Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Topics - Anders Blom

Pages: [1] 2 3 ... 5
1
News and Announcements / Bugfix update 2016.2
« on: September 26, 2016, 16:20 »
Small updates:
* The forces are now again printed for each step in an optimization run
* Slight improvements in the installer on Mac

2
News and Announcements / VNL-ATK 2015 released
« on: October 21, 2015, 12:18 »
The long-awaited VNL-ATK 2015 is now officially released!

The new version contains many exciting new features, like electron-phonon interaction, a new Job Manager, and a long range of analysis functions for molecular dynamics simulations, just to mention a few. See the full release letter for all details!

NOTE
If you are currently running version 2015.rc2, it's essential that you update to the official release, as several bugs were found recently that make 2015.rc2 perform less well and may give rise to a poor user experience, esp. in the Job Manager for instance.

ANNOUNCEMENT
Virtual NanoLab (but not ATK) is now available completely free of charge to academic users! To request your free license, and for more details about what is included, click here.

Highlighted new features in ATK-VNL 2015
  • Electron-phonon interaction - calculate inelastic current, deformation potentials, mobility
  • New Job Manager - submit parallel and serial jobs locally and on remote clusters
  • New molecular dynamics (MD) and ion dynamics functionality - many new analysis functions, potentials, performance upgrades
  • New methodology developed for optimizing interfaces and devices, based on RigidBodies + a 1D minimization of the central region length
  • Global minimization algorithm (for crystal structure or phase stability predictions)
  • NEB parallelized over images
  • Adaptive Kinetic Monte Carlo (KMC) module in ATK for finding reaction pathways, evaluating reaction rate pre-factors (HTST), and coarse-grained time-accelerated MD
  • Virtual Crystal Approximation (VCA) for DFT
  • Effective band structure analysis (zone unfolding) for supercells of e.g. random alloys
  • Ability to use different "c" parameters for different materials in the same structure with MGGA-TB09
  • Many parallelization and performance improvements (speed and memory) in the computational engine, in particular to better employ sparsity of matrices and allow the code to scale to a larger number of MPI processes
  • VNL can now easily handle 1M+ atoms both in the Builder and Viewer
  • Python interface to FHI-aims, which is also shipped with ATK
  • LAMMPS: import/export geometries + analysis of MD trajectories in the MD Analyzer and Movie Tool
  • Quantum ESPRESSO: import/export filters + analysis of electron density, band structure, DOS, etc
  • Extended VASP input/output support, and better defaults in the scripter
  • OpenBabel shipped with VNL

+ way more cool new features than fit in a short summary - you really have to check out the full release letter!

We also recommend everyone to have a look at the updated tutorials page, where many tutorials have been added to present the new features in the 2015 release.

Download VNL-ATK 2015.0 here

License
  • To run this version your license file must contain features for version 15.0 or higher.
  • Academic users may obtain a VNL license for free.
  • In order to run calculations using ATKClassical and to use the Builder modules, you must have an ATKClassical feature in your license file. This license is included in the free VNL license, and is provided free of charge, upon request, for all academic customers. It can of course also be purchased as part of a commercial license.
  • To use the FHI-aims functionality, you need a separate license feature, which can be purchased for commercial and academic (at a discount) use.
  • The VASP Scripter and analysis features require the ATKVASP license feature, which is included for free for all customers that purchase an ATK license. For users who only have a VNL license, the ATKVASP feature must be purchased separately, at €500/$600.

3
For a complete list of the fixes in 2014.3, see the release notes.

Main changes
  • A large number of fixes for spin-orbit, spin-polarized and non-collinear models
  • BSSE and Grimme now work together (so that you can run the tutorial on this topic properly)
  • nlsave/nlread MGGA-TB09 calculations now saves and restores the "c" parameter properly
  • A few updates in ATKClassical/TremoloX
  • + a few other minor have been fixed

Download VNL-ATK 2014.3 here!

For additional information about ATK 2014, see the original release letter.

4
Plugin Development / Make more bonds
« on: June 23, 2015, 22:29 »
VNL will by default only display bonds between atoms which are within a particular distance from each other, quite specifically their joint covalent radius +10% or closer. This means that if you stretch a structure a bit, the bonds may disappear. Now, since ATK 2015 and earlier (things will change in version 2016, when we add bonded force fields) doesn't use the bonds for anything in the calculations (only the actual distance between atoms are used), this has no influence on any computed results - only your pictures.

But, sometimes it's nice to see bonds anyway. So, I made a plugin that lets you adjust the "fuzz factor", i.e. the 10% mentioned above. So now you can set it to 20% if you want also atoms that are farther apart to appear bonded, or a lower value if you have too many bonds (it can happen also). The plugin works both in the Viewer and Builder and requires ATK 2014 or newer.

To install it, open the AddOn Manager, click Local install to install the zip file, then restart VNL. The new tool will appear as "Bond control" in the right-hand side plugin panel (under "Select" in the Builder). The value you can specify is 1+the fuzz factor (default 0.10).

See screenshots! The molecule obviously doesn't need a fuzz factor of 1, but illustrates the concept. The Tausonite crystal (SrTiO3 cubic, taken from the built-in database) looks a lot better with fuzz factor 0.04 (try the default value 1.1 in the plugin and you will see).

5
For a complete list of the fixes in 2014.2, see the release notes.

Main changes
  • Updates to the VASP plugins, to support more flavors of DOSCAR files, and for MD Trajectories
  • A few serious bugs, which however you were rather unlikely to encounter, have been fixed
  • Several minor bug fixes for things that were more annoying than problematic
  • Some updated features

Download VNL-ATK 2014.2 here!

For additional information about ATK 2014, see the original release letter.

6
News and Announcements / Bug in FFT2D in ATK 2014
« on: October 21, 2014, 15:24 »
An error which appears in some rare cases for the FFT2D solver has been detected, which results in the potential going to infinity, causing the calculation to crash or gives nan/inf values. These should be easy to spot in the log file.

This appears to actually be bug in the current Intel Math Kernel Library (MKL) which we use for all matrix operations. It manifests itself when your grid is 128x128, 256x256 or 512x512, in which case it will give completely wrong results.

This has already been posted by someone else on the Intel forum (https://software.intel.com/en-us/forums/topic/509936) and we are waiting for Intel to provide a fix.

The workaround for now is to increase the density grid mesh cut-off slightly - just a Hartree or two will usually do the trick!

7
News and Announcements / Addons updated for VNL 2014
« on: October 21, 2014, 13:19 »
Two addons have been updated for VNL 2014.

  • The package called "OldAnalysisReplacement" - the ability to do PDOS analysis for bulk systems was gone in 2014.1, it's now back properly
  • The VASP plugin has been updated to include import of XDATCAR (for MD trajectories) and OUTCAR

Download the updates via the AddOn Manager in VNL, and remember to check regularly for new updates.

8
News and Announcements / ATK 2014 released
« on: October 12, 2014, 20:57 »
It is my great pleasure to announce that VNL-ATK 2014.1 is now available for download.

NOTE: If you already installed 2014.0 (or any of the preview releases called a1,b2,rc1, etc), it's important to update to 2014.1 since several bugs were fixed.

The new release introduces a wide variety of new and improved features. For more details, see the complete release letter. The links below go to specific tutorials on some of the new topics.

  • A new module for empirical potentials, called ATKClassical, along with several updates to the molecular dynamics framework and integration with the ATK phonon transport capabilities.
  • Also support for plotting LAMMPS output!
  • Noncollinear spin upgraded and spin-orbit interaction added, both for ATK-SemiEmpirical and ATK-DFT.
  • New capabilities especially aimed at semiconductor materials (like InAs), as well as metal-semiconductor interfaces.
  • Several performance improvements, to reduce both calculation time and memory consumption, and to improve the graphical capacity of VNL. The new code is 35-40% faster than ATK 13.8, and even more when using some of the new special algorithms for particular simulations. The new graphics engine can build 10,000+ atom systems and visualize results like MD trajectories with 1M+ atoms.
  • Integration with VASP
  • Also some support for Quantum Espresso, GPAW, LAMMPS, Materials Studio added
  • Mac OS X version
  • And many, many other nice things – see the complete release letter for additional details!

9
News and Announcements / ATK 13.8.1 update released
« on: February 19, 2014, 09:31 »
An updated version of ATK 13.8 is available for download. A few somewhat serious bugs have been fixed, in addition to several minor ones, and there are also some new features!

See the full details at http://quantumwise.com/corporate/news-archive/246.

10
We have found out that OpenMP threading interferes with the MPI parallelization of ATK in a highly unexpected way. By using the correct OpenMP settings, instead of the default ones, it is possible to improve the parallel performance of ATK up to 10x (provided you already have enough slave licenses) without upgrading your hardware or buying more licenses.

For years, the conventional wisdom has been that ATK performs best in parallel when running 1 MPI process/socket. This feels wasteful on 4 or 8 core sockets, but it just wasn't possible to get any real speedup if we put more MPIs on the socket. Logically, we figured this was due to communications overhead on the L1/L2 cache or RAM buses, or even the Ethernet ports.

It turns out, however, that by setting

export OMP_NUM_THREADS=1
export OMP_DYNAMIC=FALSE


you can get fantastic MPI speedup in ATK!

We are working on benchmarks, but we have already seen a 10x speedup on a heavy device calculation (for the transmission spectrum calculation). Note that this is not a speedup for serial vs parallel - it's the same MPI setup in both cases (1 node with 32 cores, running 32 MPI processes), but with threading on/off.

As another example, a (smallish) bulk calculation in MGGA which takes 1000 s in serial can be improved 10x by using 16 MPIs on just 2 nodes. With threading turned on, the best we could get on 2 nodes was 2x speedup with 2 MPIs - if we put more MPIs, the calculation slows down! In fact, with threading on, using 8 MPIs on 1 node is 50% slower than running in serial, whereas if threading is OFF we get 8x speedup!

There are still serial overheads in ATK, and in fact they start to show now in a way we never saw before, because we could never get the speedup to get the runtime of the parallelized parts down low enough. The most noticeable case for this is the Multigrid method, if you run devices with gates or charged systems (these are, in fact, the only cases for which you should use Multigrid, which also in serial is a lot slower than FFT2D). We will put some focus on addressing these parts of the code for the next release.

Also, there are still cases where keeping threading ON makes sense, viz. if you are running a calculation which ATK cannot parallelize over - the clearest example would be a large supercell with 1x1x1 k-points (or molecules). In those cases one could run it on a single node, without using MPI, and get some benefit of OpenMP threading by leaving OMP_DYNAMIC=TRUE and by not setting OMP_NUM_THREADS (or set it to the number of cores).

Hopefully this new insight means that our customers can get results from ATK faster, both by the shorter calculation times and by being able to run more concurrent calculations without tying up all the nodes on the cluster (if you have more than 1 master license, that is). For 10x speedup, maybe a calculation just needs 2-3 nodes now, where it before needed 10-20 full nodes.

If you have further questions on this, don't hesitate to ask us!

11
We would like to notify all friends of ATK, Virtual NanoLab and QuantumWise about our upcoming workshop

Modeling of Electronic Devices and Materials at the Nanoscale

The event takes place in Copenhagen 1-3 July and includes:

  • 2 days conference program with talks by state-of-the-art modeling experts. See list of invited speakers below!
  • 1 day Atomistix ToolKit exclusive hands-on tutorial at QuantumWise headquarters + a chance get advice from the developers behind the code.
  • Guided tour to Niels Bohr Institute in Copenhagen. Hear about the famous physicist and see his office.

Workshop website

List of confirmed speakers
  • Asen Asenov (Gold Standard Simulations)
  • Jerry Bernholc (North Carolina State University)
  • Anders Blom (QuantumWise)
  • Volker Blum (Duke University)
  • Mads Brandbyge (DTU Nanotech)
  • Fabian Bufler (Synopsys)
  • Jan Hamaekers (Fraunhofer SCAI)
  • Hannes Jonsson (University of Iceland)
  • Gerhard Klimeck (University of Purdue)
  • Mathieu Luisier (ETH Zürich)
  • Troels Markussen (QuantumWise)
  • Branislav Nikolic (University of Delaware)
  • Geoffroy Pourtois (IMEC)
  • Gian-Marco Rignanese (U. Catholique de Louvain)
  • Alexander Schluger (University College London)
  • Kurt Stokbro (QuantumWise)
  • Tejs Vegge (DTU)

Call for abstracts
Send your abstract before 1st of April if you wish to give a contributed talk or want to present a poster at the poster session.

Related events

ATK Workshop at ACS Spring Meeting in Dallas, Texas, USA

"Atomic-scale Modelling and Electron Transport in Nanostructures" -more info

The workshop is free for ACS Spring Meeting Attendees

Where: Dallas Convention Center, Room D220
When: Wednesday, March 19, 12:30 PM - 3:00 PM

12
It has come to our attention, via the article Phys. Chem. Chem. Phys. 4, 4078 (2002) that the structures we have used in the Database in VNL for MoS2 and MoSe2 are not correct. This is, in fact, an error that has propagated into many standard references, such as Wyckoff's famous book which is what we used when generating the database.

To fix the problem manually, the correct structures should be

Code: python [Select]
# Set up lattice
lattice = Hexagonal(3.1604*Angstrom, 12.295*Angstrom)

# Define elements
elements = [Molybdenum, Molybdenum, Sulfur, Sulfur, Sulfur, Sulfur]

# Define coordinates
fractional_coordinates = [[ 0.33333333,  0.66666667,  0.25      ],
                          [ 0.66666667,  0.33333333,  0.75      ],
                          [ 0.33333333,  0.66666667,  0.621     ],
                          [ 0.66666667,  0.33333333,  0.121     ],
                          [ 0.33333333,  0.66666667,  0.879     ],
                          [ 0.66666667,  0.33333333,  0.379     ]]

# Set up configuration
bulk_configuration = BulkConfiguration(
    bravais_lattice=lattice,
    elements=elements,
    fractional_coordinates=fractional_coordinates
    )


and

Code: python [Select]
# Set up lattice
lattice = Hexagonal(3.288*Angstrom, 12.9*Angstrom)

# Define elements
elements = [Molybdenum, Molybdenum, Selenium, Selenium, Selenium, Selenium]

# Define coordinates
fractional_coordinates = [[ 0.33333333,  0.66666667,  0.25      ],
                          [ 0.66666667,  0.33333333,  0.75      ],
                          [ 0.33333333,  0.66666667,  0.621     ],
                          [ 0.66666667,  0.33333333,  0.121     ],
                          [ 0.33333333,  0.66666667,  0.879     ],
                          [ 0.66666667,  0.33333333,  0.379     ]]

# Set up configuration
bulk_configuration = BulkConfiguration(
    bravais_lattice=lattice,
    elements=elements,
    fractional_coordinates=fractional_coordinates
    )


The error will be fixed for ATK 13.8, and it should also be noted that if you run a geometry optimization (as you always should!) the mistake corrects itself. However, if you fail to do this, strange things will happen, and for instance the band gap of a MoS2 monolayer will be very wrong.

13
To install ATK on Windows 8, it is necessary to install it in compatibility mode.

See http://www.dummies.com/how-to/content/how-to-run-older-programs-under-windows-8.html for a detailed description of how to run a program in compatibility mode. Note carefully that it's the installer, not ATK itself, that should be run in compatibility mode.

14
Plugin Development / Passivator for III-V structures
« on: April 30, 2013, 16:18 »
There is a new Builder AddOn, for passivating III-V structures like the model of an 8.3 nm wide InAs FinFET (just the actual fin, not the whole FET) shown in the picture. Use like the normal passivator tool, which is designed for group IV elements, i.e. select the atoms to passivate and click the icon - simple!

To download the III-V passivator addon, use the AddOn Manager (in the Help menu) in VNL 12.8.

15
News and Announcements / ATK 12.8.2 bug fix release
« on: February 27, 2013, 21:32 »
A bug related to the license system has been discovered in ATK 12.8.1. All customers using ATK 12.8.x should upgrade to ATK 12.8.2.

You can obtain the updated 12.8.2 packages for Linux from the download page.

For more details, see the release notes.

Pages: [1] 2 3 ... 5