Mailman 3 October 2015 - Kwant-discuss

Tight binding model on a quasicrystal
by Ion Cosma Fulga Oct. 25, 2015

Oct. 25, 2015

Hi, For our last project, we used Kwant to study a tight binding model defined on an aperiodic tiling. To build the tiling, we used a code snippet written by Michael Wimmer, which allows to add sites at any position, without needing a kwant.lattice. Michael's code can be found at: https://mailman-mail5.webfaction.com/pipermail/kwant-discuss/2014-July/0001… Our own code is available at: http://arxiv.org/src/1510.06035v1/anc/quasicrystal.py Best, Cosma

1 0

Periodic BC in kwant
by R Mathew Oct. 23, 2015

Oct. 23, 2015

Hi, I am trying to simulate a finite size quantum hall strip with periodic boundary in one direction. I am using a modified code from Edx course on Topology in Condensed Matter. The system hangs when sys.finalized is called. Is it possible to have periodic system finite system in kwant? I don't want to resort to TranslationalSymmetry in the periodic direction. Here is the code snippet. # Building system lat = kwant.lattice.square() sys = kwant.Builder() sys[(lat(x, y) for x in range(L) for y in range(W))] = 4*par.t-par.mu def hopx(y, par): return -par.t*np.exp(-1j * par.B * y) for i in xrange(L): for j in xrange(W): if j > 0: sys[lat(i, j), lat(i, j - 1)] = -par.t # Hopping in x-direction if i > 0: sys[lat(i, j), lat(i - 1, j)] = hopx(j, par) #Making the system periodic for j in range(W): sys[lat(L-1, j), lat(0, j)] = hopx(j, par)*np.exp(1j * par.phi) Thanks

3 3

kkghosh
by kamal ghosh Oct. 22, 2015

Oct. 22, 2015

Dear all In the following program related to quantum wire I find a runtime error even though I get the system structure o.k. Please help me. The file is attached herewith. Thanks. K.K.Ghosh

1 0

Kwant 1.1 has been released
by Christoph Groth Oct. 22, 2015

Oct. 22, 2015

Hello, We are happy to announce the release of Kwant 1.1. The user-visible changes are: • Harmonize Bands with modes • New option add_cells of attach_lead • New method conductance_matrix • Deduction of transmission probabilities • Clearer error messages • New option pos_transform of map See [1] for details. There is nothing spectacular here, but particularly the improved error messages should make work with Kwant more pleasant. There was also substantial work “under the hood” that will allow us to release new versions with much less effort in the future. Full backward compatibility is the aim for all Kwant releases that share the same major version. Thus, all Kwant 1.0 scripts should continue to work with Kwant 1.1. Kwant 1.1 packages are available right now for Debian, Ubuntu, and Windows. They should be available soon for Mac OS & Arch Linux. Please let us know if you experience any problems. The plan is that 1.1 will be the last Kwant release for Python 2 and we switch development to Python 3. There should be soon a Kwant 1.2 that will be identical to 1.1 except that it will use Python 3. (Do not worry if you have to keep using Python 2: we plan to maintain Kwant 1.1 for several years.) Thanks to everyone who contributed to this release! Happy kwanting, Christoph on behalf of the Kwant team [1] http://kwant-project.org/doc/1.1/pre/whatsnew/1.1

2 3

APS March meeting 2016 tutorial: Introduction to computational quantum nanoelectronics
by Christoph Groth Oct. 22, 2015

Oct. 22, 2015

Dear all, This is an announcement for an APS March meeting 2016 tutorial that will feature Kwant. It can be also found online: http://www.aps.org/meetings/march/events/tutorials.cfm#t10 Please spread the word to anyone who might be interested. Christoph Tutorial #10: Introduction to Computational Quantum Nanoelectronics Quantum nanoelectronics deals with the physics of small (< 1 μm) and/or cold (down to ~10 mK) objects connected to the macroscopic world through electrodes or gates. A central question at the core of this field is how quantum effects can be observed and manipulated through the macroscopic measuring apparatus. In this tutorial, we will give a pedagogical introduction to the field. We will start with an introduction to the main theoretical concepts and a review of some seminal experiments. The central part of the tutorial will be devoted to practical training on numerical calculations: we will demonstrate how researchers can simply setup their own models and perform their own calculations. These calculations can be used for theoretical predictions, to explain experimental data or even to assist the conception of device design. The numerical part of the lecture is based on the Python programming language and the Kwant package. No particular background in programming is needed. Topics • Scattering theory of transport, Landauer formula for the conductance • Continuous and discrete models (effective mass, Dirac equation, fermion doubling theorem) • Electronic interference effects (Aharonov-Bohm effect, universal conductance fluctuations) • Topological matter, quantum Hall effect • Time-resolved nanoelectronics • Numerical calculations with Python • Hands-on tutorial on Kwant • Practical example of the modeling of an experiment When? Sunday, March 13, 2016 1:30 p.m. - 5:30 p.m. Who Should Attend? PhDs, postdocs, and faculty (both experimentalists and theorists) interested in calculating the transport properties of quantum nano-systems (nanowires, semiconducting heterostructures, graphene, topological insulators…). We particularly encourage participation of experimentalists wanting to develop their own modeling of their experiments. Basic knowledge of quantum mechanics, statistical physics and condensed matter would help to fully benefit from the tutorial. Organizer Christoph Groth, CEA Grenoble, France Instructors • Anton Akhmerov, TU Delft, Netherlands • Christoph Groth, CEA Grenoble, France • Xavier Waintal, CEA Grenoble, France • Michael Wimmer, TU Delft, Netherlands

1 0

Kwant 1.1 has been released
by Christoph Groth Oct. 22, 2015

Oct. 22, 2015

Hello, We are happy to announce the release of Kwant 1.1. The user-visible changes are: • Harmonize Bands with modes • New option add_cells of attach_lead • New method conductance_matrix • Deduction of transmission probabilities • Clearer error messages • New option pos_transform of map See [1] for details. There is nothing spectacular here, but particularly the improved error messages should make work with Kwant more pleasant. There was also substantial work “under the hood” that will allow us to release new versions with much less effort in the future. Full backward compatibility is the aim for all Kwant releases that share the same major version. Thus, all Kwant 1.0 scripts should continue to work with Kwant 1.1. Kwant 1.1 packages are available right now for Debian, Ubuntu, and Windows. They should be available soon for Mac OS & Arch Linux. Please let us know if you experience any problems. The plan is that 1.1 will be the last Kwant release for Python 2 and we switch development to Python 3. There should be soon a Kwant 1.2 that will be identical to 1.1 except that it will use Python 3. (Do not worry if you have to keep using Python 2: we plan to maintain Kwant 1.1 for several years.) Thanks to everyone who contributed to this release! Happy kwanting, Christoph on behalf of the Kwant team [1] http://kwant-project.org/doc/1.1/pre/whatsnew/1.1

1 0

Transmission and propagating modes
by Lee Kwok-Long Oct. 22, 2015

Oct. 22, 2015

Dear authors, I calculate the conductance of graphene nanoribbon by summing over all the propagating modes like: smatrix = kwant.smatrix(sys, energy) S10=smatrix.submatrix(1,0) n_modes = [len(li.momenta) for li in smatrix.lead_info] Gc=0. for nn in xrange(n_modes[1]//2): Tmode=real(S10[nn,nn]*conj(S10[nn,nn])) Gc=Gc+Tmode I compared the conductance Gc with the results calculated from "smatrix.transmission(1, 0)", and I found that they are different when we have defects (or some atoms are removed) in the scattering region. Why this happens? My code is attached to this email. Thanks in advance! Kwoklong Lee

2 3

Automatic parallelization?
by David Abergel Oct. 18, 2015

Oct. 18, 2015

Dear all, I'm 99% sure that this is a generic python issue, rather than a specific KWANT one, but perhaps you'll permit me to ask it here because I imagine it's something that crops up for other KWANT users as well. I have a program which uses KWANT to compute the conductance through a system. There are no explicit references to any parallelization commands in any of the code that I have written. When I run this program on my laptop (which has a processor with 4 cores), top tells me that the CPU utilization is consistently around 380%. When I run exactly the same code on our cluster (which has 16 cores per node) the CPU utilization is consistently around 100%. The specific program is essentially a big for loop which computes conductance at different energies, so there is large scope for parallelization there. It seems that my laptop is spotting this automatically whereas the cluseter is not. Does anyone know what is going on? Can I control this using environment variables or some other mechanism? Any help gratefully received! Here are the relevant version numbers of python and KWANT, as listed by apt-cache: Cluster: python 2.7.3-0ubuntu2.2 python-kwant 1.0.0-1~saucy Laptop: python 2.7.9-1 python-kwant 1.0.5-1~vivid Thanks! David -- David Abergel Nordita david.abergel(a)nordita.org @David_Abergel

2 2

Spin resolved quantum Hall
by Hamzah Fauzi Oct. 16, 2015

Oct. 16, 2015

Hello, I am new to kwant and I can successfully run kwant on my own. I would like to know whether it is possible to incorporate electron spin (or even localized nuclear spin) into the quantum Hall effect code as in example 6 kwant paper ( http://downloads.kwant-project.org/doc/kwant-paper.pdf). Thank you, Fauzi

2 2

Distinguish contribution from different valleys
by Michael Beconcini Oct. 15, 2015

Oct. 15, 2015

Hello everyone, I’m studying valley currents in graphene monolayers and I would like to distinguish the contribution to the transmission matrix elements and to the wavefunctions from different valleys. Until now the only idea I have had is inspired by a work made by Beenakker et al. in 2007 where, in a graphene ribbon with only zig-zag edges, they attribute the modes with longitudinal wavevector in (pi,2pi) to the first valley, whereas the modes with longitudinal wavevector in (0,pi) to the second valley. To apply this method I would need to know how kwant orders the elements of the scattering matrix. Anyway, the idea that I have exposed only partly solves my problem because the system I have in mind has both zig-zag and armchair edges. Someone could suggest a way to do that? Thanks for the attention! Best, Michael

2 2