key: cord-0045952-iemzdfyf authors: Maxville, Valerie; Sandford, Brodie title: Computational Science vs. Zombies date: 2020-05-25 journal: Computational Science - ICCS 2020 DOI: 10.1007/978-3-030-50436-6_46 sha: d722d0a0f79c73db88d041e70323f05cff8e4a4d doc_id: 45952 cord_uid: iemzdfyf Computational Science attempts to solve scientific problems through the design and application of mathematical models. Researchers and research teams need domain knowledge, along with skills in computing and, increasingly, data science expertise. We have been working to draw students into STEM, Computational Science and Data Science through our Team Zombie outreach program. The program leads the students through a scenario of a disease outbreak in their local area, which turns out to a potential zombie apocalypse. They become part of Team Zombie, a multi-disciplinary response team that investigates the outbreak; models the spread and potential interventions; works towards cures or vaccines; and provides options for detection and monitoring. Throughout the activities we make reference to real-world situations where the techniques are applied. Our program has engaged students from primary school through to university level, raising awareness of the range of approaches to problem solving through models and simulations. We hope this will inspire students to choose courses and careers in computational and data science. Around the world, educators are working to boost student interest in Science, Technology, Engineering and Maths (STEM). According to Scitech, 'As we shift to an information-based, highly connected and technologically advanced society, STEM can empower individuals and communities with the problem-solving capabilities to meet unprecedented economic, social and environmental challenges' [1] . In addition to developing interest in the STEM subjects themselves, we can inspire students by exploring scenarios that require a combination of STEM knowledge and techniques. Computational and data science provide many vibrant examples of STEM. Combining one or more science disciplines with computer simulations and data analytics can give interactive and realistic experiences. A key requirement for outreach activities is to be able to easily explain the context and be inclusive of a wide range of ages and backgrounds. To give better reuse and flexibility, we also look to allow for hiding or revealing details, and moving through a range of models/solutions. An important point to get across is that we are working with models, which will always be wrong [2] . Inspired by The Shodor Foundation's [3] resources for mathematics and science education through the application of modeling and simulation technologies, we have previously developed outreach activities exploring supercomputing and simulation [4] . In this paper we discuss the development and results of a story/scenario-based approach where various techniques are applied through the investigation of a zombie apocalypse. The initial simulation that inspired this work was an assignment set in Fundamentals of Programming, a Python programming course developed for Science and Engineering students at Curtin University [5] . The assignment gave students a simple model of disease spreading through a population, which they then had to extend to incorporate barriers, immunity/recovery and airports. Discussions around the assignment often referred to it as modelling the zombie apocalypse. The additional engagement this elicited was a sign of the potential of the scenario for outreach purposes. To give authenticity to the resources, we considered real epidemic monitoring and policy. Table 1 provides an example of the steps taken when investigating a disease outbreak. Our scenario begins by going through the steps required to identify and characterise a disease outbreak. Table 1 . Epidemiologic steps of an outbreak investigation [7] . Step Description Steps 1-4 would be done as preparatory work, looking at health, media and other reports. This is the settling in part of the outreach activity. Once the zombie diagnosis is established, we need more specific information about the type of zombie outbreak we're looking at (step 5-6): Depending on the time available, we might have an activity around researching the various types of zombies in movies and literature. From our research, these are some of the known Zombie types (transmission and behaviour): 1. Viral: An airborne virus infects the entire population. When people die, they re-animate four hours later as zombies. These zombies are gradually decomposing. They can be killed by destroying their brain. With the infection information, and an understanding of the type of zombie we are dealing with, it's time to develop and validate models of the outbreak (steps 7-10). These can extend on the models used in the assignment, described earlier. Steps 11-12 look at the response and monitoring of the situation, while the final step looks at communicating the plans and progress of the response. The steps may be carried out in parallel, and may have feedback loops to update and reassess based on additional information throughout the investigation [7] . The context for the outreach tasks is given to the students by memos, briefings and/or developed websites. After consideration of the various computational tools that could be developed for battling the Zombie Apocalypse, we focused on three key demonstrations: -Zombie Outbreak -modelling the spread of the disease -Zombie Detector -application of computer vision and machine learning to automatically identify zombies -Zombie Vaccine -once a potential vaccine has been identified and developed, modelling its dosage and distribution We will now describe the approach and implementation for each of the demonstrations. As the purpose of the model is to communicate the concepts around the spread of disease, we focused on the visual and interactive elements of the implementation. The Python simulation begins with a healthy population in a 2D grid. We use the PyGame package to provide an interactive interface where the user can click on individuals to infect them. The disease spreads via a set probability of infecting Healthy and infected cells/individuals are indicated by colour (green/red in Fig. 2) . A third option is a proportion of the population having immunity (yellow) which contains the spread of disease. Users can also enter a value to speed up the simulation. Possibly the most compelling part of the application is the underlying map of Perth, which localises the scenario and caused much excitement. Also included are natural and man-made barriers such as the river/ocean and bridges. Users can click on the bridges to break them down, allowing for isolation of the infected population. An important step in controlling a disease outbreak is the detection of infected people, preferably using a remote, automated system. When discussing zombies, a range of potential traits were considered: -Speed of movement -slowness, gait -Intelligence, or lack of it -Odour -decomposing flesh -Temperature -using thermal images -Appearance -e.g. missing limbs, expression Although temperature may have been the preferred approach, we chose to work with appearance as it would be engaging in an interactive display. To show the potential of computer vision, the premise was that zombies could be identified by their facial expression. We found no record of zombies smiling, so this could be a clear visual difference between healthy people and zombies. A Python program was developed to work with real-time camera footage to detect faces and flash up a warning if a face is determined to be a zombie (see Fig. 3 ). Packages used included: sklearn, open cv (cv2), dlib, numpy and csv. The implementation finds a face using HAAR Cascades [8] and puts a bounding box on the face. Then the system identifies 68 facial landmarks, and the positioning of the landmarks is used to determine if the face is human or zombie, based on a training data set. The third demonstration assumes that a vaccine or medication can be developed to protect healthy humans. Once we have a vaccine, we need to calculate how much of it is required, and how often, for there to be effective protection. With this individual information, we can then consider how many people can be saved, how much medication to produce, and whether there are difficult decision to be made due to limitations in supply. The demonstration used for communicating these concepts is based on the one-compartment model with repeat dosages described in [9] . This model has been used each semester in Fundamentals of Programming to explore simulations and parameter sweeps. In this case, we limit the interaction to adjusting key parameters to vary the half-life, absorption, dose and interval between doses. The diagram in Fig. 4 show the display of the demonstration code, which we are running in a Jupyter notebook (via Google Colab). In this case, the students see and interact with the code directly. The developed materials have been used in two outreach events: a local science festival and a multi-day immersive science experience. This family-friendly event took place at the Claremont Showgrounds on the 24th and 25th August. It is the largest event on the WA National Science Week calendar, with over fifty exhibitors and an attendance of 8,161 people (2018 figures). In pitching for a booth, we needed to give the focus and story for our booth (see Table 2 ). Ahead of the event, local media contacted us for an interview to help promote the booth and the science event. The resulting article [10] used the Team Zombie booth as the lead story to promote the Festival. The booth for Team Zombie was a standard 3 × 3 m setup with two tables. We brought multiple computers, explanatory material and zombie decorations (see Fig. 5 ). It was manned by Computer Science and Data Science students 10am-5pm for two days. There was an excellent response from kids, parents and teachers who enjoyed the interactive displays. We were asked about school visits, our website (in progress) and possible mobile phone applications. In terms of results and effectiveness, the attendees spoke with their feet. We had continual questions and interactions for the full day, often with people waiting to have a turn at the each of the demonstrations. Some children and parents kept coming back to chat, or to explore a different aspect of the simulations. We found they also built on the concepts in our materials, discussing herd immunity and seeing the relevance to current issues such as measles outbreaks and global epidemics. This initiative has been run by Curtin Outreach for over 15 years. The fourday event has a program of science activities for Year 9 and 10 students. The premise is the give students with an interest in science a chance to participate in a range of science activities, with guidance from scientists and engineers who can share their love for their work. We were asked by the organisers to showcase Technology/Data as a link to the State Government's priority areas. We were assigned a two hour timeslot with up to fifty students attending. The large group was split across two rooms, then into smaller teams or 4-5 to do their investigations. We scheduled the session in a collaborative learning space, giving each group a computer to share, which we prepared by pre-loading all of the software (see Fig. 6 ). The lesson plan took students through setting the context; research of zombies; discussion of transmission; interaction with Zombie Outbreak model; discussion of containment strategies; Zombie Detection and Zombie Vaccine development. Google Collab notebooks were used for interacting with the Vaccine code, while the other two demonstrations were set up to run continuously. Our temporary website [11] progressed the scenario through a series of blog posts. Each team had a Google doc to record notes on their investigations. Group B The Generic Zombie -is a person who has been killed and reanimated by a pathogen, often, but not always due to a virus -usually aggressive and curious, but disorientated, and at a loss to fully understand their environment -Their most notable trait is that they kill and eat uninfected humans Group C -Conspiracy theories -Government sent zombies to solve overpopulation -Contaminated drinks and food -Team Zombie is actually the one that is infecting everybody to gain publicity and mass chaos A selection of excerpts are reproduced in Table 3 . Note that the scenario puts forward chocolate as a potential antidote/vaccine. Overall the session kept the students engaged and, although the scenario was a bit of fun, it happened to take place when a measles epidemic was causing the deaths of many children in Samoa. By discussing the Samoan outbreak and other epidemics (e.g. Ebola, SARS and now Coronavirus) the students clearly understood the seriousness of such situations and we hope they saw the need for a diverse team from a range of disciplines, using a variety of techniques, to best meet these challenges. Ahead of the Perth Science Festival, local media contacted us for an interview to help promote the booth and the science event. Of over fifty booths at the event, the zombie theme was considered the most exciting to attract visitors. The resulting article [10] used the Team Zombie booth as the headline and lead story to promote the Festival. During the event, exhibitors for Scitech approached us to provide our 'expert' opinion on how zombies might spread across Australia. This evolved into a feature article for Particle [12] , an online magazine producing news, stories and views. Ideas explored in the article included: -Is it possible to mathematically work out how fast a zombie apocalypse would spread across the world? -If they only eat brains, would they eventually run out of food? -As an educated guess, where would it likely start? and being so isolated, does Perth/WA have an advantage? We worked through some additional models and problems, including consideration of census data (population density) to give a prediction on the hypothetical spread of zombies across Australia. We were informed that the Particle article had strong readership figures and received many queries for more information. We have described an outreach project intended to increase awareness of computational science and the multi-disciplinary teams that are required to address the challenges facing our planet. The project has been able to capture the imagination of students, teachers and parents through interaction with our models and simulations. Lengthy interactions, requests for additional resources and school visits indicate that we have succeeded in engaging our audience. The media interest supports our belief that we have a useful premise and plausible resources. Future work will be to develop an improved online presence and additional resources, including browser and/or app-based simulations, lesson plans and more detailed materials to expand on specific aspects of the scenario. Why STEM Matters Introduction to Computational Thinking Shodor: SHODOR: A National Resource for Computational Science Education Introducing: computational Science Fundamentals of programming for science and engineering Team Zombie: Zombie Identification, Activity Resources CDC: Principles of Epidemiology: Home-Self-Study Course SS1978 Face Detection using Haar Cascades Introduction to Computational Science: Modeling and Simulation for the Sciences Nerd it up this weekend in Perth zombie apocalypse simulation Team Zombie website Zombie a-Perth-calypse: a WA survival guide A squad of student volunteers joined Team Zombie to put together outreach materials and interactive displays. Our coding team were:-Zombie Outbreak -Cleye Jensen -Zombie Detector -Harry Walters -Zombie Vaccine -Brodie Sandford Cleye, Harry, Brodie, Lisa, Caitlyn and Ryan volunteered on the booth at Perth Science Festival. Team Zombie volunteers for the Science Experience were Brodie, Harry, Caitlyn, Nhan, Matthew, Cameron, Jack, Dylan, Brooklyn, Indigo, Bene and Blake.