
View Table of Contents: http://scitation.aip.org/content/aip/journal/jmp/47/4?ver=pdfcov Published by the AIP Publishing
Articles you may be interested in
Euler angles for G 2
J. Math. Phys. 46, 083512 (2005); 10.1063/1.1993549
Euler angles, direction cosines, and angular momentum Am. J. Phys. 49, 744 (1981); 10.1119/1.12421
A Note on Euler Angle Rotations
Am. J. Phys. 40, 1036 (1972); 10.1119/1.1986739
A Note on Euler's Angles
Am. J. Phys. 40, 345 (1972); 10.1119/1.1986528
MITAC Gyroscope as a Classroom Model to Illustrate the Euler Angles
Am. J. Phys. 28, 818 (1960); 10.1119/1.1936017
On the Euler angles for SU „N…
Stefano Bertinia兲
Dipartimento di Fisica dell’Università di Milano, Via Celoria 16, I-20133 Milano, Italy Sergio L. Cacciatorib兲
Dipartimento di Matematica dell’Università di Milano, Via Saldini 50, I-20133 Milano, Italy and INFN, Sezione di Milano, Via Celoria 16, I-20133 Milano, Italy
Bianca L. Cerchiaic兲
Lawrence Berkeley National Laboratory, Theory Group, Bldg 50A5104, 1 Cyclotron Rd, Berkeley, California 94720-8162 and Department of Physics, University of California, Berkeley, California 94720
共Received 9 January 2006; accepted 6 March 2006; published online 25 April 2006兲
In this paper we reconsider the problem of the Euler parametrization for the unitary groups. After constructing the generic group element in terms of generalized angles, we compute the invariant measure on SU共N兲 and then we determine the full range of the parameters, using both topological and geometrical methods. In par- ticular, we show that the given parametrization realizes the group SU共N+1兲 as a fibration of U共N兲 over the complex projective space CPn. This justifies the inter- pretation of the parameters as generalized Euler angles. © 2006 American Institute of Physics. 关DOI:10.1063/1.2190898兴
I. INTRODUCTION
The importance of group theory in all branches of physics is a well-known fact. Explicit realizations of group representations are often necessary technical tools. Often it is finite dimen- sional and compact Lie groups and then the knowledge of the associated algebra, which describes the group in a neighborhood of the identity, is enough for this purpose.
There are however cases where an explicit expression of the full global group structure is needed, as, for example, when nonperturbative computations come into play. In most of these cases, the main objectives are two: First, one would like to find a relative simple parametrization, making all the computations manageable. Second, one needs to determine the full range of the parameters, in order to be able to handle global questions.
If both such points can seem unnecessary at an abstract level, they become essential at a most concrete level, e.g., in instantonic calculus or in nonperturbative lattice gauge theory computa- tions. The necessary computer memory for simulations is in fact drastically diminished.
The case of SU共N兲 was first considered and solved by Tilma and Sudarshan, in Ref. 1. There, they provide a parametrization, in terms of angular parameters, for the unitary groups. In particu- lar, in the first paper they consider special groups, SU共N兲, together with some applications to qubit and qutrit configurations. In the second paper, they give an extension to U共N兲 groups, using the fibration structure of SU共N+1兲 as U共N兲 fiber over the complex projective space CPn.
In this paper we reconsider the problem of finding a generalized Euler parametrization for special unitary groups. The intent is to provide a fully explicit and elementary共which does not mean short兲 proof of the beautiful results of Ref. 1. Our motivation is that the determination of the range of the parameters is a quite difficult task, so that disagreements are present in the literature
a兲Electronic mail: bertiste@tiscalinet.it
b兲Electronic mail: sergio.cacciatori@mi.infn.it
c兲Electronic mail: blcerchiai@lbl.gov
47, 043510-1
0022-2488/2006/47共4兲/043510/13/$23.00 © 2006 American Institute of Physics
even for SU共3兲 共for example, in Ref. 2兲. Therefore, we think that a careful deduction is necessary in order to corroborate the results of Tilma and Sudarshan. Also, all our proofs based essentially on inductive procedures, and they are explicit, in order to be easily accessible to anyone who needs them.
Our construction is quite different from Ref. 1, and as a result our parametrization differs slightly from theirs. However, this does not affect the final expression of the invariant measure.
To illustrate the spirit of our construction, let us start by taking a look at the Euler parametri- zation for SU共2兲.
Starting from the Pauli matrices
1=冉0 11 0冊,
2=冉0 − ii 0 冊,
3=冉10 − 10 冊, 共1.1兲it is known that the generic element of SU共2兲 can be written as
g = ei3ei2ei3. 共1.2兲 Here
苸关0,
兴,
苸关0,
/ 2兴,
苸关0,2
兴 are the so-called Euler angles for SU共2兲. They are related to the well-known Euler angles traditionally used in classical mechanics to describe the motion of a spin. From the point of view of the structure of the representation,共1.2兲 is obtained starting from a one parameter subgroup exp共i
2兲 and then acting on it both from the left and from the right with a maximal subgroup of SU共2兲 which does not contain the first subgroup. We can rewrite it in the schematic form g = U共1兲exp共i
2兲U共1兲. On the other hand, the group SU共2兲 is topologically equivalent to the three-sphere S3, and admits a Hopf fibration structure with fiber S1 over the base S2⯝CP1.To recognize this fibration structure in共1.2兲, we can apply the methods used in Refs. 3 and 4.
After introducing the metric 具A兩B典=12Tr共AB兲 on the algebra, the metric on the group can be computed as ds2=12Tr J丢J, where J = −ig−1dg are the left-invariant currents. Following Ref. 3, it is possible to separate the fiber from the base by writing g = hU共1兲, where h=ei3ei2 and U共1兲=ei3. To find the metric on the fiber, let us fix the point on the base and compute the currents along the fiber, JF= −iU共1兲−1dU共1兲=d
3. The metric on the fiber is then simply given by dsF2= d
2. To determine the metric on the base, we first must project out from the current JB= −ih−1dh the component along the fiber, in order to be left with the reduced current on the basis J˜B= d
2+ sin共2
兲d
1, which then in turn provides the metricdsB2=14关d共2
兲2+ sin2共2
兲d共2
兲2兴. 共1.3兲 This corresponds in fact with the metric of a sphere of radius 12. It is easy to see that, introducing the complex coordinates z = tan
eiand their complex conjugates, the metric dsB2 reduces to the standard Fubini-Study metric forCP1.This shows that the Euler parametrization captures the Hopf fibration structure of SU共2兲, which is the starting point for our construction. Mimicking what we said about SU共2兲, let us write the generic element of SU共N+1兲 as g=U共N兲e共
兲U共N兲, where e共
兲 is a one parameter subgroup not contained in U共N兲. The first difficulty we must face here is that this expression for a generic SU共N+1兲 group has redundancies, which must be eliminated. After this problem is solved, we then must show that the parametrization respects the Hopf fibration structure of SU共N+1兲.II. THE SU„N… ALGEBRA
The generators of su共N兲 are all the N⫻N traceless Hermitian matrices. A convenient choice for a base are the generalized Gell-Mann matrices as explained in Ref. 1. Let us remind how they can be constructed using an inductive procedure. Let 兵i其i=1N2−1 be the Gell-Mann base for su共N兲:
They are N⫻N matrices which can be embedded in su共N+1兲 adding a null column and a null row
˜i=
冉
0ជ
i 00ជ 冊
. 共2.1兲We will omit the tilde from now on. The dimension of SU共N兲 being 共N+1兲2− 1, we must add 2N + 1 matrices to obtain a Gell-Mann base for su共N+1兲. This can be done as follows: Set
兵N2+2a−2其␣=
␦
␣,a␦
,N+1+␦
␣,N+1␦
,a,共2.2兲 兵N2+2a−1其␣= i共−
␦
␣,a␦
,N+1+␦
␣,N+1␦
,a兲,for a = 1 , . . . , N. The last matrix we need is diagonal and traceless so that we can take共N + 1兲2−1
=
⑀
N+1diag兵1, ... ,1,−N其.One can easily verify that the base of matrices兵I其I=1共N + 1兲2−1 so obtained satisfies the normal- ization condition Trace兵IJ其=2
␦
IJif we choose⑀
N+1=冑2 /关共N+1兲2−共N+1兲兴.These are exactly the matrices we need to generate the group elements.
III. THE EULER PARAMETRIZATION FOR SU„N+1…: INDUCTIVE CONSTRUCTION It is a well-known fact that special unitary groups SU共N+1兲 can be geometrically understood as U共N兲 fibration over the complex projective space CPN. Now U共N兲 is generated by the first N2− 1 generalized Gell-Mann matrices plus the last one 共N + 1兲2−1. Using the fact that all the remaining generators of SU共N+1兲 can be obtained from the commutators of these matrices with
N2+1, one is tempted to write the general element of SU共N+1兲 in the form
SU共N + 1兲 = U共N兲eixN2+1U共N兲. 共3.1兲 However to describe SU共N+1兲 we need 共N+1兲2− 1 parameters, while in the rhs they are 2N2+ 1: There are共N−1兲2 redundancies. Inspired at first by dimensional arguments, we propose that an U共N−1兲 subgroup can be subtracted from the left U共N兲 in the following way. Let us write U共N兲 in the form U共N兲=SU共N兲ei共N + 1兲2−1. Inductively, we can think that also SU共N兲 can be recovered from U共N−1兲ei共N − 1兲2+1U共N−1兲 eliminating the redundant parameters, so that it will have the form SU共N兲=hei共N − 1兲2+1SU共N−1兲eiN2−1. We then choose to eliminate the appearing SU共N−1兲 together with the phase ei共N + 1兲2−1. In this way the SU共N+1兲 group element can be written in the form SU共N+1兲=hei共N − 1兲2+1eiN2−1eixN2+1U共N兲. By induction, assuming N艌2 we arrive to the final form of our Ansatz about the parametrization of the general element g 苸SU共N+1兲,
g = ei13ei12兿
a=2 N
关ei共a/⑀a兲a2−1eiaa2+1兴U共N兲关
␣
1, . . . ,␣
N2兴, 共3.2兲where U共N兲关
␣
1, . . . ,␣
N2兴 is a parametrization of U共N兲 which in turn can be obtained inductively using the factU共N兲 = 关SU共N兲 ⫻ U共1兲兴/ZN. 共3.3兲
The Ansatz共3.2兲 contains the correct number of parameters. However, we need to show that it is a good Ansatz, meaning that at least locally it must generate the whole tangent space to the identity. Using the Backer-Campbell-Hausdorff formula and some properties of the Gell-Mann matrices,共essentially the fact that the commutators of 共k − 1兲2+1 with the first共k−1兲2− 1 matrices generate all the remaining matrices of the su共k兲 algebra but the last one兲 it is easy to show that
ei11ei13兿
a=2 N
关ei共a/⑀a兲a2−1eiaa2+1兴 = ei兺j=1共N + 1兲2−2ajj, 共3.4兲
where ajare all nonvanishing functions of the 2N parameters
a,
a. Thus in a change of coordi- nates共from the
a,
a to the aj兲 only 2N of the ajcan be chosen as independent parameters. We could choose the last ones, corresponding to the coefficients of the matrices兵k其k=N2N2+2N−1
. In this way, the N2free parameters for the remaining matrices come out exactly from the U共N兲 factors in 共3.2兲.
We have not entered into details here because a second simple proof of the validity of this parametrization will be given by constructing a nonsingular invariant measure from our Ansatz.
IV. INVARIANT MEASURE AND THE RANGE OF THE PARAMETERS
A. The invariant measure
To construct the invariant measure for the group starting from共3.2兲, we will adopt the same method used in Ref. 3, with UªU共N兲 as the fiber group. Let us then write 共3.2兲 as
g = h · U. 共4.1兲
Starting from the computation of the left invariant currents jh= −ih−1dh, we can define the one forms
elª12Tr关jh·N2+l−1兴, l = 1, ... ,2N, 共4.2兲 which turns out to give the Vielbein one forms of the base space of the fibration. If eគ denotes the corresponding Vielbein matrix, the invariant measure for SU共N+1兲 will then take the form
d
SU共N+1兲= det eគ · d
U共N兲, 共4.3兲d
U共N兲 being the invariant measure for U共N兲. Using 共3.3兲 with U共1兲=ei共/⑀N+1兲共N + 1兲2−1we obtain the recursion relation共note that here
is allowed to vary in the range关0,2
/ N兴兲d
SU共N+1兲= det eគ · d
SU共N兲d
⑀
N+1. 共4.4兲
Then we will concentrate on the det eគ term. To this end let us write 共3.2兲 in the form
g = hN+1关
a,
a兴 · U关␣
i兴. 共4.5兲 Here we will consider N艌3 so that the relationhN+1= hNei共N/⑀N兲N2−1eiNN2+1, 共4.6兲 is true. If we introduce the right currents JhN+1= −ihN+1−1 dhN+1then the Vielbein共4.2兲 takes the form
el兵N其= 1
2Tr兵JhN+1l其 = d
N␦
l,N2+1+ 1 2⑀
Nd
NTr兵e−iNN2+1N2−1eiNN2+1N2+l其+1
2Tr兵e−i共N/⑀N兲N2−1JhNei共N/⑀N兲N2−1eiNN2+1N2+le−iNN2+1其, 共4.7兲 and using the relations in Appendix A we find
eគ兵N其=
冢
d
00N sin
Ncos00
Nd
N 12sinsin
NNTr关ecos−i
共NN12/⑀Na=2兺N兲0NTr2−1冋
Jh␣
N1eaiJ共hNN/⑀Na兲2−1N2册
−1Mគ 兴冣
, 共4.8兲where we introduced the兵N其 index to remember that this is a 2N⫻2N matrix associated to the group SU共N+1兲. Here Mគ is a column of matrices, M2j−1=j2+1, M2j=j2, j = 1 , 2 , . . . , N − 1. For- mula共4.8兲 then reads as follows: JhNis a 1-form with components JhN,c, c = 1 , 2 , . . . , 2n − 2, with respects to the coordinates Xc, defined as X2j−1=
j, X2j=
j. To find the component共r,c兲 of 共4.8兲 one must then take the cth component of JhN,cand the rth component of Mគ before to compute the trace.The invariant measure is then
det eគ兵N其= d
Nd
Ncos
Nsin2N−1
Ndet冉12Tr关e−i共N/⑀N兲N2−1JhNei共N/⑀N兲N2−1Mគ 兴冊. 共4.9兲We now use the recurrence relation
Jh
N=共N − 1兲2+1d
N−1+ 1⑀
N−1e−iN−1共N − 1兲2+1共N − 1兲2−1eiN−1共N − 1兲2+1d
N−1+ e−iN−1共N − 1兲2+1e−i共N−1/⑀N−1兲共N − 1兲2−1JhN−1ei共N−1/⑀N−1兲共N − 1兲2−1eiN−1共N − 1兲2+1. 共4.10兲 Computing the traces different cases arise depending on whether j = N − 1 or j⬍N−1; using again the relations in Appendix A it is not too difficult to show that the last determinant is equal to
det
冢
dd
N−1N−1cos共Nsin共N
NN兲兲 −1212cos共Nsin共N
NN兲sin共2兲sin共2
N−1N−1兲d兲d
N−1N−1冣
⫻ det冉12cos
N−1Tr关e−i共N−1/⑀N−1兲共N − 1兲2−1JhN−1ei共N−1/⑀N−1兲共N − 1兲2−1Mគ 兴冊,which set into共4.9兲 in turn yields the recurrence relation
det eគ兵N其= d
Nd
Nsin2N−1
Ntan2N−4
N−1det eគ兵N−1其, 共4.11兲
which can be solved to give
det eគ兵N其= 2 d
Nd
Ncos
Nsin2N−1
N兿a=1 N−1
关sin
acos2a−1
ad
ad
a兴. 共4.12兲This is the same result as found in Ref. 1.
B. The range of the parameters
At this point we are able to determine the range of the parameters in such a way as to cover the whole group. We will do this only for the base space: The remaining ranges for the fiber can be determined recursively, as discussed above, remembering in particular that the U共1兲 phase in U共k兲 can be taken in 关0,2
/ k兴.We then proceed as in Ref. 3. We first choose the ranges so as to generate a closed 关共N+1兲2− 1兴-dimensional closed manifold which then must wrap around the group manifold of
SU共N+1兲 an integer number of times. This can be done by looking at the measure 共4.12兲 on the base manifold and noticing that it is nonsingular when 0⬍
a⬍
/ 2, whereas
acan take all the period values
a苸关0,2
兴, for all a=1, ... ,N. However, note that the angles
1,
1,
2generate the whole SU共2兲 group when 0艋
1艋
, 0⬍
a⬍
/ 2 and 0艋
2艋2
. We can then restrict
1苸关0,
兴. The rest of the variety is generated by the remaining U共N兲 part.If we call VN+1the manifold obtained this way we then find
Vol共VN+1/U共N兲兲 =冕0
d
兿a=2 N 冕0
2
d
a兿b=1 N 冕0
/2
d
b再
cos
Nsin2N−1
NN−1兿c=1关sin
ccos2c−1
c兴冎
=
NN!, 共4.13兲
or equivalently
Vol共VN+1兲 = Vol共U共N兲兲
NN!. 共4.14兲
This is exactly the recursion relation found in Appendix B. Therefore, it is the correct range of the parameters for every N艌2, if we have V3= SU共3兲, as can be easily checked directly or by com- parison with the results given in Appendix A of Ref. 2共see also Appendix B of Ref. 4兲. The next step is to determine the parametrization of SU共N+1兲 for every value of N. It is given by 共3.2兲 with
0艋
1艋
, 0艋
a艋 2
,a = 2, . . . ,N,0艋
艋2
N , 0艋
a艋
2,a = 1, . . . ,N, 共4.15兲
and the remaining parameters which cover SU共N兲 共determined inductively兲.
To prove that our parametrization is well defined we can do more: We are in fact able to show that the induced metric on the base manifold is exactly the Fubini-Study metric overCPN. V. THE GEOMETRIC ANALYSIS OF THE FIBRATION
We will now show that the metric induced on the base space takes exactly the form of the Fubini-Study metric in trigonometric coordinates as given in Appendix C. To do so we will again use inductive arguments.
The metric on the base is dsB2=关eគ兵N其兴T丢eគ兵N其, where T indicates transposition and eគ兵N其is given in共4.8兲. Using the relations in Appendix A and defining
XN=1 2兺
a=2 N
Tr关JhN
⑀
aa2−1兴 共5.1兲the metric takes the form
dsB2= d2
N+ sin2
N再
关d
N+ XN兴2+N−1兺j=1冋
12Tr共e−i共N/⑀N兲N2−1JhNei共N/⑀N兲N2−1兲j2册
2+兺
j=1
N−1
冋
12Tr共e−i共N/⑀N兲N2−1JhNei共N/⑀N兲N2−1兲j2+1册
2冎
− sin4
N关d
N+ XN兴2. 共5.2兲This is an encouraging form, which upon comparison with 共C3兲 suggests the identification
=
N. With this identification in mind, let us first remark that the following recursion relation holds:XN= cos2
N−1共d
N−1+ XN−1兲, 共5.3兲 which can be shown by inserting 共4.10兲 in 共5.1兲 and then applying 共A6兲 and 共A11兲. A direct computation yieldsX3= cos2
2共d
2+ cos共2
1兲d
1兲, 共5.4兲 from which, through repeated application of the recurrence relation共5.3兲, we obtainXN=兺
k=1
N−3
冋
兿i=1k cos2
N−i册
d
N−k+冋
N−2兿i=1 cos2
N−i册
共d
2+ cos共2
1兲d
1兲. 共5.5兲At this point we must to compare d
N+ XNwith the coefficient of sin4
in共C3兲. In fact, to bring d
N+ XNto the desired form兺i=1N 共R˜i兲2d
i, one is tempted to just set
i=
iand
=
. However, this cannot be the case because the R˜idoes not satisfy the condition兺共R˜i兲2= 1.These observations, together with explicit calculations for the case N = 4 and N = 5, suggest that we should simply take some linear combination
i=
i共
j兲. This can be done as follows: Let us introduce new variables˜
K, k = 1 , . . . , N, such that
˜N=
N,
N−k=˜
N−k−˜
N−k+1, k = 1, . . . ,N − 3,共5.6兲
1+
2=
˜1−
˜3,
1−
2=˜
3−˜
2. In this way d
N+ XNtakes the desired formd
N+ XN=兺i=1 N
共Ri共
兲兲2d
i 共5.7兲with
=
,
= 1 , . . . N − 1,
i=˜
N−i+1, i = 1 , . . . , N andR1= sin
N−1, Rk= sin
N−k兿i=1 k−1
cos
N−i, k = 2, . . . ,N − 1,共5.8兲 RN=兿
i=1 N−1
cos
N−i.These formulas agree with the expressions in Appendix C. As the last step, in Appendix D we finally show that, after performing the change of variables described above, the coefficients of sin2
and sin2
N also agree. This proves that the metric induced on the base CPN of the U共N兲 fibration by the invariant metric on SU共N+1兲 is nothing else but the natural Fubini-Study metric in trigonometric coordinates.We can now use this result as a different method to fix the range of the parameters. In fact, 共R1, . . . , RN兲 parametrize the positive orthant of a sphere, if 0⬍
i⬍
/ 2, i = 1 , . . . , N − 1. More- over, the identification of
N with
yields
苸关0,
/ 2兴. Finally, it is easy to show that the conditions ˜
i苸关0,2
兴 are equivalent to
1苸关0,
兴 and
i苸关0,2
兴, i=2, ... ,N. These are the same results obtained in共4.15兲.VI. CONCLUSIONS
In this paper, we have reconsidered the problem of constructing a generalized Euler param- etrization for SU共N兲. The parametrization we find differs slightly from the one described by Tilma and Sudarshan. In fact, comparing our results with the expression共18兲 in Ref. 1, it is possible to
see that we have chosen共k − 1兲2−1instead of3. Furthermore, we have computed the corresponding invariant measure, which turns out to coincide with the result in Ref. 1, despite the slight differ- ences in the choice of the parametrization.
To determine the range of the parameters, we have used two distinct methods, both yielding the same result. To better motivate the name “Euler angles,” we have carefully shown that the parametrization captures the Hopf fibration structure of the SU共N兲 groups. In particular the change of coordinate we found to evidentiate the fibrations, gives an explicit map between the Euler coordinates introduced starting from the generalized Gell-Mann matrices, and the ones introduced in Ref. 5 using geometrical considerations.
We have given a quite explicit proof of every assertion. Apart from corroborating the results of Tilma and Sudarshan, we think that our work is providing a complete toolbox of computation techniques useful in applied theoretical physics as well as for experimental physicists.
ACKNOWLEDGMENTS
The authors共S.B. and S.C.兲 would like to thank G. Berrino, Professor R. Ferrari and Dr. L.
Molinari for helpful conversations. One of the authors共B.L.C.兲 is supported by the INFN under Grant No. 8930/01. This work was supported in part by the Director, Office of Science, Office of High Energy and Nuclear Physics, of the U.S. Department of Energy under Contract No. DE- AC02-05CH11231 and in part by the U.S. Department of Energy, Grant No. DE-FG02- 03ER41262. The work of the author共S.C.兲 is supported by INFN, COFIN prot. 2003023852គ008 and the European Commission RTN program MRTN-CT-2004-005104, in which the author共S.C.兲 is associated to the University of Milano-Bicocca.
APPENDIX A: SOME COMMUTATORS
Using the explicit form of the generalized Gell-Mann matrices constructed with the conven- tions of Sec. II, we find the useful commutators
关N2+1,N2+2j兴 = − ij2, 关N2+1,N2+2j+1兴 = ij2+1, 共A1兲
when j = 1 , . . . , N − 1.
Others interesting relations easy to check are
关N2+1,N2兴 = − i共N + 1兲
⑀
N+1共N + 1兲2−1− i兺a=2 N
⑀
aa2−1,关N2+1,a2−1兴 = i
⑀
aN2, 共A2兲关N2+1,共N + 1兲2−1兴 = i共N + 1兲
⑀
N+1N2,where a = 1 , . . . , N, from which remembering that
⑀
k=冑2 / k共k−1兲, one also finds关N2+1,关N2+1,N2兴兴 = 4N2. 共A3兲 From the first two commutators we find the very useful relations
eixN2+1j2+1e−ixN2+1= 1
sin xN2+2j+1− 1
tan xeixN2+1N2+2j+1e−ixN2+1,
eixN2+1j2e−ixN2+1= − 1
sin xN2+2j+ 1
tan xeixN2+1N2+2je−ixN2+1, 共A4兲 when j = 1 , . . . , N − 1.
Other useful relations easy to prove using the previous relations are
兺a=2 N
⑀
a2+共N + 1兲2
⑀
N+12 = 4, 共A5兲
兺a=2 N
⑀
a2+共N + 1兲
⑀
N+12 = 2, 共A6兲
Tr关e−ixN2+1N2−1eixN2+1N2+I兴 =
⑀
N␦
I0sin共2x兲, 共A7兲1
2Tr关aeixN2+1N2+2ie−ixN2+1兴 =
␦
a,i2sin x, a艋 N2− 1, i = 1, . . . ,N − 1, 共A8兲1
2Tr关aeixN2+1N2+2i−1e−ixN2+1兴 = −
␦
a,i2+1sin x, a艋 N2− 1, i = 1, . . . ,N − 1, 共A9兲1
2Tr
冋
eixN2+1N2e−ixN2+1N兺a=12−1Caa册
= sin共2x兲12b=2兺N Tr关Cb2−1⑀
b2−1兴, 共A10兲兺a=2 N
Tr关Aeix共N − 1兲2+1a2−1e−ix共N − 1兲2+1兴 = cos2x兺
a=2 N
Tr关A
⑀
aa2−1兴, 共A11兲eixN2−1共N − 1兲2e−ixN2−1= cos共N
⑀
Nx兲共N − 1兲2− sin共N⑀
Nx兲共N − 1兲2+1, 共A12兲eixN2−1共N − 1兲2+1e−ixN2−1= cos共N
⑀
Nx兲共N − 1兲2+1+ sin共N⑀
Nx兲共N − 1兲2, 共A13兲 where we used Aª兺i=1共N − 1兲2−1Aii.APPENDIX B: THE TOTAL VOLUME OF SU„k…
The total volume for the groups SU共k兲 can be found following Macdonald in Ref. 6. First remember that, in the sense of rational cohomology, SU共k兲 is equivalent to the product of odd dimensional spheres
SU共k + 1兲 ⬃兿
j=1 k
S2i+1, 共B1兲
where we choosen k + 1 to obtain recursive relations. The total volume of the group is then uniquely determined when a metric is established on the Lie algebra. We chose the metric induced by the scalar product共A兩B兲=12Tr共AB兲, for A,B苸su共k+1兲. In this way the Gell-Mann generators are orthonormal. The formula for the total volume is共Ref. 6兲
Vol共SU共k + 1兲兲 =兿
j=1 k
Vol共S2i+1兲 · Vol共Tk兲兿
␣⬎0兩
␣
Ú兩2, 共B2兲where
␣
Ú are the coroots associated to positive roots and Vol共Tk兲 is the volume of the torus generated by the simple coroots.For su共k+1兲 the simple coroots are si= Li− Li+1, i = 1 , . . . , k where Li is the diagonal matrix with the only nonvanishing entry兵Li其ii= 1. After writing siin terms of j, as
si=兺
a=1 k 1
2Tr兵si共a + 1兲2−1其共a + 1兲2−1, 共B3兲 it is easy to prove the recursive relation
Vol共Tk兲 =冑k + 12k Vol共Tk−1兲. 共B4兲 From this we find
Vol共SU共k + 1兲兲 = Vol共SU共k兲兲2
k+1k! 冑k + 12k , 共B5兲
where we used the fact that all the positive coroots have unitary length. If we note that the phase ei共k/⑀k兲共k + 1兲2−1 generates a U共1兲 group of volume 2
冑k共k+1兲/2 and that U共k兲=关SU共k兲⫻U共1兲兴/Zk, we can finally write
Vol共SU共k + 1兲兲 = Vol共U共k兲兲
kk!, 共B6兲
APPENDIX C: THE FUBINI-STUDY METRIC FOR CPN
CPNis a Kähler manifold of complex dimension N. In a local chart, which uses holomorphic inhomogeneous coordinates兵zi其i=1N 苸C, the Kähler potential is K共zi, z¯j兲=k/2 log共1+兺i=1N 兩zi兩2兲 with k a constant. The associated Kähler metric gij¯=
2K /
zi
¯zj is thendsCPN
2 = k
冉
1 +兺i=1N兺dzi=1N i兩zdz¯i兩i2−兺i,j=1N zidz¯i¯zjdzj共1 +兺i=1N 兩zi兩2兲2
冊
. 共C1兲Notice that obviously it is not possible to cover the whole space with a single chart, but the set of points which cannot be covered has vanishing measure. For our purpose it is therefore enough to consider a single chart.
Let us now search for a trigonometric coordinatization. To this aim let us introduce the new real coordinates
,
,
i,
= 1 , . . . , N − 1, i = 1 , . . . , N, such thatzi= tan