Evolution of magnetic correlation in doped Hubbard model with altermagnetic spin splitting Yinlong Li, Rana Imran Mushtaq, Ji Liu, Wing Chi Yu, Xiaosen Yang, Cho-Tung Yip, Ho-Kin Tang Abstract The evolution of magnetic correlation in strongly correlated electron systems with altermagentic spin splitting remains largely unexplored. Here we investigate how spin splitting generated by spin-dependent next-nearest-neighbor hopping t ′ t^{\prime} reshapes the Fermi surface nesting and van Hove singularities in the two-dimensional square-lattice Hubbard model, leading evolution of magnetic instabilities. Using the constrained-path quantum Monte Carlo method, we find the dominant magnetic correlation as functions of the filling and t ′ / t t^{\prime}/t by computing the momentum-resolved spin structure factor. The analysis reveals a transition from antiferromagnetic ( π , π ) (\pi,\pi) order in the isotropic, half-filled system to non-collinear spiral ( π , q ) (\pi,q) order upon increasing the spin-dependent anisotropy or doping away from half-filling, ultimately entering a short-range correlation regime where stripe and spiral correlation coexist. These findings highlight a possible route to realizing spiral correlation in altermagnetic systems, potentially providing a platform for spintronic devices that exploit non-collinear spin textures. I I. INTRODUCTION In strongly correlated electron systems, the interplay of doping, lattice geometry, and electronic correlation engenders a rich landscape of competing magnetic phases, exemplified by the evolution from antiferromagnetism to complex spin textures in materials such as cuprates Bednorz and Müller 1986 ; Goremykin and Katanin 2024 ; Wagman et al. 2013 ; Kastner et al. 1998 . Theoretically, this magnetic reconstruction is intrinsically governed by the structure of the Fermi surface and the associated particle–hole nesting instability He et al. 2019 ; Khokhlov et al. 2020 . Specifically, when extensive regions of the Fermi surface are connected by a magnetic wave vector, the spin structure factor exhibits a pronounced peak, promoting the formation of magnetic correlation Igoshev et al. 2010 . As carrier concentration varies, the consequent deformation of the Fermi surface shifts the optimal nesting vector from the (\pi,\pi) point to spiral values, thereby driving the transition between distinct magnetic ground states, such as spiral order Fujita et al. 2002 ; Yamada et al. 1998 ; Markiewicz and Bansil 2024 ; Kang et al. 2019 . Figure 1: Schematic representation of the spin-anisotropic t^{\prime} Hubbard model and its resulting band splitting. (a) Square lattice with on-site interaction U and sign-alternating next-nearest-neighbor (NNN) hopping \pm t^{\prime} , giving opposite effective NNN hopping for the two spin species while maintaining zero net magnetization. (b) Noninteracting band dispersion along the high-symmetry path of the Brillouin zone and the corresponding spin-resolved Fermi surface (inset) for t^{\prime}=0 , where the spectrum is spin degenerate. (c) Noninteracting band dispersion and spin-resolved Fermi surface for t^{\prime}=0.5 , where the term 4t^{\prime}\sin k_{x}\sin k_{y} induces momentum-dependent spin splitting and distinct Fermi-surface contours for spin up (red) and spin down (blue). To explore these phenomena theoretically, the two-dimensional Hubbard model is widely recognised as the standard starting point for studying doped Mott insulators Dong et al. 2022 ; Scalapino 2012 . Despite its formal simplicity—comprising only kinetic terms describing electronic transitions and a local Coulomb repulsion term U —this model successfully captures the essence of strongly correlated physics: the competition between kinetic energy and strong electron interactions Arovas et al. 2022 ; Qin et al. 2022 . It is precisely this competition that gives rise to a variety of intertwined and competing quantum phases, including antiferromagnetism and charge ordering Beenen and Edwards 1995 . Consequently, the Hubbard model has become an indispensable theoretical tool for studying how strong correlation reshape competing particle–hole instabilities across doping and band-structure tuning Dong et al. 2022 . However, the minimal Hubbard model containing only nearest-neighbor (NN) hopping ( t ) exhibits particle-hole symmetry at half-filling, which contradicts the experimentally observed significant asymmetry in the electron-doping versus hole-doping phase diagrams of copper oxides Rashid and Singh 2024 . The introduction of a next-nearest-neighbor (NNN) hopping term t^{\prime} to construct the so-called t-t^{\prime} Hubbard model represents a crucial and physically meaningful extension Rashid and Singh 2024 . A non-zero t^{\prime} term not only explicitly breaks the particle-hole symmetry, making the Fermi surface more realistic, but also profoundly affects the nesting conditions Fontenele et al. 2023 . Consequently, t^{\prime} plays a decisive role in regulating the competition between different ground states—particularly between antiferromagnetic ordering and spiral magnetic ordering Beenen and Edwards 1995 —and is crucial for the evolution and distribution of magnetic order discussed herein Arzhnikov and Groshev 2012 . Although the NNN hopping t^{\prime} is known to reshape the electronic structure by shifting van Hove singularitie (VHS) and redistributing low-energy intensity in momentum space Cade et al. 2020 ; Markiewicz et al. 2023 , its precise role in driving the evolution of magnetic correlation remains unclear. This unresolved issue is central to understanding how competing magnetic instabilities emerge in the doped Hubbard model, where modest changes in the Fermi surface can qualitatively alter the dominant ordering tendencies. Previous studies indicate that t^{\prime} can influence the magnetic order, resulting in the formation of stripe or non-conjugate magnetic correlation Markiewicz and Bansil 2024 . A sufficiently large t^{\prime} has been argued to stabilize a single spiral magnetic correlation over a broad range of carrier concentrations Beenen and Edwards 1995 , whereas small t^{\prime} or increased doping typically produces a more intricate structure of the bare susceptibility, leading to near-degeneracy among multiple competing ordering vector and rendering the phase competition highly sensitive to microscopic details. In this study, we employ the Hubbard model to establish a theoretical framework that elucidates the regulatory role of the NNN hopping, t^{\prime} , in the competition between collinear and spiral magnetic correlation. By systematically analyzing the remodeling of band geometry and Fermi surface nesting induced by t^{\prime} , we aim to clarify how such kinetic frustration governs the emergence, competition, and ultimate selection of magnetic ground states in correlated electron systems. The remainder of this paper is organized as follows: Section 2 outlines the theoretical framework and computational methodology; Section 3 presents our numerical results and detailed analysis; and Section 4 provides a conclusion and discussion of our findings. II II. Model and Method II.1 A. Next-nearest-neighbor Hubbard model The Hamiltonian of the Hubbard model is \displaystyle\hat{H}= \displaystyle\Big[-t\sum_{\langle ij\rangle,\sigma}\hat{c}_{i\sigma}^{\dagger}\hat{c}_{j\sigma}-\sum_{i,\sigma}\Big(t^{\prime}_{\hat{x}+\hat{y},\sigma}\hat{c}^{\dagger}_{i\sigma}\hat{c}_{i+\hat{x}+\hat{y},\sigma} (1) \displaystyle-t^{\prime}_{\hat{x}-\hat{y},\sigma}\hat{c}^{\dagger}_{i\sigma}\hat{c}_{i+\hat{x}-\hat{y},\sigma}\Big)+h.c.\Big]+U\sum_{i}\hat{n}_{i\uparrow}\hat{n}_{i\downarrow}-\mu\sum_{i\sigma}\hat{n}_{i\sigma}, where i and j label lattice sites, \hat{c}^{\dagger}_{i\sigma} ( \hat{c}_{i\sigma} ) creates (annihilates) an electron with spin \sigma\in\{\uparrow,\downarrow\} at site i , and \hat{n}_{i\sigma}=\hat{c}^{\dagger}_{i\sigma}\hat{c}_{i\sigma} is the local number operator. The first term describes NN hopping on the square lattice, where \langle ij\rangle denotes NN bonds. We set the NN hopping t as the energy unit. For the NNN directions, for different spins, the t^{\prime} term carries different signs in different directions, for up spin, t^{\prime}_{x+y,\uparrow}=t^{\prime} and t^{\prime}_{x-y,\uparrow}=-t^{\prime} ; for down spin, t^{\prime}_{x+y,\downarrow}=-t^{\prime} , t^{\prime}_{x-y,\downarrow}=t^{\prime} . The spin-dependent NNN hopping generates a momentum-dependent spin splitting already at the single-particle level, while maintaining a vanishing net magnetization in the absence of interactions. The third term is the on-site Hubbard interaction U , and \mu is the chemical potential controlling the filling n=\sum_{\sigma}\hat{n}_{i\sigma} . We set the on-site interaction U=4 , which places the system in the intermediate coupling regime, where the interplay between kinetic energy and electron-electron interactions is most pronounced. The interaction is sufficient to induce robust magnetic correlation and potential ordering instabilities without fully localizing the charge carriers, thereby allowing for a detailed examination of the competition between Fermi surface nesting effects and strong correlation physics. We decompose the Hamiltonian into a kinetic (noninteracting) part, H_{K} , and an interaction part, H_{N} . The noninteracting term H_{K} is diagonalized in momentum space, yielding the noninteracting band structure. Its explicit form is given in Eq. ( 2 ). \varepsilon(\mathbf{k})=-2t\bigl(\cos k_{x}+\cos k_{y}\bigr)\pm 4t^{\prime}\sin k_{x}\sin k_{y}+\mu\,. (2) The altermagnetic term t^{\prime} significantly changes the shape of Fermi surface. Figure 1 schematically summarizes the Hubbard model considered in this work and the spin splitting induced by the NNN hopping t^{\prime} . To further investigate the spin splitting phenomenon in magnetism, we introduce the total momentum domain spin polarization \Delta_{tot} : \Delta_{tot}=\sum_{k}|\Delta n(k)|=\sum_{k}|n_{\uparrow}(k)-n_{\downarrow}(k)|. (3) Here, \Delta n(k) represents the spin polarization in momentum space. A nonzero value of this quantity reveals the presence of momentum-dependent spin splitting, which breaks the local spin degeneracy in reciprocal space despite the system exhibiting zero net magnetization globally. Through a comprehensive scan of the (t^{\prime}/t,n) parameter space (where t^{\prime}/t\in[0.1,1] and n\in[0.5,1] ), the system systematically reveals symmetry-dependent magnetic correlation evolution results for the two-dimensional t-t^{\prime} Hubbard model. We employ the constrained-path quantum Monte Carlo (CPQMC) method using a Hartree–Fock trial wave function Zhang et al. 1997 ; Zhang et al. 1995 , which accurately captures quantum fluctuations and many-body correlation. By imposing a constrained-path approximation, CPQMC mitigates the fermion sign problem, enabling reliable simulations of the ground state property away from half filling. We study the system with lattice size L=16\times 16 , that provides momentum space resolution to resolve key features in the Brillouin zone, including high symmetry points ( \Gamma=(0,0) , X= ( \pi ,0)). The hyper-parameters used in our CPQMC calculations are as follows: the number of walkers is N_{w}=1000 ; the number of blocks for relaxation is N_{\mathrm{rel}}=10 ; the number of blocks for growth estimate is N_{\mathrm{gr}}=3 ; the number of blocks after relaxation is N_{\mathrm{meas}}=10 ; the number of steps in each block is N_{s}=640 ; the Trotter step size is \Delta\tau=0.01 ; and the growth-control energy estimate is E_{T}=-50 . In this work, magnetic ordering tendencies are diagnosed from equal-time spin correlation, which can be evaluated from one-body Green’s functions defined by fermionic creation and annihilation operators. We introduce the (equal-time) one-body density matrix G^{\sigma}_{ij}\equiv\left\langle\hat{c}_{i\sigma}\hat{c}^{\dagger}_{j\sigma}\right\rangle, (4) where \hat{c}^{\dagger}_{i\sigma} ( \hat{c}_{i\sigma} ) creates (annihilates) an electron with spin \sigma on site i . The local spin operator is \hat{S}_{i}^{z}=\frac{1}{2}\left(\hat{n}_{i\uparrow}-\hat{n}_{i\downarrow}\right),\qquad\hat{n}_{i\sigma}=\hat{c}^{\dagger}_{i\sigma}\hat{c}_{i\sigma}. (5) The equal-time spin correlation function is then C^{zz}(i,j)\equiv\langle\hat{S}_{i}^{z}\hat{S}_{j}^{z}\rangle=\frac{1}{4}\Big\langle(\hat{n}_{i\uparrow}-\hat{n}_{i\downarrow})(\hat{n}_{j\uparrow}-\hat{n}_{j\downarrow})\Big\rangle. (6) For a Slater-determinant walker |\phi\rangle (the fundamental object in CPQMC propagation), Wick’s theorem reduces the required two-body density correlators to products of one-body Green’s functions. In particular, for i\neq j one obtains \displaystyle\langle\hat{n}_{i\sigma}\hat{n}_{j\sigma}\rangle_{\phi} \displaystyle=\langle\hat{n}_{i\sigma}\rangle_{\phi}\langle\hat{n}_{j\sigma}\rangle_{\phi}-\left|\langle\hat{c}_{i\sigma}\hat{c}^{\dagger}_{j\sigma}\rangle_{\phi}\right|^{2} (7) \displaystyle=G^{\sigma}_{ii}(\phi)G^{\sigma}_{jj}(\phi)-|G^{\sigma}_{ij}(\phi)|^{2}, \displaystyle\langle\hat{n}_{i\uparrow}\hat{n}_{j\downarrow}\rangle_{\phi}=\langle\hat{n}_{i\uparrow}\rangle_{\phi}\langle\hat{n}_{j\downarrow}\rangle_{\phi}=G^{\uparrow}_{ii}(\phi)G^{\downarrow}_{jj}(\phi), (8) where G^{\sigma}_{ij}(\phi)=\langle\hat{c}^{\dagger}_{i\sigma}\hat{c}_{j\sigma}\rangle_{\phi} is computed from the corresponding walker Green’s function. Substituting Eqs. ( 7 )–( 8 ) into Eq. ( 6 ) yields \displaystyle C^{zz}(i,j)= \displaystyle\frac{1}{4}\Big(G^{\uparrow}_{ii}-G^{\downarrow}_{ii}\Big)\Big(G^{\uparrow}_{jj}-G^{\downarrow}_{jj}\Big)- (9) \displaystyle\frac{1}{4}\Big(|G^{\uparrow}_{ij}|^{2}+|G^{\downarrow}_{ij}|^{2}\Big),\qquad(i\neq j), and the CPQMC estimate is obtained by averaging C^{zz}(i,j) over the walker ensemble (with the chosen estimator). The spin structure factor is the Fourier transform of C^{zz}(i,j) , \displaystyle S^{z}(\mathbf{k})=\frac{1}{N}\sum_{i,j}e^{i\mathbf{k}\cdot(\mathbf{r}_{i}-\mathbf{r}_{j})}\,C^{zz}(i,j) \displaystyle=\frac{1}{N}\left\langle\hat{S}^{z}_{\mathbf{k}}\hat{S}^{z}_{-\mathbf{k}}\right\rangle, (10) \displaystyle\quad\Rightarrow\quad\hat{S}^{z}_{\mathbf{k}}=\sum_{j}e^{-i\mathbf{k}\cdot\mathbf{r}_{j}}\hat{S}^{z}_{j}. (11) Finally, we define the dominant magnetic ordering vector by the location of the maximum of S^{z}(\mathbf{k}) : \mathbf{Q}=\arg\max_{\mathbf{k}}S^{z}(\mathbf{k}). (12) This provides a unified criterion to classify the ordering vector throughout the phase diagram: \displaystyle\mathbf{Q}= \displaystyle(\pi,\pi)\quad\Rightarrow\quad\text{N\'{e}el antiferromagnetism}, \displaystyle\mathbf{Q}= \displaystyle(\pi,0)\ \text{or}\ (0,\pi)\quad\Rightarrow\quad\text{Stripe magnetism}, \displaystyle\mathbf{Q}= \displaystyle(\pi,q)\ \text{or}\ (q,\pi),\ \ q\neq 0,\pi\quad\Rightarrow\quad\text{Spiral magnetism}. (13) In this way, the magnetic order classification used in this work is directly rooted in Green’s function based measurements: one-body Green’s functions determine C^{zz}(i,j) via Wick contractions, whose Fourier transform yields S^{z}(\mathbf{k}) and hence the ordering vector \mathbf{Q} . To distinguish between long-range magnetic order and short-range magnetic correlation, we employ a complementary analysis combining momentum space spectral features with real space correlation decay. In the momentum space, the static S^{z}(\mathbf{k}) serves as the primary diagnostic indicator. A system possessing long-range order is characterized by sharp peak at specific S^{z}(\mathbf{k}) , with peak intensities remaining significantly enhanced even as system size increases. Conversely, short-range order manifests as broadened or diffuse weight centered around \mathbf{Q} . The finite width of these peaks in reciprocal space is inversely proportional to the magnetic correlation length in real space; thus, a diffuse peak signifies a finite short-range order, indicating that spin coherence is limited to local clusters. Through further analysis of the real space spin-spin correlation function C_{\text{SDW}}(\mathbf{r}) , the differences between the two exhibit qualitative and distinct characteristics: In the long-range sequence C_{\text{SDW}}(\mathbf{r}) , the spatial oscillations maintain finite, nonzero amplitudes throughout the lattice, constrained only by the system boundaries in finite-size simulations. In contrast, the short-range correlation function exhibits exponential decay. Although local magnetic oscillations may persist over a few lattice spacings, the signal vanishes rapidly at large distances, confirming the absence of global phase coherence. By combining the sharpness of the S(\mathbf{k}) peak with the asymptotic behavior of C_{\text{SDW}}(\mathbf{r}) , the nature of magnetic order can be reliably identified. II.2 B. Van Hove Singularities We evaluate the partial derivatives of the band dispersion \varepsilon(\mathbf{k}) with respect to the momentum components, and determine the stationary points at which these derivatives vanish. \displaystyle\frac{\partial\varepsilon}{\partial k_{x}}=2t\sin k_{x}\pm 4t^{{}^{\prime}}\cos k_{x}\sin k_{y}, (14) \displaystyle\frac{\partial\varepsilon}{\partial k_{y}}=2t\sin k_{y}\pm 4t^{{}^{\prime}}\cos k_{y}\sin k_{x}. The VHS represent features of the band structure that occur at stationary points of the dispersion \nabla_{\mathbf{k}}\varepsilon_{\mathbf{k}\sigma}=0 Van Hove 1953 . In two dimensions, such saddle points generate a logarithmic enhancement of the density of states (DOS), which strongly amplifies low-energy phase space for both particle-hole and particle-particle processes. The nature of the VHS is dictated by the local curvature of the energy band, characterized by the Hessian matrix of the dispersion, \mathcal{H}_{xy}(\mathbf{k})=\partial^{2}\varepsilon/\partial k_{x}\partial k_{y} Yuan et al. 2019 . We computed the VHS structure for t^{\prime}\in[0.1,0.9] and summarize the complete set of results in Appendix B. In the main text, we focus on three representative values, t^{\prime}=0.1 , 0.5 , and 0.9 , which capture the key trends across this range. In the regime of small t^{\prime} (e.g., t^{\prime}=0.1t ), the VHS are located at the time-reversal invariant momenta M=(\pi,0) and (0,\pi) . At these points, the band exhibits a saddle-point geometry with finite curvature, where the band disperses upwards along one principal axis and downwards along the other. This results in a logarithmic divergence of the DOS, N(\varepsilon)\sim\ln|\varepsilon|^{-1} . The Fermi surface exhibits good nesting properties with a nesting vector \mathbf{Q}\approx(\pi,\pi) , favoring an antiferromagnetic order Hlubina et al. 1997 . A transition occurs at the critical value t^{\prime}=0.5t . Here, the VHS shift to the \Gamma point (0,0) and the K point (\pi,\pi) . Crucially, the curvature along the node direction ( k_{x}=\pm k_{y} ) approaches zero, leading to a significant enhancement in curvature divergence. This flat-band dispersion leads to a higher-order van Hove singularity (HO-VHS), characterized by a stronger, power-law divergence in the DOS, N(\varepsilon)\sim|\varepsilon|^{-1/4} Yuan et al. 2019 ; Shtyk et al. 2017 ; Isobe and Fu 2019 . According to the Stoner criterion, this divergent DOS significantly amplifies the influence of electron correlation, driving the system towards a complex non-collinear ordering, even at moderate interaction strengths Schulz 1987 . For large t^{\prime} (e.g. t^{\prime}=0.9t ), the singularities at the high-symmetry points bifurcate and migrate to spiral positions in the Brillouin zone. The curvature becomes finite again, recovering the logarithmic nature of the DOS singularity. However, the modification of the Fermi surface geometry destroys the perfect (\pi,\pi) nesting, suppressing AFM order in favor of spiral magnetic phases or collinear stripe correlation dictated by the evolved nesting vector \mathbf{Q}(t^{\prime}) . As shown in Fig. 2 , in the present spin-split NNN Hubbard model, the location and energetic relevance of the VHS are strongly controlled by the NNN hopping amplitude t^{\prime} . In particular, increasing t^{\prime} continuously shifts the saddle-point energy relative to the Fermi level and simultaneously deforms the Fermi surface, thereby modifying the dominant low-energy scattering channels. Our results reveal a qualitative boundary around t^{\prime}\!\approx\!0.5 : for t^{\prime}\lesssim 0.5 the VHS remains “close” to the conventional square-lattice saddle-point sector and the Fermi surface retains substantial diagonal connectivity, whereas for t^{\prime}\gtrsim 0.5 the band reconstruction driven by spin-dependent NNN hopping qualitatively reshapes the Fermi surface into strongly anisotropic pocket-like structures, indicating that the system has crossed into a distinct kinematic regime. In this sense, t^{\prime}\!\approx\!0.5 acts as an effective demarcation line separating two different VHS-controlled Fermi-surface topologies and, correspondingly, two different magnetic-selection mechanisms. Figure 2: The evolution of van Hove singularities (VHS) and Fermi surfaces with varying NNN hopping t^{\prime} . The panels correspond to (a) t^{\prime}=0.1 , (b) t^{\prime}=0.5 , and (c) t^{\prime}=0.9 . For each t^{\prime} , solid (dashed) contours denote the Fermi surfaces of the upper (lower) band at fillings n=0.5 and n=0.9 , while the shaded regions indicate the corresponding filled Fermi surface. Stars mark the VHS. As t^{\prime} increases, the VHS manifests not only at high symmetry points and the Fermi surfaces become strongly distorted, degrading (\pi,\pi) nesting and thereby favoring (\pi,q) or (q,\pi) spiral tendencies and, at larger t^{\prime} , stripe collinear configurations (\pi,0) or (0,\pi) . To corroborate the evolution of the electronic structure, we calculate the total density of states (DOS) and the \nabla_{\mathbf{k}}\varepsilon_{\mathbf{k}\sigma} along the nodal diagonal ( k_{x}=k_{y} ), as shown in Fig. 3 . For t^{\prime}=0.1 , the DOS exhibits a conventional logarithmic divergence characteristic of a standard saddle point Fig. 3 (a). The group velocity along the diagonal vanishes linearly only at the high-symmetry points \Gamma ( k=0 ) and M^{\prime} ( k=\pm\pi ), consistent with a quadratic band dispersion in Fig. 3 (d). A critical transition occurs at t^{\prime}=0.5 . Here, the \nabla_{\mathbf{k}}\varepsilon_{\mathbf{k}\sigma} vanishes tangentially at the zone center in Fig. 3 (e), indicating a vanishing band curvature ( \partial^{2}\varepsilon/\partial k^{2}=0 ). This flattening of the dispersion confirms the emergence of a HO-VHS, where the dispersion is dominated by quartic terms ( \sim k^{4} ). Consequently, the DOS singularity becomes sharper, reflecting the divergence of the effective mass in Fig. 3 (b). In the strong spin splitting ( t^{\prime}=0.9 ), the single VHS at \Gamma bifurcates. This splitting is explicitly captured by the group velocity profile in Fig. 3 (f), where new nodes (marked by red circles) emerge at momenta away from k=0 . These nodes correspond directly to the split peaks observed in the DOS in Fig. 3 (c), signifying the migration of VHS into the Brillouin zone interior. Figure 3: The evolution of the electronic density of states (DOS) as a function of the NNN hopping amplitude t^{\prime} . The panels (a)–(c) display the calculated total DOS for t^{\prime}=0.1 , t^{\prime}=0.5 , and t^{\prime}=0.9 , respectively. The energy E is normalized by the NN hopping t . Vertical dashed lines mark the energy E=0 , typically corresponding to the Fermi level at half-filling. The prominent peaks in the DOS spectra identify the positions of VHS, which arise from saddle points in the momentum space dispersion relation. The panels (d)–(f) depict the corresponding \nabla_{\mathbf{k}}\varepsilon_{\mathbf{k}\sigma} along the diagonal momentum cut. The red circles mark the critical points where the \nabla_{\mathbf{k}}\varepsilon_{\mathbf{k}\sigma} vanishes, directly corresponding to the VHS locations. III III. RESULTS III.1 A. Magnetic phase diagram The distribution of S^{z}(\mathbf{k}) peaks as functions of the electron filling n and NNN hopping t^{\prime} is shown in Fig. 4 . The figure clearly demonstrates the transition between different types of magnetic correlation, with antiferromagnetic order, spiral correlation, and collinear stripe magnetic correlation exhibiting consistent patterns across various lattice sizes. In the low-doping regime, the system stabilizes into Néel antiferromagnetic order, which is characterized by a pronounced S^{z}(\mathbf{k}) response at the (\pi,\pi) point in momentum space. As t^{\prime} increases, the system’s magnetism undergoes a gradual transition towards a non-collinear magnetic state, with peaks shifting to non-collinear S^{z}(\mathbf{k}) , particularly at (\pi,q) or (q,\pi) . We examined the magnetic phase diagram at different lattice sizes and found that the overall magnetic behavior exhibits robustness to variations in system size under the same analytical framework. Specifically, phase boundaries and qualitative ordering trends remain consistent across different parameter ranges, with relevant results summarized in Appendix A. These observations point to a progressive transition from Néel antiferromagnetic order to stripe and spiral states as the doping concentration n and the NNN hopping t^{\prime} are varied. The NNN hopping term t^{\prime} introduces a momentum-dependent spin splitting, causing distinct spin branches to have different energies in momentum space. This could result in the non-collinearity of the spin alignment. The phase boundary distinguishing long-range magnetic order from short-range correlation is delineated in the Fig. 4 . The long-range phase is predominantly associated with the ordering vector \mathbf{Q}=(\pi,\pi) , (\pi,q) or (q,\pi) , and is characterized by the emergence of sharp, well-defined peaks—as opposed to diffuse features—in the SDW momentum space spectrum. Notably, this long-range order demonstrates robustness against finite size effects, a property that will be elaborated upon in the subsequent sections. As t^{\prime} increases, the system transitions from a collinear magnetic ordered state to a stripe or spiral configuration, with the corresponding ordering vector shifting from (\pi,\pi) to (\pi,q) , thus demonstrating a continuous change in the system’s magnetic phase driven by the interplay between electronic interactions and the hopping parameters. For fillings n\simeq 0.5 and small NNN hopping t^{\prime} , the S^{z}(\mathbf{k}) develops pronounced peaks at both \mathbf{Q}=(\pi,\pi) and (\pi,0) or (0,\pi) ). This double-peak structure reflects a regime of strong competition between Néel order and stripe correlation tendencies accompanied by the coexistence of two magnetism. The quasi-square Fermi contour retains strong particle–hole scattering at \mathbf{Q}=(\pi,\pi) , thereby stabilizing Néel antiferromagnetism. Upon doping away from half filling, the contour develops elongated axial features that enhance low-energy scattering at \mathbf{Q}=(\pi,0) or (0,\pi) , which in turn promotes collinear stripe correlation. At high filling ( n\sim 1 ) and low NNN hopping strength ( t^{\prime}\sim 0 ), the magnetism of the system is primarily driven by electron-electron interactions. In this regime, as shown in Fig. 4 , for n=0.953 and t^{\prime}=0.1 , perfect Fermi surface nesting results in the antiferromagnetic ordering vector \mathbf{Q}=(\pi,\pi) being the most energetically favourable configuration, stabilising Néel antiferromagnetic order. As the doping concentration n decreases and t^{\prime} increases, the NNN hopping t^{\prime} introduces momentum-dependent spin splitting, where distinct spin branches exhibit different energy levels in momentum space. For instance, when n=0.789 and t^{\prime}=0.5 and n=0.945 with t^{\prime}=0.5 , the antiferromagnetic order stability is disrupted, with peaks appearing at (\pi,q) or (q,\pi) , indicating a transition towards a spiral magnetism. When n is small and t^{\prime}\sim 0.4 , the system cannot maintain its initial antiferromagnetic Néel state, increasingly tending towards non-collinear spin configurations (such as the spiral state) or collinear stripe state. The ordering vector shifts from the collinear point (\pi,\pi) towards non-collinear points (e.g., (\pi,q) ) or the collinear point (\pi,0) , when n=0.633 and t^{\prime}\leq 0.5 , Néel antiferromagnetic order coexists with collinear stripe magnetic correlation. Upon further decreasing n and increasing t^{\prime} , the system response evolves into a collinear state, where the spin distribution corresponding to collinear stripe magnetic correlation becomes uniform. As n further decreases and t^{\prime}\sim 1 , the real space spin distribution approaches uniformity, with the dominant magnetism fully transitioning to either the collinear stripe correlation (\pi,0) or (0,\pi) . Figure 4: The ordering vector distribution of dominant S^{z}(\mathbf{k}) peak locations in the (n,t^{\prime}) plane for an L=16 lattice. The color scale shows the S^{z}(\mathbf{k}) peak coordinate Q at which |S^{z}(\mathbf{k})| is maximal: Q=1 corresponds to Néel order at (\pi,\pi) , Q=0 to collinear stripe correlation at (\pi,0) or (0,\pi) , and 00.5 the VHS detach from these high–symmetry points and migrate onto the diagonals k_{x}=\pm k_{y} , giving rise to spiral saddle points. As t^{\prime} is further increased, the VHS trajectories spread out along these diagonals and the corresponding saddle energies become more strongly dispersed, reflecting the enhanced anisotropy of the band structure induced by t^{\prime} . Figure 14: The evolution of VHS and Fermi surfaces with varying NNN hopping t^{\prime} . Panels (a)–(i) correspond to t^{\prime}=0.1,0.2,\ldots,0.9 . For each t^{\prime} , solid (dashed) contours denote the Fermi surfaces of the upper (lower) band at fillings n=0.5 and n=0.9 , while the shaded regions indicate the corresponding filled Fermi surface. Stars mark momenta where \nabla_{\mathbf{k}}\varepsilon_{\mathbf{k}}=0 , i.e., VHS saddle points. As t^{\prime} increases, the VHS move away from the high-symmetry points and the Fermi surfaces become strongly distorted, degrading (\pi,\pi) nesting and thereby favoring (\pi,q) or (q,\pi) spiral tendencies at larger t^{\prime} . This systematic evolution of the VHS distribution encapsulates the way in which the NNN hopping t^{\prime} reshapes the Fermi–surface and modifies the nesting conditions. In particular, the relocation of the VHS from high–symmetry points to spiral momenta directly impacts the dominant magnetic instabilities. Combined with the main–text analysis of the S^{z}(\mathbf{k}) response, these results demonstrate that tuning t^{\prime} provides an efficient handle to control both the position and the spectral weight of the VHS, thereby governing the evolution of the system’s magnetic correlation tendencies.