2015-01-29T15:17:42Z

Zzalsmt2:

=Summary list of Survey Executive Summaries:=

==For the Global CCP survey:==

*Three packages are the most-used packages by 26% of respondents. Conversely, another 31 packages are used by one or two users and account for a further 26% of respondents.
*Producing publication quality plots is the most-used technique.
*However, the features making these packages the favourites are:
#Software that is written specifically for their domain of interest.
#Large datasets are handled efficiently.
#Scripting or other ability to extend the tool is required.
*Users second most favoured packages are general purpose visualisation tools.
*Users were given five options for selecting their most required development. None emerged as being more needed than the others.
*Conversely, large amounts of memory was clearly the most important requirement for high performance visualisation.
*The main future challenges are suggested to be
#The ability to handle large amounts of data
#The ability to operate in a distributed environment.

==For the CCPi survey:==
The following conclusions were drawn:
*The survey was focused on what changes should be made to the set of supported tools? ImageJ, Avizo and ParaView products were favoured. A previous popular tool for development, VolView, proved to be unpopular with users and therefore was dropped.
*The type of software development supported should focus primarily on the areas where the network already has skills that can be leveraged (e.g. reconstruction) with new partners being sought to fill gaps (e.g. quantification and segmentation) for the future. Training was shown to be a requirement at different levels with outreach and metadata to be a lower priority.
*Over the following six months changes have happened that have resulted in the following actions:
#Quantification visualisation tools are now being developed with API links for ImageJ, Avizo and ParaView; the most popular specific tools.
#Outreach events are continuing but at a lower effort level.
#Software show-and-tell events along with related seminars now aim to focus on guidance and use of the products as well as the research solution that is being solved. These are now becoming monthly events and also aim to keep a user and developer community connected.
#Increase in cross training sessions: there is a collaboration now across Diamond Light Source / RCaH / University of Manchester and SCD/STFC labs with links to the Harwell Imaging Partnership and ISIS (Neutron Spallation Source). Four annual training events are being held at DLS/RCaH and a further four annual training events at UoM with specialist events being held at SCD/STFC in RAL - all of these allow for cross-attendance throughout the CCPi network.
#The survey allowed for representation at celebration events; including the Queen's Anniversary Prize at Manchester and the ToScA exhibition space where we sponsor poster prizes.
#Quarterly Developers' Workshop days are now held in CCP remote institutions and focus on areas that the survey stated there were needs.

==For the CFD survey:==
The survey of tools and techniques used within project gave the following main conclusions:
#Visualisation is most important in analysis; but also useful in problem definition, and mesh generation.
#ParaView is high on the list of preferred visualisation tools. But gnuplot, Matlab, xmgrace, and Pgplot are also required.
There are strong preferences to version number as features change. This may unify in the future.

=Future Roadmap=

There have been some requests for further work and also some groups delayed implementation. Currently there are four actions:

1. '''Two sub-surveys requested:''' We surveyed both the old non-funded CCPs which included

#[http://www.ccp.ac.uk/ CCP1] The electronic structure of molecules (defunct link)
#[http://www.ccp.ac.uk/ CCP3] Computational studies of surfaces (defunct link)
#[http://www.ccp.ac.uk/ CCP6] Molecular quantum dynamics
#[http://www.ccp.ac.uk/ CCP13] Software for fibre and polymer diffraction
#[http://www.ccp14.ac.uk/ CCP14] Powder diffraction

and surveyed the national facilities lists:

# [http://www.diamond.ac.uk/Home.html STFC: Diamond Light Source]
# [http://www.isis.stfc.ac.uk/ STFC: ISIS - Neutron Spallation Source]
# [http://www.stfc.ac.uk/clf/default.aspx STFC: CLF - Central Lasers Facility]

These links gathered 37 extra responses that were merged together with the total (6 from non-funded CCPs and 31 from the user communities). We have been asked to look specifically at the CLF and the ISIS user communities separately in a short extra report. From informal comments we believe users from non-funded CCPs have transferred to a funded CCP, a facilities user list, or moved on.

2. '''Rerunning the global survey:''' A further survey is being planned for Summer 2015, which will then track temporal changes between the CCPs and also indicate the changing nature from the new CCPs just recently funded.

3. '''Space Applications Catapult Audit:''' One part of the Optional Appendices was a page on the TSB Space Application Catapult Centre - just recently rebranded to becoming a UK Innovation site. They have recently built two new visualisation nodes and installed different and distinct software links to be reevaluated. This has come about due to change of use from a specific type of data show-and-tell space, to a command-and-control center with different visualisation needs. An associate is assisting in this move and the transition and thus as there is a new set of different users an audit is proposed in Spring 2015. We have requested this list to then compare against the Global CCP survey.

4. '''Global Comparison:''' We are in email contact with similar structures and have evaluated through the CCPi in particular specific tools from the Australian National University and Curtin University. Comparative study with audits are underway with these and at the German build IBM SuperMUC.

We have been asked to look at future user communities:

* CCP5 and CCP_EM as they embark on tomographic reconstruction and visualisation of 3D volumes

* CCP-NC involvement in remote visualisation and web based visulation with Jmol for example. http://www.ccpnc.ac.uk/magresview/magresview/magres_view.html?JS

* CCP_PET/MR a new recently funded CCP that has specific 4D visulaisation needs including uncertainty visualisation.

We are also open to studies and sub-studies for other groups related to the CCP programme.

2015-01-28T16:05:55Z

Zzalsmt2:

Over 2014 the visualisation group, part of the Technology Division within SCD, has reconsidered the real visualisation needs of the computational sciences community. A series of informal and formal surveys are underway and the first ones have tackled the Tomographic Imaging and CFD (Computational Fluid Dynamics) communities. These wanted to find out which tools were actually being used and the best methods to support them.

== Executive Summary ==

Key was a difference between choice of primary and secondary package:

* Three packages are the most-used packages by 26% of respondents.
* Conversely, another 31 packages are used by one or two users and account for a further 26% of respondents.

Producing publication quality plots is the most-used technique. However, the features making these packages the favourites are:

# Software that is written specifically for their domain of interest.
# Large datasets are handled efficiently.
# Scripting or other ability to extend the tool is required.

Users second most favoured packages are general purpose visualisation tools.

Users were given five options for selecting their most required development. None emerged as being more needed than the others.

Conversely, large amounts of memory was clearly the most important requirement for high performance visualisation.

The main future challenges are suggested to be

# The ability to handle large amounts of data
# The ability to operate in a distributed environment.

== Survey Presentation ==

Surprisingly, open source was not always the most important issue but the easy creation of plug-ins, new readers and writers, as well as analysis tools have been requested.
There was also indicated a strong growth in the use of the ParaView visualisation system (http://www.paraview.org/) that is an open source, multi-platform data analysis and visualisation application where users can build systems including adding qualitative and quantitative techniques.

Q1 Home institutions

The wordle gives an indication of the distribution,

[[CCPSlide35.JPG]]

but the institution links allw us to see the extremely long tail that makes the statistics not a power law relationship.

[[CCPSlide36.JPG|500px]]

Zooming in we can see the individual major establishments.

[[CCPSlide37.JPG|500px]]

Q2 Which CCP(s) are you involved with?

Looking at the 'old' CCPs there were very few responses which indicates users have moved on to new CCPs or institution facilities.

[[CCPSlide39.JPG|500px]]

The newer and therefore more current organisations have a larger response; with CCP5 and CCP9 being popular source code repositories and ISI an CLF being popular user communities.

[[CCPSlide40.JPG|500px]]

Q3 What software do you use for visualisation of data?

Looking at the usage with both 'frequent' and 'essential' categories then we have the following curve.

[[CCPSlide41.JPG|500px]]

again we can zoom in and see in more detail the most popular tools.

[[CCPSlide42.JPG|500px]]

But removing the frequent and just looking at 'essential' tools the graph is very different and shows how different user communities use different tools.

[[CCPSlide43.JPG|500px]]

A wordle can also be useful to spot your favourite tools

[[CCPSlide44.JPG|500px]]

When asked about other tools there is an even longer tail to be found.

[[CCPSlide45.JPG|500px]]

The three most often used tools (gnuplot, MATLAB, Jmol) account for 26% of the responses and conversely, there is a tail of 31 packages that also accounts for 26% of the responses. Similarly the top four Essential tools (gnuplot, MATLAB, VMD, xmgrace) account for 42% of the responses. We need to ask why is there a long tail? Are there any useful features provided by the less popular tools that are not provided by gnuplot, MATLAB or Jmol? This is answered in the comments of questions 5 and 6.

Q4 What visualisation techniques are important to your work?

[[CCPSlide47.JPG|500px]]

This is lessradical when we also add thre 'occasional' used techniques.

[[CCPSlide48.JPG|500px]]

The most frequently used visualisation technique is to produce publication quality output. The facility to produce line graphs is equally important

Q5 Comments on the respondents' most used visualisation tool.

This section is converted into an interactive exploration. In summary 91 replies were given with the most commonly used tools, are:

* [[VMD]] - 9 replies
* [[MATLAB]] - 6 replies
* [[Avizo]] - 5 replies
* [[Gimias]] - 4 replies
* [[PyMol]] - 4 replies

Of the other tools, the major reasons for choosing them were:
* ASE: quick and easy to use
*Avogadro: open source
* gnuplot: quick to use, scriptable
* ImageJ: free, easy, plugins
* [[Paraview]]: easy of use

A few observations can be drawn from these responses;
# Users will prefer software that is written specifically for their domain of interest.
# Large datasets must be handled efficiently.
# Scripting or other ability to extend the tool is required.
# Publication quality output is a valued bonus

Q6 Same as question 5, for any other tool used.

Regarding the second most used visualisation tool 47 replies were given. The most commonly used tools, with links to answers are:
* [[gnuplot]] - 5 replies
* [[Materials Studio]] - 3 replies
* [[Matlab]] - 3 replies, but is not discussed in this section as the responses are extremely similar to those in question 5.
* [[Paraview]] - 3 replies
* [[PyMol]] - 3 replies, also not discussed here.
* [[VMD]] - 3 replies, also not discussed here.

Other tools that had two responses:
* CCPN Analysis
* Discovery Studio
* IMOD
* JMOL
* Vesta
* Visit
* xmgrace

A few observations can be drawn from these responses:

# Users seems to prefer software that is general purpose.
# Large datasets must be handled efficiently.
# Good quality documentation/tutorials is required.
# Ability to read multiple formats is useful.

We can combine the frequency of tools in Q5 and Q6 to produce an interesting frequency graph.

[[CCPSlide54.JPG|500px]]

Q7 Visualisation requirements. How important do you see the provision of the following (ed. Services)?

[[CCPSlide55.JPG|500px]]

Q8 Requirements for high performance/advanced visualisation facilities. Do you have any need for access to (ed Services)

[[CCPSlide57.JPG|500px]]

Q9 What do you see as the main challenges for visualisation in your domain now and in the near future?

[[CCPSlide58.JPG|500px]]

Q10 Any other comments?

Summary
# Code development/maintenance in various guises
# Training in using various packages is required

Raw Comments
# Would be nice if there was an obvious preferred open source tomographic reconstruction code that we (STFC/EPSRC) could recommend to users and modify ourselves for our own needs. If there are candidates here, maybe publicise the options more widely?
# Would be nice to have somebody looking into maintenance and support of useful tools and software.
# Visualisation and application of calibrations to data go hand-in-hand together, thus the greatest visualisation tool won't be used much if it isn't integrated into a data handling flow. A good example is XCrysDen, which makes all the difference between just number crunching with wien2k and actually seeing what you're doing.
# It would be great if the visulisation tool developers would start to collaborate to develop libraries, data formats etc that could be shared between them, to speed up the development of new tools and functionality, and allow users access to the full spectrum of tools they require.
# I'm happy to get involved with this CCP/project or any initiative it might lead on with. I'm passionate about data visualisation and have experience in developing such tools.
# Training on visualisation tools are really lacking. There should be more of them.
# More seminars to introduce what is available
# There is a strong need for web-based visualization methods to integrate into web-based applications we are developing.
# License sharing and advanced usage training
# Lack of tools limits the science we are able to achieve.
# Final note - Martin Turner at RAL is brilliant at cross promotion (between communities) and compiling enthusiastic newsletters and should be thoughtfully priased for his dedication to the subject. He and his team are successfully promoting they ideas that other communities are way ahead in.
# I'm not sure if Diamond are as involved as they could be. If they are it certainly isn't communicated well to Users.
# Many of the packages are great for visualization but quick mathematics (e.g. subtract two 3d data sets) requires recoding.

== Future results ==

There will be a six month review in Spring 2015 - but remember from the presentation "The user and viewer are always important"