14.09.2014 Views

pdf, 9 MiB - Infoscience - EPFL

pdf, 9 MiB - Infoscience - EPFL

pdf, 9 MiB - Infoscience - EPFL

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

4.4. RESULTS AND DISCUSSION 91<br />

where<br />

ξ(k) =t ( e i(α 2−α 1 ) + e iα 2<br />

+1 ) ,<br />

B(k) =∆/2 ( e −iφ 3<br />

cos(α 2 − α 1 )+e −iφ 2<br />

cos α 2 + e ) −iφ 1<br />

.<br />

α 1 and α 2 are the projections on the reciprocal vectors: k = α 1 /2π b 1 +α 2 /2π b 2 .<br />

We emphasize that ∆ is not the superconducting 〈 〉 〈 order parameter 〉 〈 which 〉 is<br />

actually given in our approximation by c † i↑ c† j↓<br />

= b i f † i↑ b jf † j↓<br />

≃〈b i b j 〉 f † i↑ f † j↓<br />

.<br />

〈 〉<br />

At zero temperature we have c † i↑ c† j↓<br />

≃ δ∆.<br />

4.4 Results and discussion<br />

4.4.1 VMC approach<br />

The VMC approach allows the exact evaluation (up to statistical error bars) of<br />

the energy expectation value of the Gutzwiller projected trial functions defined<br />

previously. The goal is now to find the lowest energy state at each doping level.<br />

To compare the energy expectation values of the different states we define the<br />

condensation energy e c<br />

e c = e var − e FS , (4.13)<br />

which is the energy gain per site with respect to the FS state, the Gutzwiller<br />

projected Fermi sea. The results for the 72 and the 144 sites cluster are shown<br />

in figures 4.3 and 4.4. We have checked that our variational subspace is large<br />

enough by minimizing the trial function (4.1) on the 72 sites cluster with respect<br />

to all parameters of the Hamiltonian (1.23) simultaneously, not allowing however<br />

the breaking of translational symmetry. By comparing the results for the two<br />

clusters we see that our results are rebust with respect to size effects.<br />

For all filling factors we have found that the isotropic hopping term t ij ≡ 1<br />

and θ ij ≡ 0 always give the minimal energy. Moreover for the symmetry of the<br />

RVB state we found d x 2 −y 2+id xy symmetry in the minimal energy configuration at<br />

all dopings, i.e. the phases in equation (4.2) are given by φ 1 =2π/3, φ 2 =4π/3,<br />

and φ 3 = 0. Of course one can interchange these phases and also add or subtract<br />

2π to each of them and still the same energy.<br />

In the reminder of this section we will describe in detail the different phases<br />

appearing in the honeycomb lattice from our VMC simulations.<br />

4.4.2 Half-filling<br />

At half-filling we found that the optimal energy wavefunction is the RVB/AF<br />

mixed state. This state has a considerable fraction (66%) of the classical Néel<br />

magnetism. However, our variational ansatz seems to overestimate the tendency

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!