Interplay Between d-wave Superconductivity and a Bond Density Wave in the one-band Hubbard model

# Interplay Between d-wave Superconductivity and a Bond Density Wave in the one-band Hubbard model

## Abstract

It is now well established that superconducting cuprates support a charge density wave state in the so-called underdoped region of their phase diagram. We investigate the possibility of charge order in the square-lattice Hubbard model, both alone and in coexistence with -wave superconductivity. The charge order has a period four in one direction, is centered on bonds and has a form factor. We use the variational cluster approximation, an approach based on a rigorous variational principle that treats short-range correlations exactly, with two clusters of size that together tile the infinite lattice and provide a non-biased unit for a period-four bond density wave (BDW). We find that the BDW exists in a finite range of hole doping and increases in strength from to . Its location and intensity depends strongly on the band dispersion. When probed simultaneously with -wave superconductivity, the energy is sometimes lowered by the presence of both phases, depending on the interaction strength. Whenever they coexist, a pair-density wave (a modulation of superconducting pairing with the same period and form factor as the BDW) also exists.

## I Introduction

Charge order in underdoped superconducting cuprates has been observed by many techniques and in many compounds. Nuclear magnetic resonance measurements on YBaCuO indicate the presence of a long-range, static charge order without any signature of spin order Wu et al. (2011, 2013). In BiSrCaCuO, scanning tunnelling microscopy (STM) shows a periodic modulation in the density of states Hoffman et al. (2002); Vershinin et al. (2004). Charge density wave correlations have also been observed in X-ray scattering Ghiringhelli et al. (2012); Achkar et al. (2012); Chang et al. (2012), and the charge density wave seems to be directed along copper oxygen bondsBlackburn et al. (2013). STM measurements also indicate that the charge density wave modulation resides on Cu–O–Cu bonds Vershinin et al. (2004); Kohsaka et al. (2007); Achkar et al. (2013). The dependence of the peak intensity as a function of magnetic field clearly indicates the possibility of a competition between d-wave superconductivity and charge density wave order Chang et al. (2012). More recently, the pair-density wave (PDW) that coexists with -wave superconductivity and charge order has also been observed Hamidian et al. (2016).

Theoretical investigations of charge order in cuprates roughly fall into two categories: (i) those that study the effect of static charge order on observables and (ii) those that attempt at explaining the origin of charge order from a model Hamiltonian with interactions. This work belongs to the second category. A few attempts have been made in that direction in the literature. For instance, VojtaVojta (2002) has applied mean-field theory to the - model plus extended interactions and mapped out various charge order phases that appear when is low enough, whereas -wave superconductivity dominates at higher . A pure exchange model (without correlated hopping) has also been studied at the mean field level by Sachdev and LaPlaca Sachdev and La Placa (2013). Atkinson et al. have applied the generalized RPA approximation to the full three-band Hubbard model Atkinson et al. (2015) and view charge order, like the pseudogap, as a side effect of short-range antiferromagnetic correlations. The Gutzwillwer approximation has been applied to the Hubbard model (without extended interactions) but no charge order was found with that approachMarkiewicz et al. (2011). Charge order at half-filling in the extended Hubbard model has recently been investigated with the dynamical cluster approximationTerletska et al. (2016), where it should be competition with antiferromagnetism. That competition has also been studied in the context of the Hubbard-Holstein modelBauer and Hewson (2010), in which optical phonons would favor charge order over antiferromagnetism.

There is also a vast literature on stripe order, i.e., a coexistence of charge and spin density waves, which we will not review here. Let us mention nonetheless the work of Corboz et al.Corboz et al. (2011) in which the nearest-neighbor - model is studied using the projected-entangled pair states (PEPS) variational Ansatz, and where stripe order occurs naturally in coexistence with -wave superconductivity. This is consistent with the previous work of Capello et al.Capello et al. (2008) on the same model using variational Monte Carlo. Finally, the pair density wave (PDW) state has been the focus of many studies Himeda et al. (2002); Raczkowski et al. (2007); Choubey et al. (2016); Freire et al. (2015); Wang et al. (2015) (for a recent review, see Ref.Fradkin et al. (2015)).

In this work we investigate whether a particular CDW order can arise from local repulsive interactions alone, and whether it can coexist with -wave superconductivity. To this end, we apply the variational cluster approximation (VCA) Dahnken et al. (2004) to the one-band repulsive Hubbard model. The charge density wave studied is bond centered, has a -wave form factor and will henceforth be referred to as a bond-density wave (BDW). The VCA, and other quantum cluster methods such as Cluster Dynamical Mean Field TheoryLichtenstein and Katsnelson (2000); Kotliar et al. (2001) and the Dynamical Cluster ApproximationMaier et al. (2000) already predict the presence of -wave superconductivity in the doped one-band Hubbard modelSénéchal et al. (2005); Maier et al. (2005); Kancharla et al. (2008). We find that a bond density wave is indeed possible in the doped Hubbard model and that this phase is more robust when increasing the interaction strength . Its location is also sensitive to the detailed band structure. In addition, we find that the BDW can coexist with -wave superconductivity, although both the dSC and BDW order parameters are negatively affected by their coexistence. A pair-density wave (PDW) order also sets in when BDW and dSC orders coexist.

This paper is organized as follows: in Sect. II, we describe the particular BDW studied and briefly review the VCA method; in Sect. III, we present and discuss our numerical results. Finally, we conclude in Sec. IV.

## Ii Model and method

Let us first establish some notation. The one-band Hubbard model on a square lattice is defined by the following Hamiltonian:

 H=−t∑⟨r,r′⟩,σc†r,σcr′,σ+U∑rnr,↑nr,↓−μ∑r,σnr,σ (1)

where destroys an electron of spin at site on the lattice. is the number of electrons of spin at site . The chemical potential is included in the Hamiltonian for convenience.

Following Allais et al. (2014), a general BDW operator of wavevector is defined as follows:

 ^ΨBDW=∑rσ,atq,ac†r,σcr+a,σei(q⋅r+a/2)+H.c (2)

We will probe a BDW of period four, with -form factor: , with . The -form factor is supported by STM observations Fujita et al. (2014). This BDW is illustrated on Fig. 1. This choice is motivated by its simple commensurability and its compatibility with the cluster method we will use, as explained below.

We will also probe -wave superconductivity, with a pair operator defined as

 Extra open brace or missing close brace (3)

where . If both superconductivity and charge order are present, the pair-density wave (singlet) order parameter

 ^ΨPDW=∑r,atq,a(cr,↑cr+a,↓−cr,↓cr+a,↑)ei(q⋅r+a/2)+H.c (4)

should also be nonzero.

### ii.1 The Variational Cluster Approximation

In order to probe the possibility of superconductivity and bond density wave as well as their coexistence in Model (1), we use the variational cluster approximation (VCA) with an exact diagonalization solver at zero temperature Dahnken et al. (2004). This method, which goes beyond mean-field theory by keeping the correlated character of the model, has been applied to many strongly correlated systems in connection with various broken symmetry phases, in particular -wave superconductivity Sénéchal et al. (2005); Aichhorn et al. (2007). For a detailed review of the method, see Refs Potthoff (2012, 2014).

In essence, the VCA is a variational method on the electron self-energy. It can probe various broken symmetries (or just the normal state) by exploring a space of self-energies that are the actual self-energies of Model (1), but restricted on a small cluster of sites and augmented by Weiss fields that probe broken symmetries and other one-body terms. Once the optimal self-energy in that space is found, it is added to the noninteracting Green function for the full lattice and, from there, various observables may be computed.

Like other quantum cluster methods, VCA starts by a tiling of the lattice into an infinite number of (usually identical) clusters. In VCA, one considers two systems: the original system described by the Hamiltonian , defined on the infinite lattice, and the reference system, governed by the Hamiltonian , defined on the cluster only, with the same interaction part as . Typically, will be a restriction of to the cluster (i.e., with inter-cluster hopping removed), to which various Weiss fields may be added in order to probe broken symmetries. More generally, any one-body term can be added to . The size of the cluster should be small enough for the electron Green function to be computed numerically.

The optimal one-body part of is determined by a variational principle. More precisely, the electron self-energy associated with is used as a variational self-energy, in order to construct the following Potthoff self-energy functional Potthoff (2003):

 Ω[Σ(ξ)]=Ω′[Σ(ξ)]+Trln[−(G−10−Σ(ξ))−1]−Trln(−G′(ξ)) (5)

The quantities and above are respectively the physical Green function of the cluster and the non-interacting Green function of the infinite lattice. The symbol stands for a small collection of parameters that define the one-body part of . is a functional trace, i.e., a sum over frequencies, momenta and bands, and is the grand potential of the cluster, i.e., its ground state energy, since the chemical potential is included in the Hamiltonian. and are computed numerically via the Lanczos method at zero temperature.

The Potthoff functional in Eq. (5) is computed exactly, but on a restricted space of the self-energies that are the physical self-energies of the reference Hamiltonian . We use a standard optimization method (e.g. Newton-Raphson) in the space of parameters to find the stationary value of :

 ∂Ω(ξ)∂ξ=0 (6)

This represents the best possible value of the self-energy , which is used, together with the non-interacting Green function , to construct an approximate Green function for the original lattice Hamiltonian :

 G(k,ω)=1G−10(k,ω)−Σ(ω)  . (7)

In the above the wavevector is restricted to the reduced Brillouin zone associated with the superlattice defined by the cluster, and all boldface quantities are matrices of dimension (or , if superconductivity is present), being the number of sites in the cluster. From that Green function one can compute the average of any one-body operator, in particular order parameters associated with bond density wave (BDW) or -wave superconductivity (dSC).

The competition between orders can be studied by probing the two orders separately, and then together, in a coexistence scenario. If homogeneous coexistence is possible, then generally the associated value of , which approximates the free energy in this approach, is lower than for the pure solutions for the two phases separately.

Charge density waves, or other states that break translation symmetry, present a particular challenge to cluster methods like VCA, first because the unit cell of the density wave may be larger than the largest cluster that can be practically solved numerically, and second because cluster methods breaks translation symmetry from the outset. The first difficulty is solved by aggregating different clusters that together form a ‘supercluster’ (or repeated unit) that tiles the lattice and that accommodates the charge density wave. The overcome the second difficulty, one has to choose the clusters in such a way that the method is not biased towards the broken symmetry state. In other words, if and are the amplitudes of the Weiss fields associated with the BDW and dSC, respectively, then the Potthoff functional should be an even function of and . Thus the normal solution () is always an option and the broken symmetry state occurs if a nontrivial solution exists with a lower value of . In this work, we will use the two 12-site clusters shown in Fig. 2; the amplitudes of the bond charges in the particular BDW studied are indicated by colored circles. Together, these two clusters form a supercluster that contains three unit cells of the BDW. Changing the sign of the BDW amounts to flipping each cluster horizontally, which does not affect the value of .

A few words on the computation of expectation values. The average of any one-body operator can be computed from the VCA solution in two different ways: (i) by taking its trace against the Green function or (ii) by differentiating the grand potential with respect to an external field. Let be the one-body matrix defining the operator , such that

 ^S=∑α,βsαβc†αcβ (8)

( and are compound indices, representing site together with spin or other band indices). In a cluster approach, a partial Fourier transform can be applied to the site indices in to produce a reduced expression , where belongs to the reduced Brillouin zone and the site indices are now limited to those of the repeated unit (the ‘supercluster’). In this language, the expectation value of may be evaluated as

 ⟨^S⟩=∫dω2π∫rBZd2k(2π)2tr[s(k)G(ω,k)] (9)

where the frequency integral is taken over a contour that circles the negative real axis, targeting the occupied states only. Alternatively, one may add an external field to the lattice Hamiltonian, compute the best estimate of the grand potential for and (i.e. the optimized self-energy functional), and compute the derivative

 ⟨^S⟩=∂Ω∂s≈Ω(ϵ)−Ω(0)ϵ (10)

The two approaches (9) and (10) do not necessarily yield the same answer (the first one is less computationally intensive). In the case of a local operator, like the particle number , which does not contain inter-cluster components, it can be shown that the two approaches will yield the same value if the corresponding Weiss field on the cluster (in this case , the cluster’s chemical potential) is treated as a variational parameter on the same level as the others (e.g. and ).

## Iii Results and discussion

In this section we report the results of VCA calculations on Model (1) at various values of the interaction and for a few band parameters. Since the simulations are somewhat time consuming, we could not explore the space of parameters in extenso, but our results illustrate how a bond density wave can arise at finite doping in the presence of repulsive interactions, and in coexistence with -wave superconductivity also arising from the same interaction.

### iii.1 Pure bond density wave

We start by probing a pure BDW, without superconductivity. In VCA, this amounts to solving the following Hamiltonian on the cluster system of Fig. 2:

 H′=H′0+λ^Ψ′BDW , (11)

inserting the computed Green function into the expression (5) for the Potthoff functional, and finding the value of that minimizes the functional. In the above expression, and are the restriction to the cluster of the kinetic energy operator and of the BDW operator (2), respectively. The coefficient was the only variational parameter used in optimizing the functional . At this point, we have neglected the possible interaction with other phases, such as antiferromagnetism or superconductivity.

Figure 3A shows the optimal value of the Weiss field as a function of chemical potential , for a few values of the local repulsion , and for second-neighbor hopping and third-neighbor hopping and . Note that the nearest-neighbor hopping is set to unity and thus defines the energy scale. These data constitute the ‘raw’ solution from VCA. It is more physically instructive to look at the order parameter as a function of electron density (or doping ), as is done on panels (B) and (C) of the same figure. We show the raw data on panel (A) in order to shed some light on the method itself.

Note that the transition from the BDW phase to the normal phase can proceed in many ways: (i) the Weiss field can go to zero at the phase boundary (continuous transition) or (ii) the solution can jump from a finite value of to zero, either because the solution becomes metastable at that point, or because it ceases to exist (the derivative at the minimum is no longer defined, i.e., the functional acquires a cusp-like behavior). The latter type occurs in the figure for . Fig. 4 illustrates this last point by showing plots of for a few values of . We see how the smooth minimum (indicated by arrows) disappears at some value of as acquires non-differentiable features. This behavior is an occasional drawback of the method and is likely dependent on the shape and size of the cluster, on the type of Weiss field considered, etc. In the continuous case the order parameter goes to zero smoothly, whereas it jumps discontinuously in the second case. A more conventional first-order transition, in which two distinct bona fide solutions have the same value of at some value of an external parameter, is also possible, but has not been observed here.

Panel B of Fig. 3 shows the BDW order parameter as a function of hole doping, both computed from the Green function (Eq. (9)), for the same data sets as those appearing on Panel A. Panel C shows the same data, this time with the density computed from the derivative of with respect to . The values of doping obtained by these two approaches differ by roughly a factor of two. Whatever the method of computing , it appears that the BDW is increasing in strength with and is also quite sensitive on the band dispersion; in particular, no BDW order occurs at .

An apparent way out of the ambiguity in the way to compute electron density is to treat the cluster chemical potential as a variational parameter, on the same level as the Weiss field . When this is done, the data of Fig. 5 are obtained (this was done for and only). The top panel shows the Weiss field as a function of , and the bottom panel the BDW order parameter as a function of doping. Note that the Weiss field tends to larger values on the overdoped side (smaller ). This leads us to think that the overdoped results are less reliable: the ‘canonical’ phase transition in VCA has the Weiss field go to zero at the same time as the order parameter. A discontinuity in the solution for , also mildly apparent at , reinforces this point of view.

To conclude this part: Model (1) indeed supports a pure bond density wave as the particular one studied here, but the VCA is not too reliable as to the precise doping range where it occurs. The BDW appears quite sensitive to the dispersion relation, as expected. It appears beyond a certain threshold value of and grows with up to intermediate coupling.

### iii.2 Bond density wave and d-wave Superconductivity

We now proceed to probe solutions in which d-wave superconductivity (dSC) coexists with the BDW. This is done by adding both BDW and dSC terms, as defined by Eqs (2) and (3), to the cluster Hamiltonian. We start by studying the case with nearest-neighbor hopping only (), for and . Fig. 6 shows the dSC and BDW order parameters both for the pure and the coexistence solutions as a function of the hole doping (the ‘pure’ solutions are obtained by setting one of the Weiss field s [ or ] to zero). The electron density was computed from the Green function. The pure dSC solution exists in a large doping interval, with a maximum at . Antiferromagnetic order was not put in competition with dSC order, since we are focusing on the interplay between dSC and BDW. The striking effect shown here is that the interaction of the two phases tends to suppress superconductivity at , whereas it tends to suppress charge order at .

Figure 7 shows results obtained with a more realistic band dispersion (), at . This time, the pure dSC solution always has a lower energy than the pure BDW solution, contrary to the solution found at . Nevertheless, the coexistence solution has a lower energy than the pure dSC solution from on. We do not believe the solutions beyond (not shown) to be reliable, as they display the same type of discontinuity as shown on Fig. 5. Note that the dSC solution reconnects to the pure-dSC solution when the BDW falls to zero, at .

The VCA approach used in this work has strengths and weaknesses. The main strength is that the problem can be treated at all: In the context of the intermediate couping, repulsive Hubbard model, VCA and other quantum cluster methods (CDMFT, DCA) are among the few options capable of revealing -wave superconductivity; the effective pairing interaction is dynamically generated within the clusters used. In addition, treating a period-four BDW can be done elegantly with clusters, which are too large to be treated with an exact diagonalization solver (hence at zero temperature) in any other approach than VCA. Seeing both BDW and dSC emerging from the same repulsive interaction, in coexistence, is particularly satisfying.

On the flip side, VCA, contrary to DMFT-like approaches, does not let particles in and out of the cluster (there are no bath degrees of freedom) and this leads to a less reliable estimate of the electron density. The approach is prone to (likely spurious) first-order phase transitions that occur when doping is pushed too far. In addition, the approach is numerically more delicate than DMFT and convergence may be more difficult. Finally, in practice, it can only probe orders that are defined from the outset (i.e. it is somewhat restricted).

## Iv Conclusion

Figure 7 is the most significant result of this work. It shows how a particular bond-density wave and d-wave superconductivity both emerge dynamically from the one-band Hubbard model and how they are intertwined in a finite range of hole doping, for a realistic dispersion relation.

In this work only local interactions are responsible for the establishment of both charge order and -wave superconductivity. It is reasonable to expect that extended interactions will reinforce the tendency towards charge order; in the context of quantum cluster methods, extended interactions need to be treated partly in the Hartree approximation (the so-called dynamical Hartree approximation Faye et al. (2015)). Work in that direction would be the natural next step.

###### Acknowledgements.
We thank A.-M. Tremblay for fruitful discussions. Computing resources were provided by Compute Canada and Calcul Québec. This research is supported by NSERC grant no RGPIN-2015-05598 (Canada) and by FRQNT (Québec).

### References

1. Tao Wu, Hadrien Mayaffre, Steffen Krämer, Mladen Horvatić, Claude Berthier, W. N. Hardy, Ruixing Liang, D. A. Bonn,  and Marc-Henri Julien, “Magnetic-field-induced charge-stripe order in the high-temperature superconductor YBa2Cu3Oy,” Nature, 477, 191–194 (2011), ISSN 0028-0836.
2. Tao Wu, Hadrien Mayaffre, Steffen Krämer, Mladen Horvatić, Claude Berthier, Philip L. Kuhns, Arneil P. Reyes, Ruixing Liang, W. N. Hardy, D. A. Bonn,  and Marc-Henri Julien, “Emergence of charge order from the vortex state of a high-temperature superconductor,” Nat. Commun., 4, 299–302 (2013), ISSN 2041-1723.
3. J. E. Hoffman, E. W. Hudson, K. M. Lang, V. Madhavan, H. Eisaki, S. Uchida,  and J. C. Davis, “A Four Unit Cell Periodic Pattern of Quasi-Particle States Surrounding Vortex Cores in Bi2Sr2CaCu2O8+,” Science (80-. )., 295 (2002).
4. Michael Vershinin, Shashank Misra, S. Ono, Y. Abe, Yoichi Ando,  and Ali Yazdani, “Local Ordering in the Pseudogap State of the High-Tc Superconductor Bi2Sr2CaCu2O8+,” Science (80-. )., 303 (2004).
5. G. Ghiringhelli, M. Le Tacon, M. Minola, S. Blanco-Canosa, C. Mazzoli, N. B. Brookes, G. M. De Luca, A. Frano, D. G. Hawthorn, F. He, T. Loew, M. Moretti Sala, D. C. Peets, M. Salluzzo, E. Schierle, R. Sutarto, G. A. Sawatzky, E. Weschke, B. Keimer,  and L. Braicovich, “Long-Range Incommensurate Charge Fluctuations in (Y,Nd)Ba2Cu3O6+x,” Science (80-. )., 337 (2012).
6. A. J. Achkar, R. Sutarto, X. Mao, F. He, A. Frano, S. Blanco-Canosa, M. Le Tacon, G. Ghiringhelli, L. Braicovich, M. Minola, M. Moretti Sala, C. Mazzoli, Ruixing Liang, D. A. Bonn, W. N. Hardy, B. Keimer, G. A. Sawatzky,  and D. G. Hawthorn, “Distinct Charge Orders in the Planes and Chains of Ortho-III-Ordered YBa 2 Cu 3 O 6 + Superconductors Identified by Resonant Elastic X-ray Scattering,” Phys. Rev. Lett., 109, 167001 (2012), ISSN 0031-9007.
7. J. Chang, E. Blackburn, A. T. Holmes, N. B. Christensen, J. Larsen, J. Mesot, Ruixing Liang, D. A. Bonn, W. N. Hardy, A. Watenphul, M. v. Zimmermann, E. M. Forgan,  and S. M. Hayden, “Direct observation of competition between superconductivity and charge density wave orderÂ in YBa2Cu3O6.67,” Nat. Phys., 8, 871–876 (2012), ISSN 1745-2473.
8. E. Blackburn, J. Chang, M. Hücker, A. T. Holmes, N. B. Christensen, Ruixing Liang, D. A. Bonn, W. N. Hardy, U. Rütt, O. Gutowski, M. v. Zimmermann, E. M. Forgan,  and S. M. Hayden, “X-Ray Diffraction Observations of a Charge-Density-Wave Order in Superconducting Ortho-II YBa 2 Cu 3 O 6.54 Single Crystals in Zero Magnetic Field,” Phys. Rev. Lett., 110, 137004 (2013), ISSN 0031-9007.
9. Y Kohsaka, C Taylor, K Fujita, A Schmidt, C Lupien, T Hanaguri, M Azuma, M Takano, H Eisaki, H Takagi, S Uchida,  and J C Davis, “An Intrinsic Bond-Centered Electronic Glass with Unidirectional Domains in Underdoped Cuprates,” Science (80-. )., 315, 1380–1385 (2007), ISSN 0036-8075.
10. A. J. Achkar, F. He, R. Sutarto, J. Geck, H. Zhang, Y.-J. Kim,  and D. G. Hawthorn, “Resonant X-Ray Scattering Measurements of a Spatial Modulation of the Cu 3 d and O 2 p Energies in Stripe-Ordered Cuprate Superconductors,” Phys. Rev. Lett., 110, 017001 (2013), ISSN 0031-9007.
11. M. H. Hamidian, S. D. Edkins, Sang Hyun Joo, A. Kostin, H. Eisaki, S. Uchida, M. J. Lawler, E.-A. Kim, A. P. Mackenzie, K. Fujita, Jinho Lee,  and J. C. Séamus Davis, “Detection of a Cooper-pair density wave in Bi2Sr2CaCu2O8+x,” Nature, 532, 343–347 (2016), ISSN 0028-0836.
12. Matthias Vojta, “Superconducting charge-ordered states in cuprates,” Phys. Rev. B, 66, 104505 (2002), ISSN 0163-1829.
13. Subir Sachdev and Rolando La Placa, “Bond Order in Two-Dimensional Metals with Antiferromagnetic Exchange Interactions,” Phys. Rev. Lett., 111, 027202 (2013), ISSN 0031-9007.
14. W A Atkinson, A P Kampf,  and S Bulut, “Charge order in the pseudogap phase of cuprate superconductors,” New J. Phys., 17, 013025 (2015), ISSN 1367-2630.
15. R.S. Markiewicz, J. Lorenzana, G. Seibold,  and A. Bansil, “Competing phases in the cuprates: Charge vs spin order,” J. Phys. Chem. Solids, 72, 333–336 (2011), ISSN 00223697.
16. H Terletska, T Chen,  and E Gull, “Charge ordering and correlation effects in the extended Hubbard model,” ArXiv e-prints (2016), arXiv:1611.07861 [cond-mat.str-el] .
17. Johannes Bauer and Alex C. Hewson, “Competition between antiferromagnetic and charge order in the Hubbard-Holstein model,” Phys. Rev. B, 81, 235113 (2010), ISSN 1098-0121.
18. Philippe Corboz, Steven R. White, Guifré Vidal,  and Matthias Troyer, “Stripes in the two-dimensional t - J model with infinite projected entangled-pair states,” Phys. Rev. B, 84, 041108 (2011), ISSN 1098-0121.
19. Manuela Capello, Marcin Raczkowski,  and Didier Poilblanc, “Stability of RVB hole stripes in high-temperature superconductors,” Phys. Rev. B, 77, 224502 (2008), ISSN 1098-0121.
20. A. Himeda, T. Kato,  and M. Ogata, “Stripe States with Spatially Oscillating d -Wave Superconductivity in the Two-Dimensional t â t â² â J Model,” Phys. Rev. Lett., 88, 117001 (2002), ISSN 0031-9007.
21. Marcin Raczkowski, Manuela Capello, Didier Poilblanc, Raymond Frésard,  and Andrzej M. Oleś, “Unidirectional d -wave superconducting domains in the two-dimensional t â J model,” Phys. Rev. B, 76, 140505 (2007), ISSN 1098-0121.
22. P Choubey, W.-L. Tu, T.-K. Lee,  and P.~J. Hirschfeld, “Incommensurate charge ordered states in the $t-t’-J$ model,” ArXiv e-prints (2016), arXiv:1609.07067 [cond-mat.supr-con] .
23. Hermann Freire, Vanuildo S. de Carvalho,  and Catherine Pépin, “Renormalization group analysis of the pair-density-wave and charge order within the fermionic hot-spot model for cuprate superconductors,” Phys. Rev. B, 92, 045132 (2015), ISSN 1098-0121.
24. Yuxuan Wang, Daniel F. Agterberg,  and Andrey Chubukov, “Coexistence of Charge-Density-Wave and Pair-Density-Wave Orders in Underdoped Cuprates,” Phys. Rev. Lett., 114, 197001 (2015), ISSN 0031-9007.
25. Eduardo Fradkin, Steven A. Kivelson,  and John M. Tranquada, “Colloquium : Theory of intertwined orders in high temperature superconductors,” Rev. Mod. Phys., 87, 457–482 (2015), ISSN 0034-6861.
26. C Dahnken, M Aichhorn, W Hanke, E Arrigoni,  and M Potthoff, “Variational cluster approach to spontaneous symmetry breaking: The itinerant antiferromagnet in two dimensions,” Phys. Rev. B, 70, 245110 (2004).
27. A I Lichtenstein and M I Katsnelson, “Antiferromagnetism and d-wave superconductivity in cuprates: A cluster dynamical mean-field theory,” Phys. Rev. B, 62, R9283—-R9286 (2000).
28. G Kotliar, S Y Savrasov, G Pálsson,  and G Biroli, “Cellular Dynamical Mean Field Approach to Strongly Correlated Systems,” Phys. Rev. Lett., 87, 186401 (2001).
29. Th Maier, M Jarrell, Th. Pruschke,  and J Keller, “d-wave superconductivity in the Hubbard model,” Phys. Rev. Lett., 85, 1524 – 7 (2000).
30. David Sénéchal, P. L. Lavertu, M. A. Marois,  and A.-M.S. Tremblay, “Competition between antiferromagnetism and superconductivity in high-T c cuprates,” Phys. Rev. Lett., 94, 156404 (2005), ISSN 00319007.
31. Th Maier, M Jarrell, T C Schulthess, P R C Kent,  and J B White, “Systematic Study of d-Wave Superconductivity in the 2D Repulsive Hubbard Model,” Phys. Rev. Lett., 95, 237001 (2005).
32. S S Kancharla, B Kyung, David Sénéchal, M Civelli, M Capone, G Kotliar,  and A.-M. S Tremblay, “Anomalous superconductivity and its competition with antiferromagnetism in doped Mott insulators,” Phys. Rev. B, 77, 184516 (2008).
33. Andrea Allais, Debanjan Chowdhury,  and Subir Sachdev, “Connecting high-field quantum oscillations to zero-field electron spectral functions in the underdoped cuprates,” Nat. Commun., 5, 5771 (2014).
34. Kazuhiro Fujita, Mohammad H Hamidian, Stephen D Edkins, Chung Koo Kim, Yuhki Kohsaka, Masaki Azuma, Mikio Takano, Hidenori Takagi, Hiroshi Eisaki, Shin-Ichi Uchida, Andrea Allais, Michael J Lawler, Eun-Ah Kim, Subir Sachdev,  and J C Séamus Davis, “Direct phase-sensitive identification of a d-form factor density wave in underdoped cuprates.” Proc. Natl. Acad. Sci. U. S. A., 111, E3026–32 (2014), ISSN 1091-6490.
35. M Aichhorn, E Arrigoni, M Potthoff,  and W Hanke, “Phase separation and competition of superconductivity and magnetism in the two-dimensional {Hubbard} model: From strong to weak coupling,” Phys. Rev. B, 76, 224509 (2007).
36. Michael Potthoff, “Self-energy-functional theory,” in Theor. methods Strongly Correl. Syst., Springer Series in Solid-State Sciences, Vol. 171, edited by Adolfo Avella and Ferdinando Mancini (Springer, 2012) Chap. 9.
37. Michael Potthoff, “Variational Cluster Approximation,” in DMFT 25 Infin. Dimens. Lect. Notes Autumn Sch. Correl. Electrons 2014, edited by Eva Pavarini, Erik Koch, Dieter Vollhardt,  and Alexander Lichtenstein (Forschungszentrum J{ü}lich, 2014).
38. M Potthoff, “Self-energy-functional approach to systems of correlated electrons,” Eur. Phys. J. B, 32, 429–436 (2003), ISSN 14346028, arXiv:0301137 [cond-mat] .
39. J P L Faye, P Sahebsara,  and David Sénéchal, “Chiral triplet superconductivity on the graphene lattice,” Phys. Rev. B, 92, 085121 (2015), ISSN 1098-0121.
101810