REVIEW
Quantum periodic cluster methods for strongly correlated electron systems
Strongly Correlated Electrons · Submitted 27 Feb 2006 · 22 references
Pith's verdict: A clean canonical reformulation that unifies DCA and CDMFT and a promising PCPT variant, but the O(1/Lc^2) convergence claim is conditional on an unproven constraint and one benchmark.
T0 · AI review
Why Pith says this
Highlight any text to comment on it.
Referee report
The paper
The paper claims
Cluster method converges quadratically, not linearly, in cluster size.
The reading
What carries the argument
The central object is the canonical periodization transformation, which creates fermion operators periodic in both the superlattice and cluster spaces so that the intra-cluster wave vector becomes a conserved quantum number. Combined with the constraint (16) that sets the umklapp momentum transfer from the superlattice to the cluster space to zero, it turns the original Hamiltonian into a periodized one whose hopping and interaction are diagonal in the reciprocal cluster space. This diagonalization is what avoids real-space matrix inversions and produces the small parameter: the inter-cluster hopping $\delta\bar{t}$ scales as $1/L_c$, so the hybridization in the locator expansion, which controls the error of the coarse-grained Green function, scales as $1/L_c^2$ for PCPT versus $1/L_c$ for standard CPT.
Extended reading notes
Core claim
The paper's core claim is that a canonical transformation can periodize fermions in the cluster space, and that this periodization unifies the DCA and the CDMFT while enabling a faster-converging cluster perturbation theory. The transformation maps the original fermion operators to periodic ones $\bar{c}_{a\sigma}(I)$ that are periodic in the cluster index; the associated Hamiltonian has periodized hopping $\bar{t}_{ab}(I,J)$ and interaction $\bar{U}_{ab}(I,J)$ that are cyclic in $a,b$, so all cluster quantities are diagonal in the reciprocal cluster space. The equivalence with the original Hamiltonian holds only after imposing the constraint (16), which forbids the umklapp momentum transfer from the superlattice to the cluster space. Under this constraint, applying the CDMFT to the periodized Hamiltonian reproduces the DCA equations, giving both methods a common microscopic definition. The PCPT then approximates the lattice self-energy by the exactly solved cluster self-energy $\bar{\Sigma}^c_{\bar{k}}(z)$, yielding the lattice Green function $G(K+\bar{k},z)=[z-t(K+\bar{k})-\bar{\Sigma}^c_{\bar{k}}(z)]^{-1}$. In the exactly solvable one-dimensional large-$N$ benchmark, the error in the dimer correlation $\chi$ decays as $\mathcal{O}(1/L_c^2)$ for PCPT and $\mathcal{O}(1/L_c)$ for CPT.
Load-bearing premise
The load-bearing premise is that the artificial momentum transfer created by making clusters periodic is unphysical and can be switched off, and that the periodized inter-cluster hopping then decays as $1/L_c$; both assumptions are validated only on one exactly solvable one-dimensional large-$N$ model.
Editorial extensions
If this is right
- PCPT reaches a fixed accuracy with substantially smaller clusters than CPT, because its error is $\mathcal{O}(1/L_c^2)$ instead of $\mathcal{O}(1/L_c)$.
- Because the lattice Green function is evaluated directly in reciprocal space, PCPT avoids inverting cluster Green functions in real space, cutting computation time for large clusters.
- The derivation gives DCA a real-space microscopic definition as CDMFT applied to the periodized Hamiltonian, so DCA can be formulated within the self-energy functional approach just like CDMFT.
- PCPT is exact in the limits $L_c\to\infty$, $U/t=0$, and $t/U=0$, and the claimed $1/L_c^2$ convergence of the coarse-grained Green function is stated to hold in any dimension, not only in the one-dimensional benchmark.
Reading between the lines
- The constraint (16) is doing the full work of restoring equivalence; if physical umklapp processes are present in a real material, setting this transfer to zero may discard genuine physics, and the paper's single one-dimensional benchmark is a narrow test of that danger.
- The convergence argument relies on the periodized inter-cluster hopping decaying as $1/L_c$; a natural stress test is a two-dimensional Hubbard model at intermediate coupling with a numerically exact cluster solver, checking whether the $1/L_c^2$ scaling survives away from the large-$N$ limit.
- Because PCPT is formulated in momentum space and avoids matrix inversions, it is a natural candidate to pair with continuous-time quantum Monte Carlo cluster solvers, which also work naturally in Matsubara frequencies.
- Reading DCA as CDMFT of a model with long-range periodized hoppings suggests that DCA's coarse graining is equivalent to modifying the lattice Hamiltonian rather than merely discretizing the Brillouin zone; that shift in interpretation could affect how DCA results are compared with experiment.
Original abstract
Quantum periodic cluster methods for strongly correlated electron systems are reformulated and developed. The reformulation and development are based on a canonical transformation which periodizes the fermions in the cluster space. The dynamical cluster approximation and the cellular dynamical mean field theory are related each other through the canonical transformation. A cluster perturbation theory with periodic boundary conditions is developed. It is found that the periodic cluster perturbation theory converges rapidly with corrections $\mathcal{O}(1/L_c^2)$, where $L_c$ is the linear size of the clusters, whereas the ordinary cluster perturbation theory converges with corrections $\mathcal{O}(1/L_c)$.
The compiled record
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
The abstract, sentence by sentence
Quantum periodic cluster methods for strongly correlated electron systems are reformulated and developed. The reformulation and development are based on a canonical transformation which periodizes the fermions in the cluster space. The dynamical cluster approximation and the cellular dynamical mean field theory are related each other through the canonical transformation. A cluster perturbation theory with periodic boundary conditions is developed. It is found that the periodic cluster perturbation theory converges rapidly with corrections $\mathcal{O}(1/L_c^2)$, where $L_c$ is the linear size of the clusters, whereas the ordinary cluster perturbation theory converges with corrections $\mathcal{O}(1/L_c)$.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Quantum periodic cluster methods for strongly correlated electron systems are reformulated and developed. | contribution 1.00 | 0.89 | 0.09 | asserted 1.00 |
| s1 | The reformulation and development are based on a canonical transformation which periodizes the fermions in the cluster space. | method 0.98 | 0.85 | 0.06 | asserted 1.00 |
| s2 | The dynamical cluster approximation and the cellular dynamical mean field theory are related each other through the canonical transformation. | contribution 0.59 | 0.76 | 0.12 | asserted 1.00 |
| s3 | A cluster perturbation theory with periodic boundary conditions is developed. | contribution 0.89 | 0.93 | 0.09 | asserted 1.00 |
| s4 | It is found that the periodic cluster perturbation theory converges rapidly with corrections $\mathcal{O}(1/L_c^2)$, where $L_c$ is the linear size of the clusters, whereas the ordinary cluster perturbation theory converges with corrections $\mathcal{O}(1/L_c)$. | result 1.00 | 0.92 | 0.63 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v2
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
Figures
Showing 3 of the paper's 4 figures. See all figures in the PDF ↗ Figures reproduced from arXiv:cond-mat/0602615 by the authors.
References
Discussion
Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.