The integration of a Supercomputer in the educational process improves
student’s technological skills. The aim of the paper is to study the interaction between
science, technology, engineering, and mathematics (STEM) and non-STEM subjects for
developing a course of study related to Supercomputing training. We propose a flowchart
of the process to improve the performance of students attending courses related to
Supercomputing. As a final result, this study highlights the analysis of the information
obtained by the use of HPC infrastructures in courses implemented in higher education
through a questionnaire that provides useful information about their attitudes, beliefs
and evaluations. The results help us to understand how the collaboration between
institutions enhances outcomes in the education context. The conclusion provides a
description of the resources needed for the improvement of Supercomputing Education
(SE), proposing future research directions.
Notes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
1 Introduction
High-performance technologies are an increasingly prominent theme in higher
education. The fast introduction of Supercomputing in education is a challenge that
provides a broader range of opportunities. In this sense, the possibility of managing a
high volume of data and information has prompted educators and researchers to take a
pedagogical view to promote the use of these infrastructures in teaching. Computing
Education (CE), as a popular research field, faces new challenges that require students’
motivation, which is the foundation in systems as the SAIL—a System for Adaptive
Interest-based Learning [3]. New tasks and
challenges arise for the Higher Education System (HIES) which require extensive and
complete training. This training has to face the limitations of the integration of
technologies in teaching [44, 47] and at the same time be included in the standard
curriculum as a constraint related with Supercomputing [26]. Currently, the use of Supercomputers is becoming a key element in
the improvement of certain disciplines of higher education, in spite of the lack of
experience and basic skills in this field [1] and low availability of Supercomputing facilities due to the high
costs [19].
Supercomputing Education (SE), as one of the sub-fields of CE, is
considered important for innovation, modernization and technological development,
contributing to create a new generation of professionals, companies and organizations
related to Science and Technology [11].
This area is also in charge of specific training, mainly for students of computational
science and engineering, and also for students of various fields of knowledge, that use
computational, mathematical and engineering tools for problem solving [45]. Currently, many educational institutions
worldwide are seriously introducing Computer Science (CS) concepts into the basic
training of students [2], which raises the
question of how important the training in Supercomputing for the professional future of
the students is.
Advertisement
The introduction of new information and communication technologies (ICT) in
the training sessions of students of all levels has been considered one of the main
tools in the modernization of all educational systems. It has contributed to evolution
from the “society of the information” to the “society of knowledge” [33], allowing forming students, in an appropriate
way, for the exercise of a great variety of professions. This consideration means that
CS could be suggested for the core area of many postgraduate disciplines, where the use
of a Supercomputer will be essential for solving complex data problems. The fast
developments in computer and electronics technologies provide father-reaching
developments especially in the areas of biomedical and health informatics, which are
very important for improving solutions [25]
for the needs of training sessions of these fields.
For companies and organizations in general, technological qualification is
ensuring corporate competitiveness by improving productivity and reducing the
consumption of human/physical resources through the change of strategic management
environments using ICT technology [31]. In
this, supercomputing training represents a specific learning activity. It provides an
integrated and experimental approach in the use of technologies for higher education
students in advanced contents. These contents are designed to provide experiences in the
analysis of real cases to be used as pedagogical resources. At the same time, it
increases learner engagement, making students feel involved in the course content,
improving the level of technical expertise and expectations regarding technology. The
level of training provided in these courses covers a wide range of areas that depend on
the target category of the trainees. They are supported by specialists in many fields of
knowledge that help students in solving complex computation-intensive applied problems.
SE is based on a wide variety of courses and tools, under different forms of educational
programs that permit the development of these skills. In general, courses linked to
Supercomputing differ from those of advanced training for IT specialists, and can be
classified into three groups [24]:
System architecture: the goal is to teach how to design, implement
and deploy a supercomputing platform.
Applications in Computing Science/Computing Education: based on
the implementation of high-performance programs on a supercomputing
platform.
Specific-domain courses: uses the computing power of a
Supercomputer to solve specific problems of a field, such as computational
physics [32], computational
mathematics [27], and
computational biology [8].
Regarding higher education students, the advantages of Supercomputing
training are the following:
Student explores real problems through hands-on experience
[7, 10, 20, 21,
26, 41, 43], learning the most advanced functionalities of computers,
linking the theory with a practical problem solving.
It is not always necessary, previously to the course, to have
skills in the use of a Supercomputer [38].
Supercomputing training prepares students in a continuous process
to get a level of knowledge and proficiency adequate to deal with the largest
computing systems available [7].
Supercomputing training helps in the use of mathematical models
[9, 42, 50] which allows describing a problem in a specific research
field, for extracting conclusions and evaluating performance on the proposed
problems.
Simulations and data design in a Supercomputer prepare students
for the role of leaders in research [50] by the use of very large database systems [1], covering, at the same time, a great
variety of applications [39].
Supercomputing contributes to inventing new tools and new
philosophies of work, which enhance the learning about computing [20, 29].
Bloom’s Taxonomy of Educational Objectives [12] represents a convenient way to describe the degree to which
students understand and use concepts, demonstrate particular skills, and have their
values, attitudes, and interests affected. According to this taxonomy, supercomputing
training enhances learner engagement by providing experiences that allow students to
practice application skills.
Advertisement
This article presents first an analysis of Supercomputing in education
through a review of relevant literature and secondly, a case study that illustrates the
benefits and challenges of organizing a course of high technology for higher education
students. The main goal is measuring the quality of training sessions through the use of
HPC tools by analyzing how the integration of technology, in the teaching and learning
process, engages students, providing motivation, attitudes, achievement and interactions
[48]. In order to achieve this goal and
based on both the literature review and the experience of the collaborating entities of
this study, a survey questionnaire was developed to evaluate the opinion of the students
in relation to the training received in both courses. The feedback obtained was used as
a base both to validate the flowchart proposed and to improve future sessions and, at
the same time, to look for better performance [17, 18].
The paper’s main objective is the description of a flowchart of the
educational process related to Supercomputing. The secondary objectives are: (1) to
study the main application fields identified in the review of the literature for both
STEM and non-STEM or non-computing students in High Education Institutions (HEI) where
Supercomputers are used; (2) to study the types of computational courses, related to
Supercomputing, identified through the review of the literature; (3) to provide a
relational study juxtapositioning the field or sector that uses Supercomputing and the
type of specific computational training and (4) to analyze how the use of a
Supercomputer in a practical course, organized jointly a University and a Supercomputing
Center, permits the enhancement of training in higher education environments.
The conclusions highlight the strengths of the experiential approach taken
through the case study. It is argued that technology, in general, is a powerful way to
engage students using innovative practices, thereby improving their performance.
The outline of the paper is as follows: in Sect. 2, we briefly discuss the material and methodology followed;
Sect. 3 presents, in greater detail, a
general overview of the results obtained, with data analysis related to the use of
Supercomputers in education. In Sect. 4, a case
study of training is described. An overall discussion of the results is outlined in
Sect. 5. Finally, in Sect. 6, the conclusions are presented.
2 Materials and methodology
The scope of the study, both with regard to its technical and educational
content, is suited for a combination of methods. In this sense, the design of the
targeted process combines theory with the findings of the literature review and, at the
same time, is heavily based on the case study method.
Firstly, the base of the population studied was the one used in
[22], which follows guides related to
literature reviews [23], with special
attention to systematic reviews of the literature and based on a taxonomy about
bibliographic reviews in education [15]. It
is considered as true hypothesis-based research, in which the studies are selected and
combined through the use of a predefined protocol that reduces subjectivity and the
possibility of bias of the researcher [6,
30]. The items of the study were
filtered based on a review of relevant journals, conferences and books, focusing on the
analysis of terms related to training, Supercomputing and technological subjects in
various sectors. The databases consulted considered to be the main repositories for
searches on computer science are IEEExplore Digital Library, ACM Digital Library,
Elsevier ScienceDirect, Google Scholar and Science Web. As a result of the first phase
of the search, 1911 works were extracted. Once some inclusion/exclusion criteria were
applied, 128 items were considered as relevant. In the second phase, 8 new works were
considered, bringing the items up to 136. At the end of this phase 2, quality assessment
criteria were applied, and the final number of items selected was 34.
Figure 1 describes the process.
×
The methodological framework of the case study is based on an approach
related to Supercomputing courses carried out in collaboration between a University
(University of Leon, Spain) and a Supercomputing Center (SCAYLE). The use of a case
study design is used to describe and explain what experiences based on field research
are like, in order to establish their relationship with their real-life context
[51]. In general, the case study results
offer insights and expand the reader’s experiences [34] in their construction of knowledge [40].
The courses included in the case study enhance computer competency and
improve teaching and learning processes, introducing students to the use of several
technological tools and technology integration practices, to effectively use
supercomputers.
In order to provide a better quality, the implementation of both courses
follows the National Educational Technology Standards for Teachers (NETS) from the
International Society for Technology in Education (ISTE). The ISTE (2008) standards were
categorized as follows:
(a)
facilitate and inspire student learning and creativity,
(b)
design and develop digital age learning experiences and
assessments,
(c)
model digital age work and learning,
(d)
promote and model digital citizenship and responsibility,
and
(e)
engage professional growth and leadership.
In this article, the interest lies in how experiences provided by
supercomputing training improves the performance of higher education students, instead
of analyzing deeper technical aspects in relation of HPC.
3 Results
As mentioned above, the main objective of this paper is to present the
description of a flowchart of the educational process related to Supercomputing, whose
definition was completed once some partial objectives of this first study were
developed: (1) the study of the main application fields in which students use
Supercomputers; (2) the identification of the type of computational courses related to
Supercomputing in the literature review and (3) the relational study of the field or
sector that uses Supercomputing and the type of specific computational training. In the
next section, the fourth secondary objective, the analysis based on the case study of a
practical course, is described.
Figure 2 describes a flowchart in
which the development of a Supercomputing course is analyzed, from the previous analysis
of the students’ profile to its closure. The process is based on the description of the
tasks carried out by the entities that have collaborated in the case study presented in
this work, which is closely linked to the findings of the literature review. The first
step of the process, once the student has been registered, is to check their level of
prior knowledge, especially in the use of Unix-like operating systems such as Linux. In
the event that the student does not possess the minimum level of necessary knowledge, he
will be prohibited from completing of the course. Once the course begins, special
attention will be paid to the practical cases, since students seek to know how to apply
Supercomputing in the solution of their tasks. In this process, feedback on students’
performance is provided in order to improve course development, and a digital repository
stores the cases to be used as examples in future training sessions. Before the end of
the course, the assessment of students on the different topics discussed is analyzed
and, if necessary, aspects that have not been clarified are repeated. When the teachers
consider that the students have understood all theoretical and practical concepts, the
course ends, offering the possibility of repeating some sessions.
×
The next sections will describe the solution proposed for every partial
objective, described previously:
3.1 Application fields
In order to better understand the details of the studies of each field
analyzed in the review, the International Nomenclature of Unesco was used to group
the most relevant fields of knowledge in the use of Supercomputers, according to each
article where the field is referenced.
Table 1 shows the fields that
use Supercomputers found in the literature review. The column “Category” shows a
total of 52 references found related to fields of knowledge. In the column
“Specialty”, a more specific reference of the type of activity inside every category
is included.
Table 1
Fields of knowledge using Supercomputers
Category
Specialty
Number of articles
Mathematics (4)
Mathematics and numerical analysis
1
Control problems
1
Automatic control
1
Design automation
1
Astronomy and astrophysics (3)
Astrophysics
1
Exploration of space
1
Aerospace engineering
1
Physical (8)
Physical phenomena
1
Physical processes
3
Thermal and fluid dynamics
4
Chemistry (6)
Computational chemistry, chemical engineering and
structural chemistry
6
Life sciences (8)
Biology
3
Bio-informatics
2
Genomics and proteomics
2
Molecular simulation
1
Earth and space sciences (9)
Meteorology and weather forecasting
5
Material science
3
Prediction for forest re-propagation
1
Science of technology (12)
Image processing
2
Drug design
2
Construction of vehicles (land/amphibious/aerial):
structural design, engine design, financial modeling and crash
tests
1
Engineering
2
Microelectronics
2
Sensor technology
1
Nanotechnology
1
Media
1
Economic sciences (1)
Economics and finance
1
Linguistics (1)
Linguistics
1
3.2 Type of computational courses related to Supercomputing
In the scope of the same review mentioned before, an analysis of the
computational courses related to Supercomputing was conducted.
As seen in Table 2, the column
“Category” shows a total of 97 references detected through the review of the
literature, related to the type of computational course identified. In the column
“Specialty”, a more specific reference to the content of the course is
described.
Table 2
Computational courses related to Supercomputing
Category
Specialty
Number of articles
Programming (16)
Parallel programming
10
Multiple-device programming
1
Programming techniques for supercomputers
1
Parallel and distributed programming
2
Multicore programming and performance
evaluation
1
Code optimization
1
Computing (17)
Cluster computing
3
Cloud computing
2
Multicore, multicore computing and multicore
clustering
1
Parallel computing
3
High-performance parallel computing
2
Parallel and distributed computing
2
HPC technology (hardware)
1
Scalable computing
1
Matrix calculation
1
Heterogeneous computing
1
Software (6)
Applications
1
User developer
1
Software engineering development tools
1
Software engineering
1
Software design for complex and parallel
systems
1
Software parallelism
1
Systems and architectures (17)
Architecture of parallel systems
8
Distributed and parallel system
4
System administration
1
Parallel systems
1
High-performance computing systems
1
Scientific computer systems
1
Operating systems
1
Data management (14)
Visualization of scientific data and mathematical
tools
4
Distributed and parallel database
1
Workflow management
1
Data analysis/post-processing
1
Data partitioning
1
Data extraction
1
Distributed and parallel databases
2
Big data
2
Others (28)
Parallel algorithms
6
Optimization
1
Cluster and grid middleware
1
Network technologies
1
Modeling and simulation
6
Operational issues
1
Performance analysis
1
Verification and validation
1
Domains
1
HPC application packages
1
Scientific resources
1
Parallel libraries
1
Scientific applications of parallelism
1
Load balancing
1
Checkpoints
1
Introduction to accelerators
1
Network infrastructure
1
Parallel processing
1
3.3 Relational study comparing the field of knowledge that uses Supercomputing and
the type of specific computational training
The relational study was based on the identification of 12 articles
that define together, a field of knowledge and a type of computer course. The initial
number of papers selected was reduced, because not all articles mentioned data of
both variables at the same time. To assess the relationship between the variable
“field of knowledge” and the variable “type of training”, The UNESCO criteria have
been followed to define two dimensions: Sciences (including all the science-related
subject and those related to economics and linguistics) and Engineering (includes
articles related to the sub-discipline of Mathematics, Astronomy and Astrophysics,
Chemistry and Physics). In order to achieve the proposed objectives, the construction
of two multiple regression models has been carried out, with the dependent fields
being the Sciences and Engineering fields mentioned above. In all the cases, the
independent variables were the different types of computational courses developed,
described in Table 2.
It is necessary to emphasize that in the relationships sought for the
development of the models, the number of references, or frequency in which the
variables are mentioned in the final selection of articles detailed in the index, is
analyzed as reflected in Tables 1 and
2. The final number of articles to be used
in the study was 12, due to the coincidences of references, both in the field of
knowledge, as in the type of course mentioned.
The results obtained in relation to the parameters of each model, with
a level of significance of 95%, are shown in Table 3.
Table 3
Models created according to the criterion of dependent variable
“field of knowledge”
Model
Criteria
Predictors
Coefficients (β)
Typical error
Statistic t
Model 1
Sciences
Interception
0
N/A
N/A
Programming
2.34
0.94
2.49
Computing
− 0.10
0.75
− 0.14
Software
− 0.24
2.56
− 0.09
Systems and architectures
− 0.68
0.77
− 0.89
Data management
− 1.31
1.34
0.98
Others
0.16
0.51
0.31
Model 2
Engineering
Interception
0
N/A
N/A
Programming
0.01
1.09
0.02
Computing
0.32
0.87
0.36
Software
− 0.33
2.97
− 0.11
Systems and Architectures
0.04
0.89
0.05
Data management
0.55
1.54
0.35
Others
0.79
0.59
1.32
In the case of training in the field of Sciences (Model 1), the
parameter of training in Programming is the one that is positively related in greater
proportion to explain the increase in training of computational courses. In Model 2,
the parameter that best explains the increase in the number of courses in the
Engineering is that related to Others (referring to algorithms, load balance,
simulations, etc.).
In order to evaluate the validity and the degree of correlation that
exists in the models, as well as their statistical significance, the analysis of
variances (ANOVA) is shown in Table 4. The
correlation coefficients between the dependent and independent variables of each
model in Table 5 show the percentage for
which the variables explain each other.
Table 4
Details of ANOVA results for each model
Degrees of freedom
Sum of squares
Average of squares
F
Critical value of F
Model 1
Regression
6
60.09
10.02
6.75
0.0268
Waste
6
8.91
1.48
Total
12
69
Model 2
Regression
6
21.08
3.51
1.76
0.274
Waste
6
11.92
1.99
Total
12
33
Table 5
Details of the correlation of the models
Regression statistics
Model 1
Sciences
Model 2
Engineering
Multiple correlation coefficient
0.93
0.80
Coefficient of determination R2
0.87
0.64
R2 adjusted
0.60
0.17
Typical error
1.21
1.41
Observations
12
12
From the data presented above, it can be observed that only Model 1
satisfies the condition of critical value F being less than 0.05 and, therefore, this
would be the only model of the two proposed in which the null hypothesis would be
rejected. Therefore, Model 1 can be globally considered a valid model.
Table 5 analyzes the adjusted
R2 coefficient to compare which of the proposed models best explains the correlation
among variables. In Model 1, where the dependent variable is the “Sciences” field, it
has a higher correlation percentage, which explains the degree of variation of the
dependent variable as a function of the independent variables selected. The
corresponding value of Model 2 is below 0.5 in the lower scale of the parameters that
measure this coefficient, and therefore does not adequately explain the correlation
as it happens in Model 1.
4 Case study
This section describes the case study based on the courses completed by
the students related to Supercomputing. Courses analyzed were organized by the
University of Leon and the Supercomputing Center of Castile and Leon (SCAYLE), and held
during 2016. The objectives in both cases were to provide fundamental and advanced
concepts by using very large database systems in a Supercomputer. Courses can be
classified in the “Specific-domain courses” category. The analyzed courses are in the
field of biotechnology, due to the fact that of those related to Supercomputing the
courses from this field, developed by the entities involved, held the majority.
In the description of the case study, a prior analysis is of great
importance, since, as described in this study, practically all of the findings of the
literature review are applicable to the development of the training actions
described.
The names of the courses were: ‘‘Practical Course of Introduction to the
use of Supercomputing applied to the analysis of RNA-Seq data—2nd Edition’’ (C1) and
“Practical Course on the use of Supercomputing applied to metagenomics and comparative
genomics” (C2). The most relevant characteristics of the courses are summarized in
Table 6.
Table 6
Characterization of courses (significant/meaningful
data)
C1
C2
Organizer
University of Leon
Supercomputing Center of Castile and Leon
Date
July 2016
October 2016
Number of students
16
14
Course duration
28 h
Previous Seminar of Linux for better use of Supercomputing
resources
Yes
Number of students attending the previous
seminar
16 (100%)
12 (86%)
Participants of the course
Researchers and/or higher education students interested in
genomic studies, Computer Science professionals, Biologists and/or
Biotechnologists related to genetic diagnosis and postgraduate
University Students (with technical qualifications in the experimental
field) and, in general, any person related to the topic of research,
innovation and development
Objectives
Managing and interpretation of data from global gene
expression of Next Generation Sequencing using a
Supercomputer
Training programs in Supercomputing are normally conducted by highly
qualified specialists with broad expertise in HPC systems and applications. They usually
also have skills for the use of Supercomputers for a wide range of computer science
issues. These specialists teach how to optimize the software used to work efficiently
with HPC technologies. During the development of the courses, new teaching strategies
were important in order to adopt technology based on Supercomputing. The new strategies
have to enhance the pedagogical processes and ensure that the educational approach has
been correctly applied while allowing personalization of the learning plans. They also
have to be adapted, depending on previous knowledge and individual learning goals,
described in a system in the literature where students come from a variety of
backgrounds, and specific weaknesses [4].
For better organization and quality, courses were designed, developed and tested using
the infrastructure of a real supercomputer cluster and a software appropriately
programmed and optimized to generate quick results to solve complex problems.
The analysis of the graphics shown in Figs. 3 and 4 provides information
on how the supercomputing infrastructure was used during the course. In
Fig. 3, we can see the diagram of how the
Supercomputer was used in C1 and C2 sessions. Once students complete the theoretical
phase, they begin to test practical cases, related to the specialization of the course,
sending jobs to the Supercomputer. They have an immediate feedback of the simulation
they run, and they can directly see directly the quality and efficiency of the use of
these infrastructures in the optimization of the performance of a process, enhancing the
learning.
×
×
As a basic part of the courses, students run computational simulations
using real data. A graphic tool is in charge of controlling the load of the
infrastructure in order to provide the best efficiency for the users (Fig. 4). With the support of the Supercomputer, the graph shows
a peak marked with a red circle, which represents the operations per second in the
storage system. The highlighted detail corresponds to the exact moment when this data
was processed in the practical sessions, and shows the peak of data access when students
access data stored on disk to begin calculations. It is made from SCAYLE’s monitoring
systems (Grafana, see web https://grafana.com). The screen image at the bottom shows a complete list of the jobs that
were running on the Supercomputer at the same moment, highlighting the servers that were
assigned to the course in the Cluster Queue manager. This is obtained from the SGE job
manager qstat command, which displayed all created queues and their status (available
cores, total cores, used cores, average load, etc.).
4.1 Participants and data collection
Both courses were designed for students with diverse backgrounds, and
are usually made up of 12 to 20 participants at the most, in order to enable a more
personalized education for students. The number and characteristics of the students
are shown in Table 7 below.
Table 7
Students’ profile
Student profile
C1
C2
Doctorate with experience
2
2
Post-doctorate
2
2
Researcher
11
10
Master student
1
A key step in every research is to be clear about its purpose and
scope. The implementation of methods for data collection and analysis are essential
for all types of evaluations. Data collection is an iterative process based on the
search strategies of the study. In the case of this paper, in order to complete the
quantitative analysis for obtaining information to analyze the results, a
questionnaire consisting of ten items which mixed Likert-type, closed- and open-ended
questions, was provided to students at the end of the sessions in both
courses.
The questionnaire reliability was tested previously in SCAYLE, where
this tool is used as the standard for evaluating the quality of the training provided
to the students in their own training programs. This questionnaire provides, through
the use of specific key evaluation questions, the evidence needed to make appropriate
judgements about the quality of the courses. The consistency of the items of the
questionnaire was evaluated by experts who also contributed to the design and
adaptation of data collection instruments.
4.2 Survey results
The evaluation questionnaire includes a first part that evaluates the
introductory aspects of the course, a second part requiring an evaluation of the
teachers and, in the last part, a final consideration of the whole course.
Table 8 includes the full text of
questions that were asked in the survey, and displays the results of the students who
answered the survey. Items can be classified into three groups: (i) introductory
aspects (Q1 to Q5); (ii) assessment of the teachers (Q6 and Q7); (iii) overall
satisfaction and final considerations (Q8 to Q10).
Table 8
Results of the survey
Questions
C1–N = 15
C2–N = 13
Q1
Was the introductory seminar to Linux useful for
you?
Yes: 15
Yes: 13
Q2
Average assessment of the introduction seminar teacher,
biotechnologist profile (1–5)
4.62
4.92
Q3
Average assessment of the introductory seminar teacher,
computer engineer profile (1–5)
4.60
4.40
Q4
Do you consider the teaching material as
adequate?
Yes: 12
Yes: 11
Unanswered: 3
Unanswered: 2
Q5
Has the course been adequately disseminated?
Yes: 14
Yes: 13
No: 1
No: 0
Q6
Would you change anything about the seminar?
No: 5
No: 8
Unanswered: 5
Unanswered: 4
Reduce the technical part of the startup
seminar
1
1
Include more practical aspects in the
sessions
3
Take into account that students have low computer
skills
1
Lower the level of the course
1
Go further and deeper in course content
3
Q7
Average evaluation of sessions of research professors
(1–5)
3.59
4.46
Q8
Did the course cover your expectations?
Yes: 12
Yes: 10
Unanswered: 3
Unanswered 3
Q9
Has the course methodology been attractive?
Yes: 12
Yes: 10
Unanswered: 3
Unanswered: 3
Q10
Average of overall assessment of the course
(1–10)
8.83
9.69
The analysis of the responses of every group mentioned above is the
following:
Introductory aspects The introductory
seminar was considered very useful by all the students in both courses, so
it is essential to remedy the lack of knowledge regarding aspects needed to
use a Supercomputer. In this connection, the mark of both biotechnologist
and computer engineer profile teachers of the introductory seminars was very
high, with marks close to the maximum. Finally, most of the students
consider that the other introductory aspects, such as the material provided
in the course and the dissemination, were adequate.
Assessment of the teachers In order to
ascertain the student’s perception of the courses, we asked question Q6,
relating to possible changes in the conduct of the course. The number of
unanswered questions was high, but in any case, a more practical content and
further extension and depth were proposed. In the case of the evaluation of
the specialized teachers (Q7), we can see that teachers of course C2 were
evaluated better, probably because C1 was a very intensive course, with the
same number of sessions and double the number of teachers, making the
teaching task more difficult.
Overall satisfaction and final
considerations Questions Q8 to Q10 provide the final
assessment of the course. Most of the students consider that the course met
their expectations (Q8) and that the method of teaching was attractive (Q9).
The final assessment of both courses was very high: 8.83 for C1 and 9.69 for
C2.
5 Discussion
This study first analyzes the results obtained in the literature review
and, later, those of the case study. At the end of this section, information regarding
the relationship of both studies will be provided.
5.1 Supercomputing in the educational process
The increase in supercomputing infrastructures has paralleled the
complexity of research and the problems facing today’s society and has been developed
to obtain better results in various fields of knowledge and sectors of activity.
Currently, numerical simulation for obtaining results of research is widely
practiced, both in the academic arena and in industry in general, to solve problems
in a short period of time. For this reason, training in these technologies is
essential for an adequate performance of professionals working in diverse types of
fields and activities.
For the consolidation of Supercomputing in the educational process, it
is essential to invest in the training of competent personnel in the proper use of
Supercomputers [16]. Supercomputing
training improves the results of the STEM and non-STEM, or non-computing students, in
a Higher Education Institution (HEI), among others. With the help of many tools such
as new and easy-to-use software computer simulation is made accessible to a wider
group of people.
For implementing Supercomputing, the need of cooperation between
professionals of multiple skills and of different backgrounds—an important added
value—should be based on two interrelated aspects: (1) different profiles of a
variety of fields of knowledge in different competences in STEM and non-STEM,
non-STEM students; (2) different experiences in the integration of technology in the
Higher Education system.
Our findings suggest, through the results seen in the literature
review, that including Supercomputers in the learning processes of a great variety of
fields has provided many opportunities for learners.
5.1.1 Outcomes
Many factors, as constraints in the curriculum, such as the rapid
changes in the field of Supercomputing in general, or the sophistication of the
mathematical models, influence the use of supercomputing facilities.
This study provides three main outcomes related to state-of-the-art
Supercomputing training, which respond to the objectives raised at the beginning
of this study and which are summarized below.
Objective 1: through the analysis of the 52 articles of
dealing with different fields related to the use of Supercomputers, an
important number of fields that are considered non-STEM or non-computing
were detected, such as a wide number of specialties, that are the basis
of many industrial processes.
Objective 2: it was found that many computational courses
are described in the literature, which influence activities related to
Supercomputing. These courses clearly enhance the performance of the
users of these infrastructures, by improving the processes on which
research is based, allowing a better interpretation of the
results.
Objective 3: according to the relational study, Model 1,
linked to the items related to the field of Science, show that
programming is the variable close related to this field of
knowledge.
5.1.2 Limitations
During the course of the literature review, the greatest difficulty
encountered was the heterogeneity in the type of articles, and also in the content
related with the typology of training. Another limitation of the study was the
fact that, due to the nature of the information, it did not focus on analyzing
training related to Supercomputing as an element of the teaching–learning process,
so it would be necessary to complement the study with others related to the use of
high-performance computers as new didactic tools.
Finally, another limitation is related to the methodology used in
the regression models, which is based on a correlational approach. The results
achieved thereby allow us to speak of a certain relationship between the field of
knowledge and the type of specialized computational course, but which does not
necessarily mean a causality between these variables. For further development of
this research, studies based on experimental designs will be necessary to confirm
the effect of independent variables on the development of computational training
in the fields of knowledge.
Our hope is also that the present study will encourage researchers
and institutions in general to contribute to the development of agreements which
address the challenges of using a Supercomputer as a fundamental part of the
research of many fields, through the use of a real Supercomputing infrastructure
and the inclusion in the academic curriculum of both STEM and non-STEM students
from subjects related to this issue [35].
5.2 Case study
5.2.1 Outcomes
This study of cases gives an opportunity to examine how the
procedures of training in a technological environment provide and facilitate the
engagement of the students in the improvement of their performance [14]. The issues raised in this part of the
paper should be relevant in order to provide new tools for higher education
students, thereby complementing their skills in the use of technology. We
highlight that, in spite of some cases described in which students participate in
curriculum design [13], currently,
contents in HPC are practically testimonial in the academic curriculum of higher
education [26, 37, 41].
Supercomputing training appeared to support student’s performance
and improve learning activities by providing a platform that helps in the pursuit
of real cases in many fields. In the future, the design of technologically
advanced courses should be conducted with the use of a Supercomputer, thus moving
forward with regard to the training offered and following new pedagogical
methods.
However, although these courses have, in general, received favorable
comments and positive assessment, some students expressed the need for some
aspects to be improved in the future, such as the lack of more time for practical
sessions and deeper explanations. Additionally, the high level of the course in
terms of specific computational skills presumed requires attention, because some
students declared having a low-level computational background using a
Supercomputer in an optimal manner. The most important questions that can be
considered for future courses are the following:
Students consider the support of the introductory seminar
essential. In some open questions they even express the belief that they
would need more training in some areas, prior to the beginning of the
main content of the course in order to optimize their performance.
Increasing the support of the technicians of Supercomputing
Centers, in order to improve and expand the capacity for using the
infrastructures to test the practical aspects of the courses—something
highly demanded by students.
Enhancing planned didactic aspects of the classes, to
ensure the best performance possible from teachers.
Implementing the design of an easy-to-use interface needs
to be carried out, to facilitate human–computer interaction with an
intuitive experience during the use of a Supercomputer.
5.2.2 Limitations
This study has certain limitations that should be considered when
interpreting its results and conclusions. Firstly, a larger number of participants
would be more suitable, because, due to the relatively small number of
participants of both courses (30, of which 28 responded to the survey), results
should be interpreted with caution. Secondly, the courses considered in the study
were taught by different instructors, in specific areas, and the profiles of the
students were also different, which complicates the study of a homogeneous
population. In order to minimize the effects of these limitations, similar levels
of education and experience of the teachers, and a close monitoring of the
implementation of the content in both courses, would be essential.
In the development of the work, it has been possible to verify how
most of the aspects analyzed in the literature review have been observed in the
courses analyzed in the case study. Thus, aspects such as lack of prior knowledge
in the use of a Supercomputer, the notable improvement in the qualification of the
students, or the effective availability of supercomputing infrastructures to carry
out practical cases on simulating reality, have made it possible to carry out
actions to improve the training offered.
6 Conclusions
The paper shows how the implementation of an innovative pedagogy for
teaching STEM-related subjects for non-STEM and/or non-computing students is very
important in today’s society. There is a strong potential of expansion for
interdisciplinary studies and addressing other major questions all through the networks
and connections to be created by highly qualified personnel involved. A step forward in
the integration of Supercomputing in STEM and non-STEM fields of knowledge through a
change in policies might lead to a wider innovation in the traditional way of the
educational process: improving teaching and learning.
Currently, the use of software tools for the deployment of theoretical
methods and data analysis used for mathematical modeling, and techniques of
computational simulation, are wide extended for the study of fields such as biological
systems, through high-performance computing [46]. This will create new learning environments which allow the best
use possible of the available technologies, showing a way for the improvement in the
offer of practical sessions in the programs in higher education.
In the courses related to supercomputing, teachers work in a
technology-oriented manner based on authentic situations and problems that illustrate
the potential of HPC technologies in learning practices. In order to reach the goal
proposed, the implementation of the courses should follow these principles: (1)
organization of an introductory seminar that contributes to a better performance for the
rest of the course; (2) support of specialized technicians and on-site Supercomputing
infrastructures, providing an optimal use of the computational resources for running
simulations as well as testing the capacity for solving practical problems, and (3)
instructions provided by highly qualified experts.
We can conclude that HPC technologies are good pedagogical tools, which
allow students to learn in different ways, in the most diverse fields of knowledge,
considering that training in Computer and Information Sciences (CIS) is largely
influenced by the way big data scientists are trained through the use of machine
learning and artificial intelligence techniques [49]. Currently, deep learning is also described as a new philosophy of
training based on technology [5]. By
changing the way Computer Science is advertised and taught, a more inclusive Computer
Science is possible [2].
Related to the support provided by the Supercomputing Center in the
design, implementation and monitoring of the training offered to students, we can
conclude that it contributes to the training sessions in the following positive
aspects:
By using the Supercomputing Center infrastructures for training,
the work of the students is constantly monitored. This way, the use of the
resources can be optimized, providing a higher computational capacity if
necessary.
Continuous support by technicians, experts in HPC, covering
several areas of Supercomputer use, solving problems and providing a high level
of security is assured.
Software involved in computational simulations is being tested
permanently in order to offer more reliable results.
Students carry out their software assignments with the help of
specialized supercomputing technicians, resulting in better performance.
The use of algorithms and flowcharts for the identification of students’
needs is extended, allowing the use of diagnosis decision trees through the use of a
distributed architecture that combines shared-data and client–server styles
[28]. Currently, algorithms that provide
an optimized environment for solving large scale optimization problems, helping
researchers in the application of evolutionary algorithms for a wide range of fields
[36], are commonly used. In this light,
the flowchart described in this paper shows a successfully tested way of conducting
training sessions related to Supercomputing based on the experience of the cases
studied.
Finally, future work should be focused, comprehensively and
simultaneously, on analyzing new methods and tools to evaluate the quality of
Supercomputing training in order to improve it, constantly adapted to the needs of
students in every moment. Accordingly, it is important to explore the requirements of
all the fields the use of Supercomputers could be helpful, trying to find proper ways of
promoting such training in students of higher education.
Acknowledgements
This study was supported by EU Commission (contract 2018-1-ES01-KA201-05093);
Ministry of Science, Unversities and Innovation of the Spanish Kingdom (grant
RTI2018-100683-B-I00); Ministerio de Economía y Competitividad (ES) (research project
ECO2015-63880-R); Fundación Centro de Supercomputación de Castilla y León.
Open AccessThis article is licensed under a Creative
Commons Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you
give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons licence, and indicate if changes were made. The images or other
third party material in this article are included in the article's Creative Commons
licence, unless indicated otherwise in a credit line to the material. If material is
not included in the article's Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to
obtain permission directly from the copyright holder. To view a copy of this licence,
visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.