EMMC The European Materials Modelling Council
Introduction to Interoperability
Part III
Alexandra Simperler
On-line 29.4.2019
https://emmc.info/
EMMC The EMMO round table
Emanuele Ghedini(University of Bologna)
Gerhard Goldbeck(Goldbeck Consulting)
Adham Hashibon(Fraunhofer Institut)
Georg J. Schmitz(Access)
Jesper Friis(SINTEF)
EMMC Outline
• Interoperability
• Interoperability and Ontology
• Open Simulation Platforms (OSPs)
EMMC Interoperability
the exchange of information is achieved only when a common representation is used between different systems
wrappers (in case of software systems) provide a translation between the internal representation of a system to a common representation, that can be understood by other systems and translated in their respective internal representations by their respective layers
EMMC Interoperability vs compatibility
Interoperability: ability of two or more
systems to exchange information between them through a common
representational system to perform a
complex work that cannot be done by each
single system aloneno privileged one-to-one connection
between two system types should be
implemented within the interfaces � one system can ignore the details about other
systems.
Compatibility: ability of two or more
systems to establish a one-to-one connection between them
usually due to strong similarities in
their internal representations that facilitate mutual understanding
systems are fully aware of the type and identity of the other connected
systems.
© Emanuele Ghediniand EMMC-CSA
EMMC Interoperability Problems …
… in Materials Modelling:
horizontal interoperability - the use of
different codes for a single materials
case
vertical interoperability- transferring
data between several codes and
model types used to simulate the
same material case.
EMMC Semantic Interoperability Levels
Numerical level - code representations
Material user case level -representation of the material
Scientific community level -enabling exchange between scientists
EMMC
Interoperability
and Ontology
EMMC Scientific Community Level
• establishing a common “language” (i.e. a common vocabulary, terminology and a standardised classification) covering various communities and sub-disciplines
• suitable common representation systems (human readable form): journal publications, reports, workshops, …
• First steps: Review of Materials Modelling (RoMM) and the CENWorkshop Agreement: Materials modelling – Terminology, classification and metadata (CWA), standardised documentation of simulations (MODA)
� The formalization of the EMMO in OWL-DL enables the use of tools like Protégé or OWL-API that can be used directly by a single expert to understand the standard and use it to represent the information
EMMC Material User Case Level
Material user case: univocal and detailed descriptions of problems of interest of the materials community (e.g. a material, a device, a process, …)
Includes information about the components and their typologies(e.g. fluid, composite, molecule, …), the spatial extension of each component, the evolution in time, the physical phenomena involved
To enable interoperabilitymaterial user cases must be formalized in a common representational system that allows a rigorous expression of all the details material user case should be understood by other systems (e.g. experts, software)data management is important
EMMC Material User Case Level
So, if the descriptions
we need on the
mesoscopic level as demanded by the
ontology can be given,
we could model on the mesoscopic level. If
they cannot be given,
we are forced to work
on the continuum level.
An ontological approach for the description of the user case, based on
mereotopology, is better suited for the purpose to build an
interoperability architecture:
EMMC Numerical Level
Discretization-methods: discretization techniques of the physics equations and materials relations (e.g. FEM, FVM), the discretization of space (e.g. grid generation), the generation of the numerical problem (e.g. linear system)
Numerical-solver: pure mathematical, includes the techniques used to numerically solve the problem generated by the discretization technique
Interoperability environment should enable the exchange of information between systems so that different numerical-solvers can be used to solve the problem generated by a specific discretization-method applied to a physical model.
Connected systems can be automated systems (e.g. software).
EMMC
Open
Simulation
Platforms
EMMC Architecture
Codes
several software
codes that
represent various
elements of a
simulation
pre-and post-
processing
OSP
open semantic standard
described in an ontology
supports all aspects of a
simulation, including pre-
and post-processing and
computational
representations of
models
Wrappers
support different
levels of linking and
coupling
EMMC Coupling/Linking Taxonomy
• High-level wrappers/Low-level integration: Software tools are black-boxes (commercial, no access to the runtime model data ) - Wrappers work at data level only and are essentially data translators to provide standard format inputs and outputs dataset to and from the third-party software tool.
• Mid-level wrappers/Mid-level integration: Software tools (grey boxes, commercial plus API, open source) provide partial access to internal data through API that are used to build the wrapper that will connect them to the OSP.
• Low level wrappers/High-level integration: Software tools (white-boxes) allow tight coupling.
De
ve
lop
er K
no
wle
dg
e
OS
P F
lex
ibility
EMMC Ontology and Metadata
38.5 °C ?
Syntactic interoperability: YES - declaration of the format
Semantic interoperability: NO - the meaning of the data is yet unclear
EMMC Ontology and Metadata
Ontologies can provide a meaning for data by the definition of concepts to which data belongs
The metadata-based architecture for interoperability can then be derived directly upon representation of concepts in a formal ontology.
Not easy …• several different languages (e.g. OWL, Common Logic, CycL) • logical frameworks can be different (e.g. OWL-DL, FOL, HOL)• Wrong choice in the design phase• Need to agree on metadata schema• Unestablished automated procedure to move from concepts to
metadata
EMMC Example: SimPhoNy
© Adham Hashibonand
EMMC Example: SimPhoNy
Common Universal Data Structures (CUDS) were developed � metadata schema for representing all
information needed to perform a simulation
Common Universal Basic Attributes (CUBA) � constrained
list of vocabularies transcending all models
Data structures generation from an ontological representation, using a code generator to build Python classes to build data structures from a YAML (YAML Ain'tMarkup Language) formalized ontology and thus achieve interoperability.
EMMC EMMO and Interoperability
Mereology-based multiscale interoperability
Granularity (better than scale!) as tool to navigate
through scales (vertical interoperability)
User case and connections with several models
One user case and multiple models scenario
(horizontal interoperability)
EMMC
… a
nd
be
yo
nd
!
Interoperability Interoperability Interoperability Interoperability
on a large scaleon a large scaleon a large scaleon a large scale
Modelling Modelling Modelling Modelling
integrated in integrated in integrated in integrated in
Business Business Business Business
DecisionsDecisionsDecisionsDecisions
Where are we going?
Digital Digital Digital Digital
MarketplacesMarketplacesMarketplacesMarketplaces
EMMC
EMMC-CSA project has received funding from the European Union's Horizon 2020
research and innovation programme, under Grant Agreement No. 723867.