canSAS-X

From canSAS
Revision as of 16:49, 11 December 2018 by Butler (talk | contribs) (→‎NXcanSAS 2.0)

canSAS-X Workshop

  • date: Sunday, October 7, 2018
  • place: Peninsula Room C, Grand Traverse Resort & Spa, Traverse City, Michigan, USA
  • (just before SAS2018)

Background

canSAS (Collective Action for Nomadic Small Angle Scatters) workshops and activities are self assembled/bootstrap efforts to provide a forum for SAS users and providers to come together to discuss common needs and issues, to learn from each others experiences, discuss current best practices, hear about emerging techniques, and importantly to identify areas where the field/state of the art could be improved by bringing together the people that are interested and able to work cooperatively to solve the problem (see canSAS-IX#Goals).

As of canSAS VIII in Tokai, Japan, the goal is to have a full independent workshop every three years, half way between the triennial SAS meetings, followed by some kind of forum at the SAS meetings themselves. These last have ranged in form from a half day session to just a lunch session to disseminate the status of activities. The intention is that the work agreed upon to solve current common problems will take place in the intervals between meetings, either at focused targeted workshops (see canSAS-2012), at ad-hoc gatherings of self-assembled working groups in person, or remotely.

At SAS2018 the organizers have arranged for space for sattelites the days prior and subsequent to the meeting proper and provided canSAS a day long block prior to the meeting itself.


Goals

The goals for this one-day workshop are to get a status report on activities kicked off at canSAS IX and since, prepare for the next full, free standing canSAS workshop to be held in Freising, Germany from 8 to 10 July 2019, and provide a discussion forum for current active topics.


Agenda

time activity title
9:30 - 10:00 plenary Welcome and Introduction to canSAS - Paul Butler (NIST-CNR) and Adrian Rennie (Uppsala)
10:00 - 10:45 discussion Update/discussion on canSAS-IX activities - moderator: Grethe Jensen (University of Delaware/NIST-CNR)
10:45 - 11:15 break coffee
11:15 - 11:30 plenary July 2019 canSAS XI meeting update - Henrich Frielinghaus (Julich)
11:30 - 12:00 discussion NXcanSAS 2.0 discussion - Leader: Tim Snow (Diamond Light Source)
12:00 - 13:30 lunch ...
13:30 - 14:20 Discussion:
Standards/Round Robins
- ISO standards Development- Leader: Michael Krumrey (PTB)
- Geosciences validation and round robin - Leader: Paul Butler(NIST-CNR) for Larry Anovitz (ORNL)
- Other standards/round robin possiblities - Leader: Adrian Rennie (Uppsala Uni.)
-
14:20 - 14:35 discussion Education and Information: SAS video competition +... - moderator Pete Jemian (ANL)
14:35 - 14:40 Group Photo Outside if possible
14:40 - 15:10 break coffee
15:10 - 15:50 Discussion:
Analysis Software
collaborative Opportunities
- Correlative Analysis and COST action - Leader: Brian Pauw (BAM) Sub: Tim Snow (Diamond)
- Data and/or code repositories - Leader: Eliot Gann (NIST-NSLSII)
- Other opportunities (machine learning/MD simulations etc) - Leader: Paul Butler(NIST-CNR)
-
15:50 - 16:10 discussion Open Mic: Other topics of interest to attendees - moderator Andrew Jackson (ESS)
16:10 - 17:00 discussion Discussion on topics and organization of canSAS-XI - moderator: Brian Pauw (BAM)
17:00 ADJOURN Adjourn to SAS2018 Opening Reception


Venue and Registration

  • date: Sunday, October 7, 2018
  • place: Peninsula Room C, Grand Traverse Resort & Spa, Traverse City, Michigan, USA
  • (just before SAS2018)

Note that as a "grass roots" community effort there are generally no specific funds to support the local organizing expenses of such meetings. Instead these expenses will be covered through mostly small registration fees asked of each participant. (Expect $30 to cover the two coffee breaks). Lunch is not included.


Participants

canSAS-X participants.JPG


Meeting Topics

The topics being addressed at the canSAS X workshop are:

  • Status of ongoing activities
  • canSAS XI planning. Practical information at http://www.fz-juelich.de/jcns/EN/Leistungen/ConferencesAndWorkshops/canSASXI/_node.html
  • NXcanSAS data formats and proposed improvements.
    • improve the description of how Q should be provided in the HDF5 file
    • Other needs?
  • Planning new round-robin samples/measurements/data treatment
    • Geosciences USAS: rocks/shales
    • particle sizing standards?
    • Other candidates?
  • Collaborative opportunities for advancing data analysis tools
    • COST action
    • International "networking" or "collaboration grants
    • Other
  • disseminate information about canSAS and to recruit active participants to canSAS activities
  • Other topics as might be of interest to those present - everyone is welcome to raise ideas as to what needs to be done and how to progress. There is a specific 'Open Mic' period in the timetable.


Report on the Meeting

Everyone is encouraged to contribute to the meeting report and to post slides and other relevant material.

Welcome and Introduction

The introduction to canSAS (P. Butler, A. Rennie) described briefly the activities and the success over the past 20 years since the first meeting in Grenoble, France. Particular aims for the day were reiterated with an emphasis that there are many on-going and new activities. Today there should be reports on status and plans for actions. A significant aim should be to prepare for the workshop next year in Freising, Germany (8-10 July 2019): http://www.fz-juelich.de/jcns/EN/Leistungen/ConferencesAndWorkshops/canSASXI/_node.html

Introductory Slides. Presentation

It was noted that some people were delayed in their arrival and others would have to mention relevant themes.

Update/discussion on canSAS-IX activities

The progress on tasks identified at canSAS-IX was reviewed. Grethe Jensen led this discussion. Topics marked with stars are marked for further discussion later in the day

  • Reduced data formats – including meta data with extensibility. NXcanSAS update *
    • The NXcanSAS 1.0 format, an n-dimensional reduced scattering data format which is extensible and provides for highly rich meta data, was approved by NeXus and released in early 2017 and announced at the canSAS-X meeting in Berkeley in June 2017.
    • Since then, several groups have been working on implementing it as an input/output to their date reduction/data analysis packages: Mantid & DAWN currently output NXcanSAS, while NIST reduction will do so soon as the new reduction software is completed; NSLSII is moving away from data formats to databases but the standard as written could still be used; APS (Jan Ilavsky) is working on output and reading; and SasView reads it, while Born Again is starting to look at reading it.
    • With one year of experience now several issues have come up which will be discussed/reported on more fully in the relevant section below.
  • Reproducibility and Reliability
    • New experimental calibration standards and round robin data collection. Including for GISANS and soft X-ray
      • Andy Smith has worked on a 100 nm Silicon Nitride calibration grating for SAXS. Data has been collected and it works (accuracy seems to be ~ 1%) however the gratings are rather fragile and the cost of manufacture is ~ 3500 GBP.
        • Action: Andy to contact Silson out about possibility and cost of bulk order
      • Joe Kline has followed up as promised by engaging the NIST nanofab which is currently working to produce platinum filled Silicon Oxide troughs and pillar, backed with nitride and polished. If successful the next step will be to produce randomly arranged pillars.
      • Adrian update on nanocrystalline glasses: First set of samples were a bit too thick for x-rays. Some were sent around but been mislaid. In the meantime need to work on getting them thinner.
        • Action: Adrian needs to get back to the company on thickness that can be produced
      • The working group should continue working and set up protocols for using the standards developed.
    • Documentation of best practices for data collection and reduction
      • The paper on the topic Brian Pauw offered to spearhead was published last year: B. R. Pauw, A. J. Smith, T. Snow, N. J. Terrill, A. F. Thünemann, The modular SAXS data correction sequence for solids and dispersions, Journal of Applied Crystallography, 50: 1800–1811 (2017), DOI: 10.1107/S1600576717015096. Tim Snow will be speaking about it later at this SAS2018 meeting.
    • Data repository – including metadata *
      • Brian Pauw has started a data repository at BAM and initiated a collaboration with Diamond. Work is ongoing.
      • Chris Tissone (SLAC) and NSLS-II are also looking at this for Machine Learning purposes and have some local repos started.
      • Could also think of adding data under scattering-central on github.
      • At this point need some experimentation to find best approach(es)
    • Multiple scattering effects - warning and model fitting
      • The ideas presented by Grethe Jensen are now published (G. V. Jensen and J. G. Barker, Effects of multiple scattering encountered for various small-angle scattering model functions, Journal of Applied Crystallography, 51, 1455-1466 (2018), doi:10.1107/S1600576718010816), while Henrich Freilinghaus is also working on a paper on the topic.
      • Brian Pauw is working on data quality flags and Grethe is working with Paul Kienzle to incorporate multiple scattering into SasView. Henrich is also hoping to develop software tools.
  • Data Analysis and Emerging Techniques
    • Multi method data analysis
    • Software repository (workshop?) *
  • Experimental Techniques
    • ISSE standards for interoperability
    • Better supplier support for widely used equipment
    • Quick guides, manuals, logs with updates. Sharing helpful?
  • Information and Outreach *
    • Advertize the SAS portal: smallangle.org
    • Software and data repositories
    • List of papers with scattering models
    • Case studies
    • Use of video format

canSAS XI meeting planning update

The location and arrangements for canSAS-XI were presented by Henrich Frielinghaus. Slides The website is up and linked from the wiki. The Venue will be Freising (near Munich airport) July 8-10, 2019. The registration fee will be 250€ and include meals. A single large room has been reserved. However, given the discussoin on the benefit of having small groups to promote active discussion, Henrich will look into booking some smaller rooms which should be possible.

Let Henrich know if people need an official invitation.

NXcanSAS 2.0

After several years of work, the NXcanSAS standard was approved By the NeXus International Advisory Group (NIAG) in early 2017 and announced at the Berkeley canSAS meeting. In the year since that announcement the status of implementation, the Diamond reduction software DAWN, the neutron package MANTID used by a growing number of neutron sources, while the NIST IGOR based reduction macros are in the process of implementing an NXcanSAS outputand. The APS IGOR based reduction NIKA and analysis IRENA packages written by Jan Ilavsky have implemented both the reading and writing of their reduced data in the NXcanSAS format. Finally, on the purely analysis side, [www.sasview.org| SasView] has implemented the reading of such files and Born Again is considering its use. Finally, it was noted that NSLS II is starting to move away from data files in favor of databases. This however could easily be used to create NXcanSAS data on the fly as needed. Two themes however emerged in the ensuing discussion:

  • From the point of view of the producers and consumers of data, there is an urgent need for clarification of the spec, or at least a meeting to understand why the various versions implemented to date do not all seem to inter-operate. This may simply require some discussion and careful reading of the full specification or it could require adding some clarifying language to the current spec (which could require a version 1.1).
  • On the other hand, from the point of view of the rapidly growing communities interested in data curation/databases, machine learning and automated data analysis tools, there is an urgent need to add specific tag definitions for common metadata in order for the utility of such a complete standard to be fulfilled. While the standard allows for any metadata to be included for an individual’s specific purpose, its utility is very limited without an agreement that all producers and consumers of the data can understand.

As part of this last point, the interest ( see software discussion below) in intensifying efforts to analyse and refine multiple data sets from different analytical techniques against each other, the question of handling multi-modal data sets and how to store them was raised. Of particular interest was the growing use of multiple techniques “simultaneously” at the beam line. Indeed a subject of some discussion was the fact that some measurements may be truly simultaneous while others may precede or follow the SAS measurement (e.g. inline chromatography) and how to distinguish between time-stamps vs time differentials. Some proposals were put forward. It was suggested these could be implemented within the current standard though it was not clear to everyone how that would work. Some example use cases need to be developed to demonstrate. Furthermore, the consumers of the data (analysis software) would needed that understand the information and can handle plotting and resampling of such data.


Additionally discussion was had about:

  • Handling uncertainties, specifically highlighting scalers, dI and dQ datasets, and maintaining their provenance and propagation throughout correction, reduction and analysis
  • A key point made by the industrial participants was the need to ensure that the requirement for any metadata be kept to an absolute minimum. In particular the data and metadata need to be stored in separate files when dealing with data from proprietary measurements. The philosophy used in developing the current specification was discussed. This was to require an abosulte minimum, namely an I and a q, but allow for producers and consumers of data to communicate at as high a leve as desired by defining a variety of useful but optional tags. This should continue in the next iteration of the specification.


It was agreed that besides clarifying the current standard, the working group should develop some more challenging example use cases, and in particular how included extended data for multimodal data for example. Moreover it was noted that it was time to “bite the bullet” and try to develop wrappers/converters to and from NXcanSAS and other formats

Reproducibility and Reliability: Standards and Round Robins

ISO standards Development

Michael Krumery briefly described the ISO structure and processes and the status of its sas related activities. ISO is the international standards organization whose current membership includes 162 national standards bodies. At present there are 784 technical committees and subcommittees of which TC-24: particle characterisation including sieving is the one of intereste to SAS. He described how the standards go through a 3 year development track and are reviewed every 5 years. He discussed the current standard which is a technical standard that defines how to do a SAXS measurement to obtain the mean particle diameter (ISO 17867). Finally, he mentioned a new proposal to create a total surface area standard based on the SAXS invariant analysis.

In the ensuing discussion it became clear that the creation of ISO standards require a lot of work are generally driven by industrial/commercial needs to for example be able to certify equipment and is not really suitable for academic use in for, for example, validating data to be published. While not impossible to extend to SANS officially, the commercial need case is hard to make. It was suggested that it might be interesting for SANS instruments to try to meet the standard informally. It should be noted that ISO standards are about process not standard samples. That said, canSAS activities could probably contribute information into the ISO process, for example through its round robin effort.

Geosciences validation and round robin

Paul Butler pointed out that a large fraction of the USANS proposals at NIST at least are geosciences related but that there seems to be a need and an interest to help validate geosciences methodologies. These are mostly based on surface area and pore volume analysis but suffer from a variety of issues, not the least of which is very strong scattering, often leading to multiple scattering. Sometimes these are addressed by trying to make thinner samples but there is concern about the effect of grinding the surface on the results obtained. Another method often employed is to crush the samples into a "powder" to avoid the potentially damaging effects of grinding, but this can create a lot of new surfaces which will scatter in their own right. Finally the analysis methods can differ. It is proposed that a set of simple model geosamples be produced in a variety of the standard forms and be measured as round robins and the data also analyzed in various ways by various people to explore the limits of technique. An Initial set of USAS instruments at NIST, ANSTO and Argone will be used. Others will be contacted in due time but anyone interested in participating immediately is welcome. Paul Butler (butlerpd@udel.edu) will try to co-ordinate this effort.

The suggestion was also made that depending on how the project progresses, the outcomes could be useful to the new ISO effort on surface area.

Other Standard/round robin discussion

Education and Information: SAS video competition etc.

Analysis Software and Collaborative Opportunities

Correlative Analysis and COST Action

In April 2018 a COST action was submitted under the title of Combining Everything: Developing Rigorous Pan Metrological Correlative Analyses (Pan-CORAL). The aim of this action was to bring together a network of people to work on a series of information optimisers, qualifying and quantifying data present in a given dataset and then bringing together this information from multiple data sources into a global optimiser to provide a top-level view of the information gathered about a sample. Conversely, ultimately, the system would also be able to propose suitable analyses to conduct on a sample, given input from a user about what it is they may want to quantify.

Aside from the academic network, the proposal also had provision for a training and outreach network designed to engage the community and build a mass of users behind it for when the initial tranche of funding came to a close.

Unfortunately, the proposal failed the first submission with the authors and proposers awaiting its review documentation. It is anticipated that, subject to revisions, it will be resubmitted within the next 12 months.

Data and/or code repositories

Other opportunities (machine learning/MD simulations etc)

Open Microphone Session

  • An open question was posed - given the potential to develop/build an algorithm that could sift through, and/or analyse, user data in real-time:
    • a) What would the canSAS group wish to get out of such an algorithm - how far should it go?
    • b) How would it be possible to prevent it becoming a black-box?
    • c) How would one display to users of such an algorithm that there is, usually, no one correct answer and that you may have to subsequently look at, or cross-correlate, a given result?


Local Organizing Committee

  • SAS2018 organizers


Program Committee

  • Paul Butler (NIST, US)
  • Adrian Rennie (Uppsala University, Sweden)
  • Pete Jemian (APS, US)
  • Henrich Frielinghaus (JCNS, Germany)