PERSISTENT CHALLENGES OF QUANTUM CHROMODYNAMICS

Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. International Journal of Modern Physics A Vol. 21, Nos. 28 & 29 (2006) 5695–5719 c World Scientific Publishing Company PERSISTENT CHALLENGES OF QUANTUM CHROMODYNAMICS∗ M. SHIFMAN William I. Fine Theoretical Physics Institute, University of Minnesota, Minneapolis, MN 55455, USA Received 27 October 2006 Unlike some models whose relevance to Nature is still a big question mark, Quantum Chromodynamics (QCD) will stay with us forever. QCD, born in 1973, is a very rich theory supposed to describe the widest range of strong interaction phenomena: from nuclear physics to Regge behavior at large E, from color confinement to quark–gluon matter at high densities/temperatures (neutron stars); the vast horizons of the hadronic world: chiral dynamics, glueballs, exotics, light and heavy quarkonia and mixtures thereof, exclusive and inclusive phenomena, interplay between strong forces and weak interactions, etc. Efforts aimed at solving the underlying theory, QCD, continue. In a remarkable entanglement, theoretical constructions of the 1970’s and 1990’s combine with today’s ideas based on holographic description and strong–weak coupling duality, to provide new insights and a deeper understanding. Keywords: Quantum chromodynamics; color confinement; hadron physics; review. Unlike some models whose relevance to Nature is still a big question mark, Quantum Chromodynamics (QCD) will stay with us forever. QCD is a very rich theory supposed to describe the widest range of strong interaction phenomena: from nuclear physics to Regge behavior at large E, from color confinement to quark–gluon matter at high densities/temperatures (neutron stars); the vast horizons of the hadronic world: chiral dynamics, glueballs, exotics, light and heavy quarkonia and mixtures thereof, exclusive and inclusive phenomena, interplay between strong forces and weak interactions, etc. Given the remarkable variety of phenomena governed by QCD dynamics, it seems unlikely that an exact solution will be ever found. But do we really need it? ∗ Julius Edgar Lilienfeld Prize Lecture at the April Meeting of American Physical Society, Dallas, Texas, April 22–25, 2006. 5695 5696 M. Shifman Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Birth and Adolescence QCD was born in 1973, with the discovery of asymptotic freedom by David Gross, Frank Wilczek and David Politzer. This discovery was marked by the Nobel Prize in 2004. In three decades that elapsed from the beginning of this exciting journey, QCD went a long way. Although its full analytic solution has never been found (and, most likely, never will be), the progress is enormous, and so are the problems which still await their solutions. From success to challenge to new discovery — this is the logic. I was asked to prepare the Lilienfeld Prize Lecture. This talk gives me a good opportunity to summarize the main elements of a big picture that emerged after 1973 and outline some promising problems for the future, as I see it now. Rather than aiming at an exhaustive coverage — which would certainly be impossible — I will focus on trends drawing them in “broad touches.” I will cite no original works, referring the reader to selected books, review papers and lectures which can fill this gap. What came out? Something like “A Brief History of Quantum Chromodynamics.” This is not a treatise of an impartial historian. I am certainly biased and tend to emphasize those contributions which produced a strong impact on me personally.a To ease my task, I will divide the subject into three time intervals, covering three decades — from 1973 to 1983, from 1984 to 1993, and from 1994 to the present, to be referred to as Eras I, II, and III, respectively. The status of QCD by the end of Era II is summarized in Ref. 1, and its status at the beginning of the new millennium in Refs. 2 and 3. The most recent developments are reviewed in Ref. 4. The first triumph that came with the creation of QCD was understanding those processes where the dominant role belongs to short-distance dynamics, such as deep inelastic scattering, or the total cross-section of the e+ e− annihilation. The fact that such processes could be described, to a good approximation, by the quark– gluon perturbation theory, was noted by the fathers of QCD. The reason is the famous asymptotic freedom: the effective quark–gluon coupling becomes weak at short distances. The boundary between weak coupling and strong coupling lies at Λ−1 where Λ is a dynamical scale not seen in the Lagrangian. It occurs through a dimensional transmutation. The phenomenon of asymptotic freedom is very counterintuitive. Generations of field theory practitioners believed that in any field theory a probe charge placed in vacuum gets screened by opposite charges appearing from vacuum fluctuations. This is intuitively clear. This is certainly the case in quantum electrodynamics (QED). If so, the effective charge seen by a large-distance observer falls off with distance, leading to infrared freedom. Remarkable as it is, in QCD (and non-Abelian gauge theories at large) it is not screening but rather antiscreening takes place. The origin a During the APS Meeting presentation I skipped many topics to meet the time constraint; to make up for that I added quite a large number of pictures which are omitted in the written version. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5697 of antiscreening is hard to visualize. Perhaps, that is the reason why the discovery of asymptotic freedom was such a surprise. Unlike all “conventional” theories, in QCD the effective coupling constant falls off at short distances — the opposite of infrared freedom of QED.b In the early days of QCD people referred to this phenomenon as infrared slavery. Although conceptually similar to that of QED, the quark–gluon perturbation theory is technically more contrived. Understanding how to use perturbation theory when color is permanently confined at large distances, and quarks and gluons do not appear in the physical spectrum, as well as adequate techniques, emerged gradually.5–8 Perturbative QCD, or pQCD as it became known later, currently deals with a broad range of issues, from ∆T = 1/2 rule in kaon decays to small-x physics at HERA, from widths of heavy quarkonia to jet physics. In spite of an advanced age, this area continues to grow: recently, insights and inspirations from string theory resulted in an explosive development in the multiparton amplitudes. I will review this issue later. In spite of remarkable successes in pQCD, an issue of great practical importance is not yet solved. In any short-distance-dominated process there is a stage where quarks and gluons are transformed into hadrons. The corresponding dynamics are essentially Minkowskian. Even if theoretical pQCD predictions can be formulated in terms of Euclidean quantities (such as the moments of the structure functions in deep inelastic scattering), the nonperturbative nature of QCD shows up in the form of exponential (in momentum transfers) corrections which are very difficult to control theoretically. The situation becomes much worse in those processes in which no Euclidean description is available, for instance, in jet physics. Or, if we need to know the structure functions themselves, rather than their moments. In this case pQCD results must be supplemented by corrections which are likely to be oscillating and suppressed by powers of large energies and momentum transfers, rather than exponentially. It is these largely unknown corrections that limit the accuracy of theoretical predictions so that in many instances they lag behind experimentally achieved accuracy. The problem goes under the name of quark–hadron duality violation. 9 It presents a serious and persistent challenge inherited from Eras I and II, a stumbling block which is impossible to bypass. A spontaneously broken axial symmetry in hadronic physics resulting in the occurrence of the (pseudo-)Goldstone bosons was conjectured in the 1960’s, well before the advent of QCD. In fact, in 1957 Marvin Goldberger and Sam Treiman studied the nucleon matrix element of the axial current, including the pion pole at t = Mπ2 . Assuming the pole dominance they obtained the celebrated Goldberger–Treiman relation gπN N = gA MN Fπ−1 . In 1960 Nambu identified pions bI should add that QED is logically incomplete because of the Landau zero charge. It must be viewed as a part of a larger asymptotically free theory. At the same time, QCD is perfect by itself, with a single exception of the CP problem which will be discussed below. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5698 M. Shifman as (pseudo-)Goldstone bosons. A rapid development of the soft-pion technique ensued, allowing one to analyze a large number of processes in low-energy hadronic physics. The advent of QCD gave new life to all these studies. It would be fair to say that a macroscopic approach was replaced by a microscopic one. As an example, let me mention a theory-defying enhancement of ∆T = 1/2 amplitudes in K → 2π, 3π decays, observed in the late 1950’s. It remained a mystery for years. Wilson’s renormalization group ideas10 applied in QCD, in conjunction with the lightness of the u, d and s quark masses,11 led to a discovery of the “penguin graphs” (Shifman, Vainshtein, Zakharov, 1974) giving rise to ∆T = 1/2 operators with a mixed chiral structure that are indeed strongly enhanced.12 Needless to say, without knowledge of underlying dynamics nothing can be said as to why the axial SU(3)flavor symmetry (for u, d, s quarks) is spontaneously broken. Since QCD is the theory of hadrons, it should explain this phenomenon. In 1980 Coleman and Witten, combining the ’t Hooft matching condition with the ’t Hooft large-N limit (which I will discuss shortly), proved that the axial symmetry must be spontaneously broken, indeed. At that time, calculation of the order parameter, the quark condensate hq̄qi, was beyond reach. During Era I, the soft pion technique13–15 evolved into a well-organized system combining two structural elements: effective low-energy Lagrangians and chiral perturbation theory. A highlight of this evolution line was Witten’s discovery in 1983 of the fact that the chiral Lagrangian supports solitons — they had been known as Skyrmions — which could be treated quasiclassically in the ’t Hooft large-N limit. Edward Witten demonstrated that the quasiclassical Skyrmions (collective excitations of the Goldstone bosons) are in one-to-one correspondence with baryons of multicolor QCD. This gave rise to the Skyrmion paradigm,16,17 a model of baryons which experienced an explosive development in the beginning of Era II. Much later, at the end of Era II, it was realized that one could use chiral Lagrangians to describe the interaction of soft pions with hadrons containing a heavy quark.18 The Skyrme model presents an elegant description of the QCD baryons at large N . At the same time, it carries a challenge. Assume that we replace conventional massless quarks in the fundamental representation of SU(N )color by unconventional quarks, in a different representation of color, e.g. two-index antisymmetric.c The pattern of the spontaneous breaking of the chiral symmetry in this gedanken case is well known. The corresponding chiral Lagrangian is not drastically different from that of QCD. It supports Skyrmions too. And yet — in this case there is no apparent match between Skyrmions and baryons.19 Why? A possible way out was suggested by Stefano Bolognesi just a few days ago. Now, it is time to dwell on one of the most crucial developments of Era I — the invention of the ’t Hooft 1/N expansion in 1974. It was further extended by Witten in 1979.20–22 c In the actual world N = 3. At N = 3 the quark in the fundamental representation is identical to that in the two-index antisymmetric representation. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5699 Why it is so hard to deal with QCD, and why are new advancements so painfully slow? This is due to the fact that in the vast majority of the hadronic problems there is no apparent expansion parameter. In hard processes Λ/E plays the role of such a parameter. This explains the successes of pQCD. However, the core of the hadronic physics operates with a different set of questions, for instance, what are the values of the ρππ constant, ωφ mixing, Σ-hyperon magnetic moment? To which extent are the Regge trajectories linear? Can one calculate their slopes? What can be said about glueballs and why they are so resilient against experimental detection? Where are four-quark states and pentaquarks? What is the structure of the newly discovered charmonium resonances? This list goes on and on . . . . In all these cases we do not see any obvious expansion parameter. Gerard ’t Hooftd came up with a brilliant idea that the number of colors N (in our world N = 3) can be treated as a large parameter. Consider multicolor QCD with the gauge group SU(N ), instead of SU(3), in the limit N → ∞, while the product λ ≡ g 2 N fixed, where g 2 is the gauge coupling constant. This limit is referred to as the ’t Hooft limit, and λ as the ’t Hooft coupling. The quarks are assumed to lie in the fundamental representation of SU(N ). A remarkable feature of the ’t Hooft 1/N expansion is that each term of the expansion is in one-to-one correspondence with topology of the relevant Feynman graphs. The leading order in 1/N describes all planar graphs, the next-to-leading order all graphs that can be drawn on a surface with one handle (torus), the next-tonext-to-leading order requires two handles, and so on. Moreover, each extra quark loop is suppressed by 1/N . Thus, multicolor QCD (in the ’t Hooft limit) is significantly simpler than QCD per se. Although consideration of planar graphs dramatically reduces the number of graphs, this is still a vast class of diagrams. Despite numerous attempts, no solution of planar QCD was ever found. Nevertheless, the 1/N expansion proved to be a powerful tool. At the qualitative level it allowed one to understand a variety of regularities inherent to the hadronic world which seemed rather mysterious for years. These regularities are: an infinite number of the meson resonances for given J P C and given flavor content; the Zweig rule (suppression of transitions between the q̄q pairs of different flavors); a relative smallness of the meson widths; the rarity of the four-quark mesons, and so on. The general picture emerging from the 1/N expansion reminds one of the dual resonance model of the 1960’s and early 1970’s which gave birth to string theory. 23 This parallel, noted already by ’t Hooft, is no accident. It gave hope that a stringbased description of “soft” QCD could be found. We will discuss QCD strings later; here I would like to note that today this dream of generations of QCD practitioners no longer seems Utopian, although, most likely, the equivalence will not be exact. Jumping ahead of myself, I will add that a version of planar QCD has been recently d By the way, ’t Hooft in English means the Head; isn’t it symbolic? Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5700 M. Shifman proven to be equivalent to supersymmetric Yang–Mills (SYM) theory,19 with rather nontrivial consequences that ensued immediately. The 1/N expansion as we knew it at that time, was applicable for qualitative, not quantitative explorations, with the single exception of the η ′ meson problem, or the puzzle of the missing ninth (pseudo-)Goldstone boson, of which I will speak later. A quantitative (or, at least, a semiquantitative) method allowing one to address many questions of hadronic physics from the list presented above was invented in 1978. It goes under the name of Shifman–Vainstein–Zakharov (SVZ) sum rules. Although the name is quite awkward, the underlying idea is simple and transparent. The most peculiar features of QCD, such as color confinement and spontaneous breaking of the chiral symmetry, critical for the formation of the hadronic spectrum and basic hadronic characteristics, must be reflected in the structure of the QCD vacuum. Although this structure is contrived, with luck its salient features could be encoded in a few of the “most important” vacuum condensates (I already mentioned one of them, hq̄qi; another is the gluon condensate). If so, one could try to relate a wealth of the low-energy hadronic parameters to these few condensates 24 through the operator product expansion (OPE). The notion of factorization of short and large distances, the central idea of OPE, was borrowed from Ken Wilson. The focus of Wilson’s work was on statistical physics, where the program is also known as the block-spin approach. Surprisingly, in high-energy physics of the early-to-mid 1970’s the framework of OPE was essentially narrowed down to perturbation theory. Seemingly, we were the first to adapt the general Wilsonian construction to QCD to systematically include powersuppressed effects, thus bridging the gap between short and large distances. This “bridging” did not lose its significance till this day. I will comment more on that later, in connection with AdS/QCD. This route — matching between the short distance expansion and long distance representation — led to remarkable successes. The SVZ method was tested, and proved to be fruitful in analyzing practically every static property of all established low-lying hadronic states, both mesons and baryons. Needless to say, with just a few vacuum condensates included in the analysis one cannot expect predictions to be exact, they are bound to be approximate. However, in many instances agreement between theoretical results and experimental data exceeded optimistic expectations. As usual there was a cloud on the horizon, a challenge which gave rise to a new development. We discovered that channels with the vacuum quantum numbers (more exactly, J P = 0± ) are drastically different from all others. In the 1981 papere entitled Are All Hadrons Alike? , we observed that these were precisely the channels where the 1/N counting fails too. Indeed, the flavor mixing in the scalar q̄q mesons is maximal, and so is mixing with the gluon degrees of freedom. There is no trace of the Zweig rule. In the 0− q̄q channel Veneziano and Witten predicted Mη2′ to e This paper was written with V. Novikov. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5701 be suppressed by 1/N , while in actuality it exceeds Mρ2 which does not scale with N . The scalar glueball whose decay width is predicted to be suppressed by 1/N 2 is in fact much broader than, say, the ρ meson whose decay width ∼ 1/N , etc., etc., etc. On the other hand, in the same paper we noted that in these particular channels the impact of “direct” instantons (instantons will be discussed shortly) is the strongest. If in all other cases, by and large, it could be neglected in the domain of validity of the SVZ sum rules, for the 0± quarkonia and glueballs the dominant nonperturbative effect was obviously correlated with the instantons. In a bid to quantify this circumstance, Shuryak; and Diakonov and Petrov engineered the instanton liquid model.25 Now I have to return to 1975 when Belavin, Polyakov, Schwarz and Tyupkin (BPST) discovered instantons in non-Abelian Yang–Mills theories, only two years after the advent of QCD. Originally Sasha Polyakov hoped that instantons could solve the problem of confinement. Although it did not happen that way (at, least, not in four dimensions) the conceptual impact of instantons was radical. First of all, they revealed a nontrivial vacuum structure in non-Abelian Yang–Mills theories. They demonstrated that an infinitely-dimensional space of fields has one particular direction which is topologically nontrivial; it is curled up in a circle. Quantum mechanics of systems living on a circle is peculiar. As well known from solid state physics, in such systems one has to introduce a hidden parameter of an angular type, a quasimomentum, which is not determined from the Lagrangian, but, rather, from the boundary conditions on the Bloch-type wave functions. In QCD this parameter is called the θ angle, or the vacuum angle. Instantons represent tunneling trajectories (in imaginary time) winding around the circle. The tunneling interpretation and the necessity of the emergence of the θ parameter was suggested by Gribov; Callan, Dashen and Gross; and Jackiw and Rebbi, independently, shortly after the BPST work. Instantons are quasiclassical objects. The qualitative insight they provide is difficult to overestimate. However, in the quantitative aspect the BPST instantons (or, more generally, the so-called instanton gas) proved to be rather useless in QCD. The reason is obvious: QCD — the real thing — is governed by strong coupling. And still, can one make definite predictions regarding the θ dependence of physical quantities in the hadronic world? Quite an exhaustive answer to this question was given on the basis of QCD low-energy theorems. Low-energy theorems are familiar to field theorists from the 1950’s. QCD gave rise to new ones, which were found, one by one, in the late 1970’s. Using them as a tool, Witten in 1980 exposed quite a sophisticated θ dependence of the QCD vacuum. Much later, in 1998 (Era III) he significantly advanced understanding of this issue, this time using a string perspective. On the field theory side, one can apply supersymmetry-based methods, see, for example, Ref. 26. They are especially fruitful at large N and fully confirm Witten’s conclusions regarding the intertwined vacuum family and the corresponding θ dependence consisting of N branches. Unfortunately, here I have no time to dwell on this topic. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5702 M. Shifman The advent of QCD put the theory of hadrons on solid footing. It brought a new problem from an unexpected side, however. Before QCD people believed CP conservation to be a natural feature of strong interactions. Alas, it is lost in QCD if θ 6= 0 and the quark masses do not vanish (we know they do not). At θ 6= 0 the theory breaks P and T symmetries. From the absence of CP breaking in strong interactions one concludes that experimentally θ < 10 −9 . One can hardly think that this incredible smallness of θ is just an accident. Can one find a reason for it? For quite some time Polyakov thought that this was not an issue. No matter what the bare value of the vacuum angle θ0 at the ultraviolet scale is, it will be screened to zero by the same effects that lead to color confinement at large distances. Polyakov even asked a student of his to prove this hypothesis. Well, this dream never came true. In 1980 we proved f that the observability of CP -odd effects at θ0 6= 0 is in one-to-one correspondence with the solution of U(1) problem. Namely, assuming that θ0 is completely screened would require restoring the Goldstone status of η ′ . Since this is impossible on empiric grounds (this was shown by S. Weinberg in 1974), θ0 cannot be screened. We are back to square one. Peccei and Quinn suggested an elegant way out — a mechanism that would screen θ0 no matter what. A “vacuum relaxation” and vanishing of the physical θ term is automatic in this mechanism. Almost immediately Weinberg and Wilczek noted that the idea leads, with necessity, to a new particle, the axion. Their magnificent work described a cute, little, almost massless axion which was good in all respects, except that it was incompatible with data. It was not so difficult to eliminate this shortcoming. In 1980 we introduced a phantom axion, a version of what is now called an “invisible axion” (this was done simultaneously and independently by Jihn Kim27,28 ). The invisible axion became a standard feature of the present-day theory. There are two versions of invisible axions; both preserve positive features of the original axion and, simultaneously, avoid unwanted contradictions.g Above I have mentioned the U(1) problem more than once. It is also referred to as the problem of the missing ninth Goldstone meson. The problem dates back to pre-QCD years, when current algebra was one of just a few tools available to theorists in strong interactions. The essence of the issue is excellently summarized in Steven Weinberg’s talk at the XVII International Conference on High Energy Physics.29 With three massless quark flavors one can construct nine (classically conserved) axial currents — eight forming a flavor octet, plus a flavor-singlet current. None of the corresponding symmetries is realized linearly in nature. As far as the flavor-octet currents are concerned, the corresponding symmetries are spontaneously broken. This implies the emergence of eight (pseudo-)Goldstone bosons which are very well known: π, η and K. However, there is no Goldstone boson that f By we I mean Vainshtein, Zakharov and myself. invisible axion of the second kind was devised by Dine, Fischler, Srednicki; and Zhitnitsky. g The Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5703 would correspond to the flavor-singlet current. A natural candidate, the η ′ meson, is too heavy to do the job. The question of where the ninth Goldstone boson hides was a big mystery. In 1975 ’t Hooft was the first to note that the GG̃ anomaly in the divergence of the flavor-singlet current is not harmless; it pushes out the Goldstone pole from the physical sector of the theory to an unphysical gauge noninvariant part of the Hilbert space. Thus, the particle spectrum of QCD was not supposed to contain the ninth Goldstone boson in the first place. In 1979 Witten and Veneziano made the next step. Using ’t Hooft’s 1/N expansion and the chiral anomaly formula they managed to obtain an expression relating the η ′ mass to the topological susceptibility of the vacuum in pure Yang–Mills theory (without quarks). They found that in ’t Hooft’s 1/N expansion m2η′ scales as 1/N . Moreover, later the vacuum topological susceptibility was calculated in the instanton liquid model and on lattices. A reasonably good agreement with the empiric value of m2η′ was obtained. Now I turn to the most important aspect of QCD (after the discovery itself) — color confinement. The founding fathers of QCD — Gross, Wilczek and Politzer — after observing the growth of the gauge coupling constant at large distances, speculated that this growth might be responsible for the fact that quarks and gluons, clearly detectable at short distances, never appear as asymptotic states. All hadrons that are seen in nature are color-singlet combinations of the quark and gluon fields. Experiment as well as computer simulations in lattice QCD show that if one considers a quark–antiquark pair separated by a distance L the energy of this system grows linearly with L. However, Gross, Wilczek and Politzer could not suggest a mechanism that would explain this phenomenon, linear confinement. Are we aware of any dynamical systems that could model or serve as analogs for this phenomenon, inseparability of constituents? In the mid-1970’s Nambu, ’t Hooft and Mandelstam put forward a hypothe30,31 sis which goes under the name of the dual Meissner effect (why it is dual will become clear shortly). They were inspired by a natural phenomenon which takes place in superconductors. What happens to a superconducting sample if it is placed in a magnetic field? As well known, superconductors expel magnetic flux. A superconducting medium tolerates no magnetic field inside. Assume we have two long magnets with wellseparated plus and minus poles. Or, better still, we find a couple of magnetic monopoles (of opposite magnetic charges) somewhere in space and bring them here to experiment with them. Next, suppose we insert this monopole–antimonopole pair into a superconducting sample, and place them at a large distance L from each other. The monopole is the source of the magnetic flux, the antimonopole is a sink, and in the empty space the flux would spread out to create a Coulomb attraction. However, inside the superconductor the magnetic flux cannot spread out, since it is expelled from the superconducting medium. The energetically favorable solution to this problem is as follows: a thin flux tube forms between the monopole and Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5704 M. Shifman the antimonopole. Inside this tube, known as the Abrikosov vortex,h superconductivity is ruined. The Abrikosov flux tube has a nonvanishing energy per unit length, a string tension. Once the string is formed the energy needed to separate the monopole and antimonopole grows linearly with L. In superconductivity, the formation of the Abrikosov tubes carrying a quantized flux of the magnetic field is called the Meissner effect. Unlike QED, the gauge group in QCD is non-Abelian. The quarks are sources of the chromoelectric field, rather than chromomagnetic. Thus, color confinement of quarks through string formation would require chromoelectric flux tubes. This is why the Nambu–’t Hooft–Mandelstam conjecture represents the dual Meissner effect. The Meissner effect assumes condensation of the electric charges and confinement of magnetic monopoles. The dual Meissner effect assumes condensation of “chromomagnetic” charges and confinement of “chromoelectric” objects. In the 1970’s and 1980’s the conjecture was nothing more than a vague idea, since people had no clue as to non-Abelian monopoles and non-Abelian strings. Any quantitative development was out of reach. Through Era II I will sail rather quickly through the 1980’s since these were relatively quiet years for QCD. I will dwell on just a few developments. So much was said about the construction of consistent OPE in QCD because, after this was done in connection with the SVZ sum rules, it gained a life of its own! The very same OPE constitutes the basis of the heavy quark expansions which blossomed in the 1990’s in the framework of the heavy quark theory.18,32 This is a branch of QCD where a direct live feedback from experiment still exists, which gives special weight to any advancement in theoretical understanding and accuracy of predictions. Conceptually the expansion in inverse powers of the heavy quark masses m Q is similar to other applications of OPE. Technically, exploring physics of mesons with open charm/beauty one has to deal with a number of peculiarities. The vacuum condensates are replaced by expectation values of certain local operators over the heavy meson states. The most important are the kinetic energy and chromomagnetic operators which are responsible for corrections proportional to m−2 Q . The OPE-based description of heavy hadrons, such as B mesons, conceived in the 1980’s was further expanded in the early-to-mid 1990’s, with new elements added and fresh findings incorporated. One such finding was a heavy quark symmetry. It is also known as the Isgur–Wise symmetry. A special case of this symmetry, manifesting itself in the b-to-c transition at zero recoil, was worked out earlier by Voloshin and myself. The Isgur–Wise consideration covers generic kinematics in h Sometimes it is also referred to as the Abrikosov–Nielsen–Olesen flux tube, or the ANO string. Nielsen and Olesen considered this topological defect in the context of relativistic field theory. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5705 the limit mQ → ∞. The heavy quark symmetry combining both flavor and spin symmetries acts in the heavy quark sector which during Era II became the focus of experimental studies in which unprecedented accuracy was achieved. By and large I can say that in the 1990’s a quantitative theory of decays of c and b-flavored hadrons was constructed that successfully matched the experimental accuracy. Many people contributed to this success. Among others I would like to mention Georgi; Bigi, Uraltsev and Vainshtein (I belonged to this group too), Manohar, Wise; and Voloshin. Let me single out one of the most elegant results established in this way in the heavy quark physics: the absence of the 1/mQ correction to the inclusive decay widths of the heavy-flavor hadrons. This theorem (the Bigi–Uraltsev–Vainshtein theorem) made its way into textbooks, let alone its practical importance for the precision determination of Vcb from data. The second development is a significant advancement of the 1/N ideas in application to baryons. It was noted by Gervais and Sakita (1984) and then thoroughly developed by Dashen, Jenkins and Manohar (1993–1994, see Ref. 33) that a large number of model-independent relations among baryonic amplitudes follow from large-N consistency conditions. The essence of these relations is as follows: at N → ∞ and ms → 0 the SU(6) spin-flavor symmetry that connects the six states u ↑, u ↓, d ↑, d ↓, and s ↑, s ↓ becomes exact, implying mass and width degeneracies among baryons of various quantum numbers, as well as relations for magnetic moments, axial couplings, and so on. Corrections to this limit can be systematically treated by combining 1/N and ms expansions. Next, we witnessed a gradual development — spanning at least a decade — of the instanton liquid model which grew into a consistent many-body (four-dimensional) problem that was solved numerically by Shuryak and collaborators. One may view it as a summation of fermion interactions to all orders in the ’t Hooft instantoninduced vertex. The fourth development to be mentioned here proved to be influential, in hindsight, although it was not perceived as such at the time. I mean the inception of supersymmetry-based methods in gauge theories at strong coupling.34 The inception of these ideas can be traced back to 1983, when the exact β function (the so-called Novikov–Shifman–Vainshtein–Zakharov, or NSVZ β function) was found in supersymmetric gluodynamics, and the first exact calculation of the gluino condensate was carried out. The basic ingredient of the above work was the use of holomorphy in the chiral sector of the SUSY gauge theories. In 1984 Affleck, Dine and Seiberg added light matter fields and came up with a beautiful superpotential emerging in supersymmetric theories with Nf = N − 1 which bears their name. (Subsequent numerous results of this group were focused mainly on the issue of spontaneous SUSY breaking. This topic lies beyond the scope of the present paper.) The issue of whether the 1983 exact result for the gluino condensate was also correct continued to preoccupy Arkady Vainshtein and me. In 1987 we engineered a Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5706 M. Shifman strategy main elements of which could be considered as precursors of the advancedto-perfection Seiberg and Seiberg–Witten programs. Although our final target was strongly coupled supersymmetric gluodynamics, we deformed the theory by introducing additional matter with a small mass term m, in such a way as to guarantee full Higgsing of the theory. Then it became weakly coupled. Building on the Affleck– Dine–Seiberg superpotential we exactly calculated the gluino condensate at weak coupling, where each and every step is under theoretical control. We then used the holomorphic dependence of the gluino condensate on the mass parameter to analytically continue to m → ∞, where the original supersymmetric gluodynamics is recovered.i By itself, this was a modest result. It is not the gluino condensate itself, but, rather, the emerging methods of SUSY-based analyses that had serious implications in the 1990’s. Concluding this part I would like to mention Seiberg’s 1988 calculation of the leading nonperturbative correction in the prepotential of N = 2 SUSY Yang–Mills theory — apparently, a starting point of a journey which culminated in 1994 when Seiberg and Witten found their celebrated solution for N = 2 theories. Maturity It would be fair to say that Era III started with Seiberg–Witten’s breakthrough. String theorists seemingly adore the word revolution, at least with regards to their own discipline. I do not like it because in real life revolutions never solve problems; instead, they only bring suffering. That is why, in characterizing the Seiberg– Witten construction and its consequences, the most appropriate phrase that comes to my mind is a long-awaited breakthrough. They considered SU(2) super-Yang– Mills theory with extended supersymmetry, N = 2. Extended SUSY is even more powerful than the minimal one. Basing on holomorphy, analytic properties following from extended supersymmetry and continuation from weak to strong coupling, Seiberg and Witten essentially solved the theory in the chiral sector at low energies. They proved that SU(2) is spontaneously broken down to U(1) everywhere on the moduli space. Thus, the magnetic monopoles and dyons are supported everywhere on the moduli space; at certain points they become massless. Deforming the theory by introducing a small mass term breaking N = 2 to N = 1 Seiberg and Witten forced the monopoles (dyons) to condense at these points, triggering the dual Meissner effect. This was the first honest-to-God demonstration ever that the dual Meissner effect can indeed take place in non-Abelian gauge theories. Shortly after, in 1998, Hanany, Strassler and Zaffaroni discussed formation and structure of the electric flux tubes in the Seiberg–Witten model which, being stretched between probe charges, confine them. Linear confinement in fourdimensional non-Abelian theory became a reality! i En route, the so-called 4/5 problem surfaced, which is not solved till today. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5707 At this time, euphoria of the first breakthrough years gave place to a more sober attitude. A more careful examination showed that details of the Seiberg– Witten confinement are quite different from those we expect in QCD-like theories. This is due to the fact that in the Seiberg–Witten solution the SU(N ) gauge symmetry is spontaneously broken in two steps. At a high scale SU(N ) is broken down to U(1)N −1 . Then complete breaking occurs at a much lower scale, where the monopoles (dyons) condense. Correspondingly, the confining strings in the Seiberg–Witten model are, in fact, the Abelian strings of the Abrikosov–Nielsen– Olesen type. This results in a “wrong” confinement; the “hadronic” spectrum in the Seiberg–Witten model is much richer than that in QCD. Only recently people started getting ideas about non-Abelian strings, so far mostly at weak coupling (for reviews see Refs. 35 and 36). Hanany and Tong; and Auzzi, Bolognesi, Evslin, Konishi and Yung found in 2003 that such strings are supported in certain regimes in N = 2 supersymmetric gauge theories. Their most crucial feature is that they have orientational zero modes associated with rotation of their color flux inside a non-Abelian SU(N ). These orientational modes make these strings genuinely non-Abelian. They are supposed to be dual to QCD strings. Shifman and Yung; and Hanany and Tong observed in 2004 that the above nonAbelian strings trap non-Abelian magnetic monopoles. In the dual description the trapped magnetic monopoles should be identified as gluelumps, of which lattice QCD practitioners have been speaking since 1985.37 A relatively simple weakly coupled non-Abelian model was found which can serve as a laboratory for studying the Meissner effect in a controllable setting (Gorsky, Shifman, Yung, 2004). So far I have scarcely mentioned lattice QCD. This is not because I do not value its achievements but, rather, because I believe this is a totally different discipline than analytic QCD. Over the years lattice practitioners have invested ample efforts in numerical studies of QCD strings. The very fact of their existence was firmly confirmed. At the same time, properties of the QCD strings, especially fine structure, are not easily accessible in numerical simulations. I will give just one example, k-strings. The notion of the k-strings was introduced largely in the context of lattice QCD.37 These are the strings that connect probe color sources with n-ality k. For instance, if we use two very heavy quarks sitting on top of each other as the first color source, and two heavy antiquarks as the second, the string forming between them is the 2-string. Until the present day the lattice studies did not reveal a fundamental property of QCD prescribing the k-string tensions to depend only on the n-ality, rather than on the particular representation of the probe color sources. For the abovementioned 2-strings one gets symmetric and antisymmetric representation of color. These representations are not identical, but the string tension σ2 is expected to be the same in both cases. This is expected but is not observed. In 2003 Adi Armoni and I revisited this long-standing problem (reviewed in Ref. 38). Our consideration was based on 1/N expansion implying a wealth of quasistable, “wrong” strings. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5708 M. Shifman It was realized that although the “wrong” excited strings eventually must decay into the “right” ones, whose string tension depends only on the n-ality, in many instances the lifetimes of the excited strings scale with N exponentially,j as exp(N 2 ). By and large, this could explain the failure to confirm universality of the k-string tensions in numerical simulations. On the theoretical side of explorations of the non-Abelian strings at strong coupling, all we have at the moment are various conjectures. In 1995 Douglas and Shenker suggested a much debated Sine formula for the k-string tension which replaced a Casimir scaling hypothesis that prevailed previously. Douglas–Shenker’s arguments were based on N = 2 super-Yang–Mills model. The Sine formula got further support from MQCD, and, later (in 2003), from a conjectured relation between the k-wall tension in N = 1 Yang–Mills theory and the string tension.38 Moreover, Armoni and I observed that the Sine formula is consistent with the general large-N expansion while the Casimir scaling is not — a rather obvious circumstance previously overlooked. Summarizing the issue of the QCD strings I will just say that although contours of the future construction became, perhaps, visible, a huge challenge remains — transforming these contours into a fully controllable quantitative construction. Let us turn now to a recent development of the ’t Hooft large-N ideas at the interface of supersymmetry and QCD, known as planar equivalence.19 Genesis of planar equivalence can be traced to string theory. In 1998 Kachru and Silverstein studied various orbifolds of R6 within the AdS/CFT correspondence, of which I will speak later. Starting from N = 4, they obtained distinct — but equivalent in the infinite-N limit — four-dimensional daughter gauge field theories with matter, with varying degree of supersymmetry, all with vanishing β functions.k The next step was made by Bershadsky, Johansen and Vafa. These authors eventually abandoned AdS/CFT, and string methods at large. Analyzing gauge field theories per se they proved that an infinite set of amplitudes in the orbifold daughters of the parent N = 4 theory in the large-N limit coincide with those of the parent theory, order by order in the gauge coupling. Thus, explicitly different theories have the same planar limit, at least perturbatively. After a few years of relative oblivion, interest in the issue of planar equivalence was revived by Strassler in 2001. In the inspiring paper entitled On Methods for Extracting Exact Nonperturbative Results in Nonsupersymmetric Gauge Theories, he shifted the emphasis away from the search for supersymmetric daughters, towards engineering QCD-like daughters. Unfortunately, the orbifold daughters considered by Strassler proved to be rather useless. However, the idea gained momentum, and in 2003 planar equivalence, both perturbative and nonperturbative, was demonstrated to be valid for orientifold daughters (Armoni, Shifman, Veneziano). The orientifold daughter of SUSY gluodynamics is a nonsupersymmetric Yang–Mills j The decay occurs through production of a pair of gluelumps, of which I spoke previously. statement is slightly inaccurate; I do not want to dwell on subtleties. k This Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5709 theory with one Dirac fermion in the two-index antisymmetric representation of SU(N ). At N = 3 the orientifold daughter identically reduces to one-flavor QCD! Thus, one-flavor QCD is planar-equivalent to SUSY gluodynamics. This remarkable circumstance allows one to copy results of these theories from one to another. For instance, color confinement of one-flavor QCD to supersymmetric Yang–Mills, and the exact gluino condensate in the opposite direction. This is how the quark condensate was calculated, for the first time analytically, in one-flavor QCD (Armoni, Shifman, Veneziano, 2003). Above I mentioned that in the 1980’s and 1990’s applications of the ’t Hooft 1/N expansion proliferated. Although the 1/N expansion definitely captures basic regularities of the hadronic world, it seems to underestimate the role of quark loops. Take, for instance, the quark dependence of the vacuum energy. In the ’t Hooft limit the vacuum energy density is obviously independent of the quark mass, since all quark loops die out. At the same time, an estimate based on QCD low-energy theorems tells us that changing the strange-quark mass from ∼ 150 MeV to zero would roughly double the value of the vacuum energy density. An alternative orientifold large-N expansion suggested by Armoni, Veneziano and myself fixes this problem. However, by and large, phenomenological implications of the orientifold 1/N expansion have not yet been studied. Here I would like to make a brief digression about surprises. Surprises accidentally occur even in old disciplines. This is what happened with the quark–gluon plasma (QGP) state of matter conjectured in the 1970’s.25 For thirty years QGP was expected to be a simple near-ideal gas. When it was discovered at RHIC, just above the phase transition it turned out39,40 to be strongly coupled! Theorists working in this area compare this event with a (hypothetical) discovery of a sizable previously unknown island, a terra incognita, in the middle of the Atlantic. Recently a hypothesis was formulated according to which the strongly coupled QGP is a plasma of both electric and magnetic charges.40 In the remainder of this talk I will focus exclusively on interrelations between string theory and Yang–Mills field theories. Some of them have been already mentioned above. My task is to complete this outline. String theory which emerged from dual hadronic models in the late 1960’s and 1970’s, elevated to the “theory of everything” in the 1980’s and early 1990’s, when it experienced an unprecedented expansion, seemingly entered, in the beginning of Era III, a “return-to-roots” stage. Results and techniques of string/D-brane theory, being applied to non-Abelian field theories (both, supersymmetric and nonsupersymmetric), have generated numerous predictions of various degree of relevance for gauge theories at strong coupling. If the latter are, in a sense, dual to string/Dbrane theory — as is generally believed to be the case — they must support domain walls (of the D-brane type). In addition, string/D-brane theory teaches us that a fundamental string that starts on a confined quark, can end on such a domain wall. These features are interesting not just by themselves; one can hope that, being established, they will shed light on regularities inherent to QCD (and now Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5710 M. Shifman we know, they do). The task of finding solutions to “down-to-earth” problems of QCD and other gauge theories by using results and techniques of string/D-brane theory is currently recognized by many as the goal of the community. On the other hand, one can hope that the internal logic of development of string theory will be fertilized by insights and hints obtained from field theory. D-Branes in Field Theory In 1996 Dvali and I published a paper entitled Domain Walls in Strongly Coupled Theories. We reanalyzed supersymmetric gluodynamics, found an anomalous (1, 0) central charge in superalgebra, not seen at the classical level, and argued that this central charge will be saturated by domain walls interpolating between vacua with distinct values of the order parameter, the gluino condensate hλλi, labeling N distinct vacua of the theory. We obtained an exact relation expressing the wall tension in terms of the gluino condensate.26,34 Minimal walls interpolate between vacua n and n + 1, while k-walls interpolate between n and n + k. In this paper we also suggested a mechanism for localizing gauge fields on the wall through bulk confinement. Later this mechanism was implemented in many models. In the 1997 paper Branes and the Dynamics of QCD, Witten interpreted the above BPS walls as analogs of D-branes. This is because their tension scales as N ∼ 1/gs rather than 1/gs2 typical of solitonic objects (here gs is the string constant). Many promising consequences ensued. One of them was the Acharya–Vafa derivation of the wall worldvolume theory (2001). Using a wrapped D-brane picture and certain dualities they identified the k-wall worldvolume theory as (1 + 2)dimensional U(k) gauge theory with the field content of N = 2 and the Chern– Simons term at level N breaking N = 2 down to N = 1. Later Armoni and Hollowood exploited this setup to calculate the wall–wall binding energy. Beginning from 2002 Alësha Yung and I developed a benchmark N = 2 model, weakly coupled in the bulk (and, thus, fully controllable), which supports both BPS walls and BPS flux tubes. We demonstrated that a gauge field is indeed localized on the wall; for the minimal wall this is a U(1) field while for nonminimal walls the localized gauge field is non-Abelian. We also found a BPS wall–string junction related to the gauge field localization. The field-theory string does end on the BPS wall, after all! The end-point of the string on the wall, after Polyakov’s dualization, becomes a source of the electric field localized on the wall. In 2005 Norisuke Sakai and David Tong analyzed generic wall–string configurations. Following condensed matter physicists they called them boojums. Summarizing, we are witnessing a very healthy process of cross-fertilization between string and field theories. At first, the relation between string theory and supersymmetric gauge theories was mostly a “one-way street” — from strings to field theory. Now it is becoming exceedingly more evident that field-theoretic methods and results, in their turn, provide insights in string theory. Persistent Challenges of Quantum Chromodynamics 5711 Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Multiparton Amplitudes We already know that Era I was the triumph of perturbative QCD. At the same time, obtaining high orders in the perturbative expansion needed for evaluation of the multiparton scattering amplitudes was an immense technical challenge. To understand the scale of the problem suffice it to have a look at a single color factor in the five-gluon tree amplitude in terms of dot products of momentum and polarization vectors, see Fig. 1 in Ref. 41. Due to the gauge nature of interactions in QCD, the final expressions for the multiparton scattering amplitudes are orders of magnitude simpler than intermediate expressions. In 1986 Parke and Taylor proposed a closed formula for the scattering process of the type “two gluons of negative helicity −→ (n − 2) gluons of positive helicity,” where n is arbitrary. This is called the maximal helicity violating (MHV) amplitude. Using off-shell recursion relations Berends and Giele then provided a proof of the Parke–Taylor proposal. In the 1990’s Bern, Dixon and Kosower pioneered applying string methods to obtain loop amplitudes in supersymmetric theories and pure Yang–Mills. The observed simplicity of these results led to an even more powerful approach based on unitarity. Their work resulted in an advanced helicity formalism exhibiting a feature of the amplitudes, not apparent from the Feynman rules, an astonishing simplicity. In 2003 Witten uncovered a hidden and elegant mathematical structure in terms of algebraic curves in terms of twistor variables in gluon scattering amplitudes: he argued that the unexpected simplicity could be understood in terms of twistor string theory.l This observation created a diverse and thriving community of theorists advancing towards full calculation of multiparton amplitudes at tree-level and beyond, as it became clear that loop diagrams in gauge theories have their own hidden symmetry structure. Most of these results do not directly rely on twistors and twistor string theory, except for some crucial inspiration. So far, there is no good name for this subject. Marcus Spradlin noted that an unusually large fraction of contributors’ names start with the letter B.m Therefore, perhaps, we should call it B theory, with B standing for beautiful, much in the same way as M in M theory stands for magic. I could mention a third reason for “B theory”: Witten linked the scattering amplitudes to a topological string known as the “B model.” B theory revived, at a new level, many methods of the pre-QCD era, when S-matrix ideas ruled the world. For instance, in a powerful paper due to Britto, Cachazo, Feng and Witten (2005), tree-level on-shell amplitudes were shown in a very simple and general way to obey recursion relations. lA precursor of this, for the special case of MHV amplitudes, was given by Nair 15 years earlier. example, Badger, Bedford, Berger, Bern, Bidder, Bjerrum-Bohr, Brandhuber, Britto, Buchbinder, . . . (of course, one should not forget about Cachazo, Dixon, Feng, Forde, Khoze, Kosower, Roiban, Spradlin, Svrček, Travaglini, Vaman, Volovich, . . .). This reminds me of a joke of a proof given by a physicist that almost all numbers are prime: one is prime, two is prime, three is prime, five is prime, while four is an exception just supporting the general rule. m For Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5712 M. Shifman Returning to topological string theory in twistor space let me note that it is dual to a weakly coupled N = 4 gauge theory. Evaluation of the string-theory instanton contributions gave MHV scattering amplitudes for an arbitrary number of partons (Cachazo, Svrček, Witten). Other amplitudes were presented as integrals over the moduli space of holomorphic curves in the twistor space (Roiban, Spradlin, Volovich). In essence, the formalism that came into being in this way reduces calculations of gauge amplitudes to an effective scalar perturbation theory. Currently the boundaries of the explored territory are expanding into loop amplitudes, and there is even a proposal for an all-loop-order resummation of MHV planar amplitudes. A suspicion emerged that planar N = 4 gauge theory may prove to be integrable! For reviews see Refs. 42 and 43. Spin Chains and Integrability QCD practitioners “observed experimentally” rather long ago that a hidden integrability unexpectedly shows up in problems associated with certain limits of QCD, e.g. high energy Regge behavior of scattering amplitudes (Lipatov; Faddeev and Korchemsky, 1994) and the spectrum of anomalous dimensions of operators appearing in deep inelastic scattering (Braun, Derkachov and Manashov, 1998). It turns out that in both cases evolution equations (in the logarithm of the appropriate energy scale) can be be identified with time evolution governed by Hamiltonians of various integrable quantum spin chains, generalizations of the Heisenberg spin magnet.44 Historically, integrability was first discovered in the Regge limit of QCD. Its relation to evolution equations for maximal-helicity operators remains unclear till today. In pure Yang–Mills theories integrability is firmly established to be a property of one- and two-loop planar evolution equations.n At finite N nonplanar corrections break it. One can hardly doubt that integrability is a consequence of a general hidden symmetry of all Yang–Mills theories in the limit N → ∞, not seen at the classical level; it appears dynamically at the quantum level. What is its origin? That is where, as many believe, insights from string theory could help. A few years later, in 2002–2003, Minahan and Zarembo; and Beisert, Kristjansen and Staudacher rediscovered the same phenomenon from a different side. These theorists, motivated by gauge–string duality (which will be discussed shortly), studied45 renormalization of composite operators in the maximally supersymmetric field theory, N = 4. Note that the spectrum of the anomalous dimensions of appropriately chosen operators is ideally suited for mapping. The dilatation operator on the field-theory side is identified with a Hamiltonian on the string-theory side. Then anomalous dimensions of the operators under considerations are mapped onto the energy of the corresponding string configurations (Berenstein, Maldacena, Nastase; and Gubser, Klebanov and Polyakov, 2002, see Refs. 46 and 47). There are independent reasons n So far nobody knows what exactly happens beyond two loops. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5713 to expect integrability on the string side too (Bena, Polchinski and Roiban, 2003). If so, it should be valid both for small and large values of the ’t Hooft coupling. This is an excellent example of how QCD and string theory work hand in hand, and, being combined, give rise to applications going far beyond these two theories. Indeed, before this development the studies of the Heisenberg magnets in solid state physics were limited to spins in the finite-dimensional (compact) representations of SU(2). (The original Heisenberg model solved through the Bethe ansatz was built of the Pauli matrices). In the context of the operator spectrum problem one encounters a novel type of Heisenberg magnets, with spin operators that are generators of the (super)conformal group in the underlying gauge theory. The corresponding representations are necessarily infinite-dimensional and, as a consequence, the corresponding Heisenberg magnets turn out to be noncompact. Noncompact spin chains that “descended” from gauge theories, have a number of stunning features, interesting on their own. An ongoing fusion of both communities gives hope that integrability of certain problems of Yang–Mills will be explained by a hidden symmetry of a (non)critical string. I would like to emphasize that QCD in its entirety is not integrable, beyond any doubt. Are there broader implications of the hidden integrability, going beyond the two problems mentioned above? Is there hope that spin chain dynamics will make scattering of, say, five or ten gluons exactly calculable? AdS/CFT or String-Gauge Holographic Duality Now I turn the page and open a new chapter, which, although not yet fully written, caused a lot of excitement. It may or may not become yet another breakthrough in QCD. We will see . . . . It all started in 1998 when Maldacena; Gubser, Klebanov and Polyakov; and Witten argued (conjectured) that certain four-dimensional super-Yang–Mills theories at large N could be viewed as holographic images of higher-dimensional string theory. In the limit of a large ’t Hooft coupling the latter was shown to reduce to anti-de-Sitter supergravity. The framework got the name “Anti-deSitter/Conformal Field Theory (AdS/CFT) correspondence.” Duality is not something totally new in non-Abelian gauge theories. In fact, the first observation of the Montonen–Olive duality dates back to 1977, i.e. Era I. Montonen and Olive suggested that in four-dimensional N = 4 super-Yang–Mills theory, replacing everything “electric” by everything “magnetic,” one obtains an equivalent theory provided that simultaneously, g is replaced by 1/g. This is an example of the electric–magnetic duality; its N = 2 cousin played an important role in the Seiberg–Witten demonstration of the dual Meissner effect.o AdS/CFT is a totally different kind of duality — it is holographic. o The Montonen–Olive duality is the oldest known example of S-duality, or a strong-weak duality. I must admit that in 1977 I did not appreciate the importance of this result, since I could not imagine, even in my wildest dreams, that extended supersymmetry could become relevant to QCD. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5714 M. Shifman By now, it is generally believed that ten-dimensional string theory in suitable space–time backgrounds can have a dual, holographic description in terms of superconformal gauge field theories in four dimensions. Conceptually, the idea of a stringgauge duality ascends to ’t Hooft, who realized that the perturbative expansion of SU(N ) gauge field theory in the large N limit (with the ’t Hooft coupling fixed) can be reinterpreted as a genus expansion of discretized two-dimensional surfaces built from the field theory Feynman diagrams. This expansion resembles the string theory perturbative expansion in the string coupling constant. The AdS/CFT correspondence is a quantitative realization of this idea for four-dimensional gauge theories. In its purest form it identifies the “fundamental type IIB superstring in a ten-dimensional anti-de-Sitter space–time background AdS5 × S 5 with the maximally supersymmetric N = 4 Yang–Mills theory with gauge group SU(N ) in four dimensions.” The latter theory is superconformal. At this point I planned, originally, to make a few explanatory remarks. Fortunately, I realized in time that this will be just another Zhukovsky anecdote. Let me tell you this joke (some say that was a true story). Nikolai Zhukovsky was a famous Russian scientist in the fields of gas and fluid mechanics and aeronautics, the theoretical father of Russian aviation. Aeronautics was a very popular subject at the end of the 19th century; Zhukovsky began a trend of lecturing for the general public. At one of the lectures he delivered at the Polytechnical Museum in Moscow the audience was mainly composed of middle-aged wives of Russian nobility. He wanted to explain the Bernoulli law, using as an example a sphere in a gas flow. When he said “sphere” he understood immediately that he lost the audience. So, he patiently explained, “a sphere is a round object like a ball your children play with.” He saw smiles and relief on the faces in the audience, and quickly continued: “thus, you take this sphere and integrate pressure over its surface . . . .” I am not going to repeat such a mistake, and will skip explanations, referring the reader to the review papers.48–51 The main task is to leave conformality and get as close to real QCD as possible. Currently there are two (hopefully convergent) lines of though. Chronologically the first was the top-down approach pioneered by Witten; Polchinski and Strassler; Klebanov and Strassler; Maldacena and Nuñez, and others. Here people try to obtain honest-to-God solutions of the ten-dimensional equations of motion, often in the limit of a large ’t Hooft coupling when on the string side of the theory one deals with supergravity limit. The problem is: in many instances these solutions are dual to . . . sort of QCD, rather than QCD as we know it. For instance, Witten’s setup or the Maldacena–Nuñez solution guarantee color confinement but the asymptotically free regime of QCD is not attained. The Klebanov–Strassler supergravity solution is near AdS5 in the ultraviolet limit, a crucial property for the existence of a dual four-dimensional gauge theory. In the ultraviolet this theory exhibits logarithmic running of the couplings which goes under the name of duality cascade. They start from string theory on a warped deformed conifold and discover a cascade of SU(kM ) × SU((k − 1)M ) supersymmetric Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5715 gauge theories on the other side. As the theory flows in the infrared, k repeatedly changes by unity, see the review paper Ref. 52. In the infrared this theory exhibits a dynamical generation of the scale parameter Λ, which manifests itself in the deformation of the conifold on the string side. There is a variant of the top-down approach in which the requirement of the exact solution of the supergravity equations is “minimally” relaxed. Confinement is enforced through a crude cutoff of the AdS bulk in the infrared, at z0 , where z is the fifth dimension. This leads to a “wrong” confinement. In particular, the Regge trajectories do not come out linear.p Keeping the full-blown string theory but still adhering to the above hard-wall approximation one restores asymptotic linearity of the Regge trajectories at large angular momenta J or excitation numbers n. In this limit one can then calculate, say, the meson decay rates, as was done recently by Sonnenschein and collaborators, who recovered the 1979 Casher– Neuberger–Nussinov (CNN) quasiclassical formula! Then, I would like to mention the 2006 paper of Brower, Polchinski, Strassler and Tan entitled The Pomeron and Gauge-String Duality.q Are we witnessing a come-back of the large-scale activities in Pomeron physics whose golden years seemingly ended with the advent of QCD? Incorporating fundamental quarks, with the spontaneously broken chiral symmetry (χSB), is a separate problem which received much attention. First, fundamental quarks were introduced, via probe branes, by Karch and Katz, in duals of the Coulomb phase. Quarks in confining scenarios were introduced, in the context of the Klebanov–Strassler confining background, through D7 branes, by Sakai and Sonnenschein. A model that admits a full-blown χSB was developed by Sakai and Sugimoto who embedded D8 (anti)branes in Witten’s setup. By and large, I cannot say that at present AdS/CFT gives a better (or more insightful) description of the hadronic world, than, say the “old” SVZ condensatebased method. Given a rather crude character of the hard-wall and similar approximations, perhaps, today one may hope to extract only universal information on hadronic dynamics, steering clear of all details. From AdS/CFT to AdS/QCD? This assessment, shared by many, gave rise to an alternative movement, which goes under the name of AdS/QCD. This bottom-up approach was pioneered by Son and Stephanov who were motivated, initially, by an observation made by Bando, Kugo and Yamawaki. The starting point of AdS/QCD is a “marriage” between the holographic representation and OPE-based methods, plus χSB, plus all other pA year ago, preparing for a talk, I suddenly realized that the meson spectrum obtained in this way identically coincides with the 30-year-old result of Alexander Migdal, who, sure enough, had no thoughts of supergravity in five dimensions. His idea was to approximate logarithms of perturbation theory by an infinite sum of poles in the “best possible way.” Then this strategy was abandoned since it contradicts OPE. Now it has been resurrected in a new incarnation. The reason for the coincidence of the 1977 and 2005 results is fully clear, see Erlich et al., 2006. q Lenny Susskind referred to it in jest as “Strassler-this-Strassler-that.” Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. 5716 M. Shifman ideas that were developed during Era I. The strategy is as follows: instead of solving the ten-dimensional theory, the theorist is supposed to make various conjectures in order to guess an appropriate five-dimensional metric encoding as much information on real QCD as possible, and then, with this metric in hands, get new insights and make new predictions. In this direction I personally was most impressed by works of Joshua Erlich, Andreas Karch, Emanuel Katz, Dam Son, and Misha Stephanov in which many features of the SVZ expansions were recovered (as well as the linearity of the Regge trajectories) from a very simple ansatz for the scalar factor in the fivedimensional metric. It remains to be seen how far this road will lead us. Concluding, I would like to suggest for consideration of the proponents a couple of trial questions: • Find the mass splitting of high radial excitations in the chiral pairs (e.g. ρn and A1n ); • Find the next-to-leading term in the 1/n expansion of the width-to-mass ratio, say, Γ(ρn )/M (ρn ). The leading term is given by the CNN formula. Instead of Conclusions By the year 1980 : • OPE-based methods were on the rise; • Some crucial low-energy theorems shedding light on the QCD vacuum structure established; • Dual Meissner effect for color confinement conjectured; • 1/N expansion as a useful classification tool suggested; • SUSY gauge theories constructed and studied (almost exclusively, in the perturbative sector); • Instantons/monopoles discovered; • Hypothesis of the monopole–particle duality in N = 4 put forward. This is all. Hints were there, but who could have guessed? Now : • OPE-based methods culminated in the 1990’s; • 1/N expansion became semiquantitative in some problems; • Triumph of SUSY-based methods for QCD cousins is unquestionable (A significant tool kit developed; the dual Meissner effect in N = 2∗ proven! Dualities in N = 1 discovered!); • Non-Abelian strings discovered and understood; a large number of parallels between string theory/D-branes and non-Abelian gauge theories revealed from the field theory side; • AdS/QCD, although still in its infancy, starts bringing fruits; • String and QCD practitioners are finally talking to each other, to their mutual benefit. Persistent Challenges of Quantum Chromodynamics 5717 Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Predictions: (Indirectly depend on external factors, such as SUSY discovery at LHC, . . . .) • SUSY-based methods will proliferate, allowing one to treat closer relatives of QCD, as well as important aspects of QCD per se; • These methods will spread to other strongly-coupled theories, e.g. those relevant to condensed matter physics; • The gap between string theories and realistic strong-coupling gauge theories will continue to narrow, with the two-way exchange of ideas; • 1/N expansion and holographic descriptions of QCD will grow into quantitative tools, whose accuracy will be under complete theoretical control. Acknowledgments I am deeply grateful to my coauthors, for joy of working together and discussing physics an infinite number of times, from the very first days of QCD. I would like to thank A. Armoni, Z. Bern, V. Braun, G. Korchemsky, A. Ritz, E. Shuryak, M. Stephanov, and A. Tseytlin for valuable comments on the manuscript. The sketch below entitled “Uphill Road” belongs to Yuri Korjevsky. 5718 M. Shifman Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. References 1. P. M. Zerwas and H. A. Kastrup (eds.), QCD — Twenty Years Later, Vols. 1 and 2 (World Scientific, Singapore, 1993). 2. M. Shifman (ed.), At the Frontiers of Particle Physics: Handbook of QCD, Vols. 1–4 (World Scientific, Singapore, 2001–2002). 3. M. Shifman, A. Vainshtein and J. Wheater (eds.), From Fields to Strings: Circumnavigating Theoretical Physics, Vols. 1–3 (World Scientific, Singapore, 2005). 4. M. Peloso and M. Shifman (eds.), Continuous Advances in QCD 2006 (World Scientific, Singapore, 2006). 5. R. D. Field, Applications of Perturbative QCD (Addison-Wesley, 1989). 6. Yu. Dokshitzer, V. A. Khoze, A. H. Mueller and S. I. Troyan, Basics of Perturbative QCD (Editions Frontieres, 1991). 7. G. Altarelli, The Development of Perturbative QCD (World Scientific, Singapore, 1994). 8. K. Ellis, J. Stirling and B. Webber, QCD and Collider Physics (Cambridge University Press, 1996). 9. M. Shifman, Quark-hadron duality, in At the Frontiers of Particle Physics: Handbook of QCD, Vol. 3, ed. M. Shifman (World Scientific, Singapore, 2001), p. 1447, hepph/0009131. 10. K. G. Wilson and J. B. Kogut, Phys. Rep. 12, 75 (1974). 11. J. Gasser and H. Leutwyler, Phys. Rep. 87, 77 (1982). 12. A. I. Vainshtein, Int. J. Mod. Phys. A 14, 4705 (1999), hep-ph/9906263. 13. A. I. Vainshtein and V. I. Zakharov, Sov. Phys. Usp. 13, 73 (1970). 14. S. Treiman, R. Jackiw, B. Zumino and E. Witten, Current Algebra and Anomalies (Princeton University Press, 1985). 15. S. Coleman, Soft pions, in Aspects of Symmetry (Cambridge University Press, 1985), p. 36. 16. A. P. Balachandran, Skyrmions, in High Energy Physics 1985, Vol. 1, eds. M. J. Bowick and F. Gürsey (World Scientific, Singapore, 1985), p. 1. 17. K.-F. Liu, Chiral Solitons (World Scientific, Singapore, 1987). 18. A. V. Manohar and M. B. Wise, Heavy Quark Physics (Cambridge University Press, 2000). 19. A. Armoni, M. Shifman and G. Veneziano, From super-Yang–Mills theory to QCD: Planar equivalence and its implications, in From Fields to Strings: Circumnavigating Theoretical Physics, Vol. 1, eds. M. Shifman, A. Vainshtein and J. Wheater (World Scientific, Singapore, 2005), p. 353, hep-th/0403071. 20. G. ’t Hooft, Under the Spell of the Gauge Principle (World Scientific, Singapore, 1994). 21. E. Witten, The 1/N expansion in atomic and particle physics, in Recent Developments in Gauge Theories, eds. G. ’t Hooft, C. Itzykson, A. Jaffe, H. Lehmann, P. K. Mitter, I. M. Singer and R. Stora (Plenum Press, 1980), p. 403. 22. S. Coleman, 1/N , in Aspects of Symmetry (Cambridge University Press, 1985), p. 351. 23. M. B. Green, J. H. Schwarz and E. Witten, Superstring Theory (Cambridge University Press, 1987). 24. M. Shifman (ed.), Vacuum Structure and QCD Sum Rules (North-Holland, Amsterdam, 1992). 25. E. Shuryak, The QCD Vacuum, Hadrons and Superdense Matter, 2nd edn. (World Scientific, Singapore, 2004). 26. M. Shifman, Prog. Part. Nucl. Phys. 39, 1 (1997), hep-th/9704114. Int. J. Mod. Phys. A 2006.21:5695-5719. Downloaded from www.worldscientific.com by UNIVERSITY OF MINNESOTA TWIN CITIES on 04/27/22. Re-use and distribution is strictly not permitted, except for Open Access articles. Persistent Challenges of Quantum Chromodynamics 5719 27. M. Srednicki, Axions: past, present, and future, in Continuous Advances in QCD 2002/ARKADYFEST, eds. K. Olive, M. Shifman and M. Voloshin (World Scientific, Singapore, 2002), p. 509, hep-th/0210172. 28. G. Gabadadze and M. Shifman, Int. J. Mod. Phys. A 17, 3689 (2002), hepph/0206123. 29. S. Weinberg, Problems in gauge field theories, Talk at the XVII Int. Conf. on High Energy Physics, London, July 1974. 30. S. Mandelstam, Phys. Rep. 23, 245 (1976). 31. G. ’t Hooft, The confinement phenomenon in quantum field theory, in Under the Spell of the Gauge Principle (World Scientific, Singapore, 1994), p. 514. 32. I. I. Y. Bigi, M. A. Shifman and N. Uraltsev, Ann. Rev. Nucl. Part. Sci. 47, 591 (1997), hep-ph/9703290. 33. A. Manohar, Hadrons in the 1/N expansion, Ref. 2, p. 507. 34. M. Shifman and A. Vainshtein, Instantons versus supersymmetry, fifteen years later, in ITEP Lectures on Particle Physics and Field Theory, Vol. 2 (World Scientific, Singapore, 1999), p. 485, hep-th/9902018. 35. D. Tong, TASI lectures on solitons, hep-th/0509216. 36. M. Eto, Y. Isozumi, M. Nitta, K. Ohashi and N. Sakai, J. Phys. A 39, R315 (2006), hep-th/0602170. 37. J. Greensite, Prog. Part. Nucl. Phys. 51, 1 (2003), hep-lat/0301023. 38. M. Shifman, Acta Phys. Pol. B 36, 3805 (2005), hep-ph/0510098. 39. E. Shuryak, Nucl. Phys. A 750, 64 (2005), hep-ph/0405066. 40. E. Shuryak, Ref. 4, pp. 3–16. 41. Z. Bern, Multiparton amplitudes in QCD, Ref. 2, p. 995. 42. V. V. Khoze, Gauge theory amplitudes, scalar graphs and twistor space, Ref. 3, p. 622, hep-th/0408233. 43. F. Cachazo and P. Svrček, Lectures on twistor strings and perturbative Yang–Mills theory, hep-th/0504194. 44. A. V. Belitsky, V. M. Braun, A. S. Gorsky and G. P. Korchemsky, Integrability in QCD and beyond, Ref. 3, p. 266, hep-th/0407232. 45. N. Beisert, Phys. Rep. 405, 1 (2005), hep-th/0407277. 46. A. Tseytlin, Spinning strings and AdS/CFT duality, Ref. 3, Vol. 2, p. 1648, hepth/0311139. 47. J. Plefka, Spinning strings and integrable spin chains in the AdS/CFT correspondence, hep-th/0507136. 48. J. M. Maldacena, Lectures on AdS/CFT, Ref. 3, p. 1473. 49. O. Aharony, S. S. Gubser, J. M. Maldacena, H. Ooguri and Y. Oz, Phys. Rep. 323, 183 (2000), hep-th/9905111. 50. P. Di Vecchia, Large N gauge theories and ADS/CFT correspondence, in Superstrings and Related Matters, eds. C. Bachas, J. Maldacena, K. S. Narain and S. RandjbarDaemi (World Scientific, Singapore, 2001), p. 1, hep-th/9908148. 51. E. D’Hoker and D. Z. Freedman, Supersymmetric gauge theories and the AdS/CFT correspondence, in Strings, Branes and Extra Dimensions, eds. S. Gubser and J. D. Lykken (World Scientific, Singapore, 2004), p. 3, hep-th/0201253. 52. M. J. Strassler, The duality cascade, in Progress in String Theory: TASI 2003 Lecture Notes, ed. J. Maldacena (World Scientific, Singapore, 2005), pp. 419–510, hepth/0505153.