Evolving requirements for materials modelling software and underlying method developments: an inventory and future outlook

This European Materials Modelling Council (EMMC) study provides an outline of the survey intent and ambitions, followed by an analysis of the results and a follow up discussion, focused on the future perspectives of the EMMC. The survey covers materials modelling and characterisation communities in both academia and industry. It provides a profile of the surveyed players in these communities and a scaled measure on their usage of computational methodologies. The survey outcomes include: (i) summary views of the recent as well as perceived future trends of materials modelling and its associated fields, with respect to two focus areas surveyed, Model Development and Software, (ii) the main adoption factors and associated bottlenecks for computational methods and software, (iii) the most targeted materials properties and digital twins approaches, and (iv) the wider communities expectations of how EMMC can help facilitate, fulfil and drive further the European Materials Modelling Roadmap to the benefit of the European Commission’s (ECs’) research and innovation.


Amendments from Version 1
The new version is a minor revision which includes typographic corrections and clarifications based on referees' comments: -clarification that for some questions respondents could select multiple answers (e.g.multiple roles) and hence the sum of percentages is not weighed and hence not 100%.
-clarification of the discussion on Table 5 results.
-Scaling Table 4 and Table 5 5 to have the same dimensions in the printed version (pdf).
-removing a doubled full stop of Figure 4 and Figure � � captions.
Any further responses from the reviewers can be found at the end of the article

Plain language summary
The European Materials Modelling Council (EMMC) focus areas of Model Development and Software lead the EMMC's activities in their domains and the key needs of their representative researcher communities are summarised as follows.For Software these are more accurate, robust, well-documented and validated/verified software, with better availability of parameters data (easy generation of input), applicability (general purpose/demonstrators), scalable performance & lower complexity to use.For Model Development these are improved capability, accessibility and performance of methods, with better applicability (general purpose over specificity).

Introduction
This report is based on the kick-off survey initiative, instigated by the Software and Model Development focus areas (FAs) of the European Materials Modelling Council (EMMC) upon re-establishing the council from a H2020 coordinated support action project and association into a professional community as a non-profit organisation (association sans but lucratif [ASBL]).The two FAs made a fresh ad hoc start in May 2020 and were interested in better identifying their perspective communities and their needs and expectations in order to drive their own development initiatives and help develop EMMC ASBL as a bottom-up organisation.Ilian Todorov, acting as co-chair of the EMMC Organisational Assembly and of the Software FA, proposed that a common survey from the two FAs "is of benefit to both as they are ultimately linked, connected and bound by the subject of scientific methodology".
Despite the large commonality and overlap of these two FAs' audiences, their main stakeholder groups are very different and can be summed as a modelling and simulation researchers, modellers and theorists/theoreticians for the Model Development FA; and software owners, computational scientists, computer scientists and research software engineers for the Software FA.It is worth noting that both FAs include data scientists and experimentalists (characterisation) who are also intrinsically associated with the stakeholder roles described above.
The survey was carried out between 21 November 2020 and 12 January 2021, as commissioned and endorsed by EMMC.
Its main purpose was to map the landscape of materials modelling by (i) collecting a focused feedback from the wider community of materials modellers with respect to the two FAs (a cross-section of many communities that work in these areas or benefit from them directly) and (ii) identifying and clarifying the communities' interests for the benefit of the EMMC.The survey outcomes are summarised in the following sections and give a clear mandate to the FA chairs as well as to the EMMC board of directors (BoD) to act upon these.

Methods
This survey was the very first one organised on behalf of the EMMC ASBL.As such it aimed to classify and identify the participating audience, their interest with respect to the EMMC main areas of business and their background.This classification included identification of funding support and models of operation of the participants' hosting institutions.The small size of the survey was deliberately chosen in order to keep participants focused.Several questions on identification, strengths and weaknesses offered a close preselected choice of options in order to produce compact and easy to identify and analyse trends.However, many questions were offered in an open format and concerned the wide spread of EMMC terminology and methodologies, e.g., the MOdelling DAta (MODA) (de Baas, 2017)., Psi-k -Ab initio (from electronic structure) calculation of complex processes in materials Network, MOLECULAR-DYNAMICS-NEWS @ JISC, EMMC) and social platforms such as LinkedIn and Twitter.
The participants were made aware of EMMC and EMMC ASBL ambitions and were provided with a link to identify the relevant FAs and their description.The Model Development FA was defined as: 'Everything that has to do with the capabilities and qualities of the materials models and the modelling workflows: development, validation and application.'The Software FA was defined as: 'Successful materials modelling software uses the best algorithms, it is numerically robust, carefully validated, well documented, easy to use, and continuously maintained and supported over decades.'.The survey was conducted by a web-form, the collected data from which were anonymised and subsequently analysed as described below.

Results
The following sections summarise the results of the survey (Todorov, 2022), which had 98 respondents in total.

The players
This section is concerned with the background and composition of the respondents.Approximately 46% of them were affiliated with the EMMC ASBL as members.As depicted in Figure 1, 58.2% were based in academia (universities/ higher education), 24.5% in research and technology organisations (RTOs) and 13.3% in industry; 3% answered as individuals and only ~1% were affiliated with national research laboratories.
RTOs (Charles & Ciampi Stancova, 2015) have been established in many European countries with the aim of assisting local industry around a specific set of industrial technologies or focusing on specific industrial sectors to advance their development and digitalisation.They are often founded with governmental funding and are either planned to self-sustain in the long-term or are strongly encouraged to diversify within a few years and develop opportunities to rely more on non-government funding from local and international sources, e.g., industry or EC research and innovation (R&I) funding calls.Hence, we took an inventory of all respondents from RTOs to gain an understanding of how their funding situation unfolded at this very time (from 15 December 2020 to 15 January 2021), since the uptake of materials modelling would always require some initial monetary input.In total, 27.7% of the respondents qualified for this additional survey and were given the option to choose more than one funding source, if and as applicable.
In Figure 2, one can see that the RTOs had similar amounts of funding from government, academia and industry, while strategic (own) funding streams were emerging.
A closer look into the funding landscape outcomes from the RTO respondents, as depicted in Figure 3, reveals that more often a mixture of several types of funding was relied upon than just a single type.Overall 40.7% claimed their funding from one source, 29.6% from two sources, 22.2% from three sources and 7.4% from four sources.Approximately 60% of all RTO respondents still relied on some amount of government funding, which was on par with funding coming from research councils (academic in nature).However, the funding landscape clearly showed, with ~50% industrial co-funding, that the private sector played a significant role in the business model of RTOs.
Of the 13 industrial participants, nine identified themselves as being from the software and services domain (~70%), two were from materials manufacturing companies (~15%) and one each were from the chemical and the oil and gas industries.
The respondents were asked to share which roles they take on (multiple choice of roles was allowed) in their respective organisations, which are shown in Table 1.
Figure 4 shows that the majority of our respondents were using or developing software (~56%) and 23 of them were in a managerial role (~12% of the total responses).A few individuals (~5% of the total) were practising digitalisation and interoperability, similar to the number of translators/consultants we could capture in this study.Only a few, ~2.5% of the total responses, declared their role as business (and innovation) developers.These results are evidence that some participants could be identified as more closely related to other FAs within the EMMC, ultimately showing that all FAs are connected to the ones surveyed here.
The majority of our respondents (61.2%) had been in their roles for more than 10 years, while only 8.2% were relatively new to their activities, as shown in Figure 5.This gives high confidence and relevance of the outcomes of this survey to the FAs' themes of interest.
The respondents were asked to share what types of research they are involved in within their organisations.The results, summarised in Table 2, show that the majority (~65%) are involved in the computationally and methodology driven environments of the two FAs surveyed.However, there are significant fractions of experimentally assisted (~15%) and data driven (~13%) modellers.It is also relevant to note that a  non-negligible fraction (6%) of respondents are associated with digitalisation and interoperability aspects of modelling.

Model scales and purposes of modelling
Figure 6 summarises the responses for the levels of models used by the respondents' institutions.Given the large proportion of respondents based in academia, it is not surprising that atomistic and electronic scales dominated over mesoscopic and continuum scales, which are known to be more popular with industry and hence more often commercially applied and exploited.However, it is also clear that a significant proportion of the respondents' institutions carried out research where scales were coupled, which also links to the FA for interoperability.
Figure 7 clearly shows the main purposes for which research was carried out.The results were aligned with the types of work the respondents based predominantly in academia (~58%) and RTO (~25%) carry out; rating materials design and theoretical purposes as the main drivers, followed by virtual screening and testing.More industrially relevant purposes, such as process performance, component design and evaluation, scored comparatively low, which was unsurprising given the smaller fraction of industrially based respondents (~13%).
The respondents were asked to share the types of materials their modelling research lines focused upon.Table 3 summarises the responses scoring more than 4% of the total selections over all material types.The respondents were allowed to select multiple types and add new materials they considered specific to their research lines, hence the dominance of these in the "Other" category.The general trend seemed to be an equal, 50/50, split between solid state and soft matter, with a small fraction (~4%) of meta-materials.

The future of modelling
The most important part of the survey tried to capture the perceived modus operandi of how modelling had changed in the past 5 years, shown in Table 4, and what the expectations Researcher/modeller (using software for producing research data) ResUser 59 30.9 Digitalisation and Interoperability practitioner -semantic workflows (Horsch et al., 2020) (ontologies [Goldbeck et al., 2019;Horsch, 2020], metadata, cloud integration), code coupling for multi-physics and multiscale  of the surveyed audience were for the next 5 years, given in Table 5.
It is clear from Table 4 that nearly half of the respondents had experienced no big changes in the type of modelling work carried out, nor the tools used, in the past 5 years.This is not entirely surprising given the identification make-up of the respondents in Table 1.Of the rest of the respondents, 14% had adopted and invested in workflow and multiscale technologies.
A similar proportion (13%) had ventured into data driven modelling, with around 8% putting more emphasis on this area than on traditional modelling.Coincidentally, the same proportion fits the Digitalisation and interoperability driven (interscale and workflows) in Table 2.Only 8% acknowledged adoption of better models and software, and 4% exploited the benefits brought by increased HPC power.The remainder confirms that the smallest changes were seen in developments of accuracy and scalability performance of models as well as in the use of cloud computing.
The forward look in Table 5 is quite different.The only similarities with previous changes were the smallest expectations placed on the development of models' accuracy and scalability performance, which are possibly the most difficult to achieve and require significant labour investment.The expectations of using cloud computing rose slightly (4%),  as did those for increasing computer power (7%), which now included quantum computing.
The biggest changes in the forward outlook are in the distribution at the top.The number of responses of no change nearly halved (from 50% to 28%), while databased modelling using ML and AI nearly doubled (13% to 28%).The expectations for multiscale modelling and workflows stayed constant (13%), although these now also included an accent on interoperability.What also is considered a change is the establishment of data driven modelling (12%, which includes both physical equation based modelling as well as characterisation driven experiments) as commonplace.Furthermore, expectations were placed on sophisticated infrastructures via more advanced software, workflows and automatisation (11%).

Software adoption factors
This section is concerned with the identification and rating of the factors deemed as important for the adoption of research software, which is of particular interest to the EMMC's Software FA. Figure 8, software adoption factors, shows the aggregated performance of all factors that we asked the respondents to grade on a scale from 1 (least important) to 10 (most important).For visualisation purposes the rankings are grouped into three categories: the low (1-3), medium (4-7) and high (7-10) importance bands.The most important factor is clearly accuracy, while the least important is consulting; this is not surprising given the domination of academic and RTO profiles of the sampled audience in Further analysis on the data here is better coupled with the outcomes in the next section.

Software proliferation bottlenecks
This section is concerned with the identification and rating of the factors that are deemed as bottlenecks for the wider spread and utilisation of research software, again of particular interest to the EMMC's Software FA. Figure 9 shows the aggregated performance of all bottlenecks that we asked the respondents to grade on a scale from 1 (least important) to 10 (most important).For visualisation purposes the rankings are grouped into three categories: the low (1-3), medium (4-7) and high (7-10) importance bands.The most limiting factor is clearly personnel cost, while the least limiting is lack of success stories.The former is quite intriguing and possibly points towards the emerging trends of using software as a service (SaaS) rather than setting up and starting a new RSE group with an institution.At the present, the latter (Usher et al., 2020) is assumed as a strategic consideration by many research organisations; however, in some cases such groups can be split over geographically neighbouring institutions as the home institution subcontracts the services to the others.
As in the previous section, if we define top factors as those with responses where more than half (½) are high, then the major bottlenecks to sustaining success for research software are Personnel Cost and Lack of Parameters.These are closely followed by other important areas.Limited applications points towards further requirements for versatility, such as software suites providing for interoperable and/or multiscale workflows with possible data driven orchestration.
Computational costs refers to the efficient utilisation of modern architectures (hybridised with accelerators) and programming paradigms (including modernisation of computer languages and extensions for accelerated parallelism).Complicated use refers to the cost of production of documentation (at all levels) and the delivery of targeted training (learning prerequisites for new users).Last but not least, we must acknowledge that all bottlenecks have the personnel cost dimension in common.

New methodology adoption incentives
This section is concerned with the identification and rating of the incentives deemed as important for the adoption of new methodology, which is of particular interest to the EMMC's Method Development FA. Figure 10 shows the aggregated performance of all incentives that we asked the respondents to grade on a scale from 1 (least important) to 10 (most important).
For visualisation purposes the rankings are grouped in three categories: the low (1-3), medium (4-7) and high (7-10) importance bands.The most important factor is clearly improved capability and the least important existing MODAs.The latter is surprising; we believe it is actually connected to adoption by the user community.

New methodology adoption deterrence
This section is concerned with the identification and rating of the factors that are deemed as bottlenecks for the adoption of new methodology, again of particular interest to the EMMC's Method Development FA. Figure 11 shows the aggregated performance of all deterrence factors that we asked the respondents to grade on a scale from 1 (least important) to 10 (most important).For visualisation purposes the rankings are grouped in three categories: the low (1-3), medium (4-7) and high (7-10) importance bands.The biggest bottleneck clearly is limited applicability, while the smallest one is missing MODAs.The top bottleneck is understood to originate from new methods being more narrowly specific as opposed to being general purpose.However, the second factor, non essential, suggests that general purpose methods are somewhat more easily adopted and applied, and sufficient for the bulk of current modelling and simulation research.
Properties and digital twins Properties prediction.We asked the respondents for the most important materials properties that their research aimed to measure and predict by using computer simulations or by addressing via developing materials modelling methodologies (methods, models, workflows, software).The respondents were allowed to select multiple types and add new ones they considered specific to their research lines.

Conclusions
The EMMC focus areas Model Development and Software lead the EMMC's activities in their domains and the key needs of their representative researcher communities are summarised as follows.For Software these were defined as more accurate, robust, well-documented and validated/verified software, with better availability of parameters data (easy generation of input), applicability (general purpose/demonstrators), scalable performance & lower complexity to use.For Model Development these were defined as improved capability, accessibility and performance of methods, with better applicability (general purpose over specificity).
This survey endorsed the point that digital tools drive the future of materials in chemical and manufacturing industries, by providing agility and speed in the development process.The materials digitalisation wave relies on using physics-based modelling approaches as well as data-driven approaches and their use together, to accurately predict and optimise industrial products in an early design stage.The materials research community, including methodology developers and software owners, provides cutting edge materials modelling methods and software that both require continued support by funding bodies and organisations, nationally and internationally.

Data availability
eData: EMMC Software and Modelling Questionnaire.http:// dx.doi.org/10.5286/edata/754[7] This project contains the following underlying data: Software and Modelling Questionnaire_v2.xlsx(the data file is in a MicroSoft ® Excel format and contains the anonymised raw data from the survey form as well as in separate data sheets the extracted and aggregated data and the data operations from which the tables and figures in this study were produced).
My major concern is the exceedingly small number of industrial participants and the resulting strong bias towards RTA's and academic researchers.This fact is clearly not due to a lack of effort by the author, but rather reflects the still very low use of materials modelling in industry.It would be useful to provide a more detailed analysis of this fact in this work.Is the gap too wide between actual industrial requirements and current capabilities of (discrete) materials modelling?Are there other factors such as a lack of skilled industrial modelers?
Overall, this is a very useful piece of work and I recommend its indexing.

Is the work clearly and accurately presented and does it cite the current literature? Yes
Is the study design appropriate and does the work have academic merit?Yes

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate?Partly Are all the source data underlying the results available to ensure full reproducibility?Yes Are the conclusions drawn adequately supported by the results?Yes Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Computational materials science and industrial applications.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
Reviewer Report 19 July 2022 https://doi.org/10.21956/openreseurope.16033.r29659 The manuscript entitled "Evolving requirements for materials modelling software and underlying method developments: an inventory and future outlook" reports the results of a survey on material modelling.The survey was aimed towards members of the material modelling community, and is successful in mapping the tendencies, bottlenecks, and potential future developments in this area of interest.The manuscript is well written and structured, while the results and methods are well explained.Therefore, my suggestion is that the article is approved.A few points can be addressed to improve the article.
The plain language summary is exactly the same as the conclusion.It is rather suggested that the conclusions may include points that can be taken out from the survey results. 1.
In the "Results" Section, under the Subsection "The Players", it is mentioned that 58.2% of the survey participants are based in Academia but within the subsection (and the corresponding Figures and Tables) there is no mention or discussion about the findings for Academia participants.

2.
In page 4, it is written that "the majority of our respondents were using or developing software (~56%) and 23 of them were in a managerial role (~12% of the total responses).".Since the total number of participants were 98, 23 respondents cannot consist of the 12% of the total responses.Please explain this small inconsistency.

3.
In page 5, in the 5 th line of the paragraph at the left column (and at Figure 7), one potential choice of the survey concerning the main research purposes is "theoretical purposes".What it is denoted by this term? 4.
The percentages in tables 3 and 6 don't sum up to 100%, which could create confusion.5.
The Table 4 which is discussed under the Subsection "The future of modelling" might be also correlated with Table 2 and not only with Table 1.For example, it is mentioned that "A similar proportion (13%) had ventured into data driven modelling, […]" which fits the percentage provided in Table 2 (12,8%).

6.
In page 8, in the paragraph at the right column, it is mentioned that "the number of responses of no change nearly halved (from 50% to 28%), while data driven modelling nearly doubled (13% to 28%)."Three lines later, it is written "These were closely followed by data driven modelling (12%) […]".The data driven modelling cannot be 28% and 12% at the same time, in the first phrase the term "data driven modelling" should be replaced by the term databased modelling including ML and AI to avoid confusion.

7.
In Figures 8 and 9, for software adoption factors and software proliferation bottlenecks, respectively, how is it possible to adopt a software due to its versatility (52/43/5, high/medium/low importance, respectively) and at the same time claim that the software is for limited applications?

8.
Where the term "Lack of parameters" shown in Figure 9 is referring to? 9.
Since the biggest bottleneck for software proliferation is the personnel cost according to the findings of the survey, how it is suggested to overcome this bottleneck and what can be potentially the role of EMMC on that?10.

Are sufficient details of methods and analysis provided to allow replication by others? Yes
If applicable, is the statistical analysis and its interpretation appropriate?Yes Are all the source data underlying the results available to ensure full reproducibility?

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Material Science and Computational Material Science I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.potential choice of the survey concerning the main research purposes is "theoretical purposes".What it is denoted by this term?A4: Theoretical purposes is often referred as basic study or research, that are often regarded as fundamental and also not yet experimentally accessible or relevant.

Q5:
The percentages in tables 3 and 6 don't sum up to 100%, which could create confusion.
A5: This is related to the Q3 and answered in A3.

Q6:
The Table 4 which is discussed under the Subsection "The future of modelling" might be also correlated with Table 2 and not only with Table 1.For example, it is mentioned that "A similar proportion (13%) had ventured into data driven modelling, […]" which fits the percentage provided in Table 2 (12,8%).

A6:
That is a useful observation although there is no per se direct correlation between the two data sets as the former is an integrated change of research change observations over 5 years whereas the other set(s) static view of respondents' attribution to a particular role.It is true that both describe the emergence of multiscale workflows.

Q7:
In page 8, in the paragraph at the right column, it is mentioned that "the number of responses of no change nearly halved (from 50% to 28%), while data driven modelling nearly doubled (13% to 28%)."Three lines later, it is written "These were closely followed by data driven modelling (12%) […]".The data driven modelling cannot be 28% and 12% at the same time, in the first phrase the term "data driven modelling" should be replaced by the term databased modelling including ML and AI to avoid confusion.A7: You are absolutely correct here as there is indeed a confusion brought up by the wording.The paragraph will change as: "The biggest changes in the forward outlook are in the distribution at the top.The number of responses of no change nearly halved (from 50% to 28%), whereas the databased modelling using ML and AI have doubled (13% to 28%).The expectations for multiscale modelling and workflows stayed constant (13%), although this category now also included an accent on interoperability.What also is considered a change is the establishment of data driven modelling (12%, which includes both physical equation based modelling as well as characterisation driven experiments) as commonplace.Furthermore, expectations are placed on sophisticated infrastructures via more advanced software, workflows and automatisation (11%)."It would be nice to make Table 4 and 5 are of the same dimensions in the print.Going over the captions I have also detected that some double full stops have occurred, which also needs correcting.

Q8:
In Figures 8 and 9, for software adoption factors and software proliferation bottlenecks, respectively, how is it possible to adopt a software due to its versatility (52/43/5, high/medium/low importance, respectively) and at the same time claim that the software is for limited applications?A8: If software is adopted then it does the job well for what it is needed and versatility is possibly not so essential and relevant as it does not score high -it is at the bottom of the scale.However, one would adopt a software solution that is versatile to offer a number of solutions/options/methods/workflows under its hood, i.e. quantum (QM) driver, atomistic (MM) driver, and mesoscopic (DPD/LB) scale driver.On the other hand "limited applications" does not refer to versatility directly but indirectly.It really refers to applicability of the solution to specific systems (as do DFT functionals) and the availability of benchmarks and use cases to demonstrate a more extended applicability (variety of models, model systems, functional, integration schemes/ensembles), beyond those driven by the software creators' research agenda.

Q9:
Where the term "Lack of parameters" shown in Figure 9 is referring to?A9: The general meaning of those is the material relation parameters for the specific physics equation coupling parameters.Hence, these refer to model parameters in general, which in the MD/DPD world are those of the force-field (interaction).DFTB it is also very similar.The same considerations apply for mesoscopic scale (DPD/LB) and continuum software (FEM,CFD, etc.) which solve equations of state dependent on parameters that need to be supplied by the modeller.Q10: Since the biggest bottleneck for software proliferation is the personnel cost according to the findings of the survey, how it is suggested to overcome this bottleneck and what can be potentially the role of EMMC on that?A10: The paper was designed to show unbiased outcomes of the survey and not supposed to suggest solutions, especially as the one you are asking for.I can hazard many answers as I have pleaded to senior officials of funding bodies and in academia over the last 2 decades.Hence, the existence of SSI (UK) and such like organisations world wide (RSSI in the USA) and RSE like associations (Society of RSE in the UK, WSSPE in USA, etc.) Kind Regards, Ilian Todorov Competing Interests: No competing interests were disclosed.

Figure 2 .
Figure 2. FUNDING FOR RESEARCH AND TECHNOLOGY ORGANISATIONS (RTOS) WITH THE Y-AXIS SHOWING THE SOURCE OF FUNDING AND THE X-AXIS THE NUMBER OF TIMES A FUNDING SOURCE WAS SELECTED.THE 27 RESPONDENTS WERE ALLOWED TO SELECT MORE THAN ONE SOURCE.

Figure 3 .
Figure 3. NUMBER OF FUNDING SOURCES IN % CLAIMED BY RESEARCH AND TECHNOLOGY ORGANISATIONS (RTOS) SIMULTANEOUSLY.

Figure 4 .
Figure 4. EXPERTISE/ROLES OF THE RESPONDENTS WITH THE Y-AXIS SHOWING THE ROLE AND THE X-AXIS THE NUMBER OF TIMES A ROLE WAS SELECTED.THE 98 RESPONDENTS WERE EACH ALLOWED TO SELECT MORE THAN ONE ROLE (RSE -RESEARCH SOFTWARE DEVELOPER, RESDEV -RESEARCHER/MODELLER DEVELOPING MODELS/WORKFLOWS, RESUSER -RESEARCHER/MODELLER [USING SOFTWARE FOR PRODUCING RESEARCH DATA], INTEROP -DIGITALISATION AND INTEROPERABILITY PRACTITIONER, BID -BUSINESS AND INNOVATION DEVELOPER, SOFTDIST -SOFTWARE DISTRIBUTOR).

Figure 6 .
Figure 5. TIME RESPONDENTS SPENT PERFORMING THEIR ROLES.

Figure 7 .
Figure 7. MAIN PURPOSES OF RESEARCH WITHIN THE RESPONDENTS' ORGANISATIONS.THE Y-AXIS SHOWS THE PURPOSE AND THE X-AXIS THE NUMBER OF TIMES IT WAS SELECTED.THE RESPONDENTS WERE ALLOWED TO SELECT MORE THAN ONE.
Figure 1.If we define top factors as those with responses where more than three quarters (¾) of them are high, then what defines success for research software engineers (RSE) and can be considered as of utmost importance to the EMMC's Software FA are accuracy, robustness, documentation and verification/validation.

Table 6
The survey asked "In the context of the comprehensive digital twin, with which disciplines should Materials Modelling software be integrated more closely?".The question offered two, non-mutually exclusive options and an optional comments section to provide further detail on the respondents' choices.
summarises the responses, which may be split roughly into three main areas of almost equal interest:• Chemical: chemical and thermodynamic properties, scoring 34.6% of the total interest• Physical: electrical, optical and magnetic properties, scoring 32.2% of the total interest• Engineering: a collection of mechanical, fluidic and interfacial properties, scoring 33.2% of the total interest.Digital twins.•Offerinformation,well-structured and easy to find Case studies (advanced, liked to IP, an actual [product, …)Demonstrator casesLink to organisations who may be able to contribute beyond EMMC's offerings Link to funding bodies who could enable "better materials modelling software"