Introducing HPC-Enabled Computational Research Training in Africa: A Pilot Study

Imagine the world of scientific discovery as a massive, high-stakes kitchen. For decades, chefs (scientists) could only cook with small, personal stoves. They could make a great soup or a decent cake, but if they wanted to bake a thousand loaves of bread at once or simulate the weather patterns of an entire continent, their stoves just weren't big enough. Enter High-Performance Computing (HPC). Think of HPC not as a single super-stove, but as a gigantic, industrial-scale kitchen with hundreds of burners working in perfect sync. It allows researchers to tackle "impossible" problems, from designing new medicines to predicting climate change.

However, just having access to this industrial kitchen isn't enough. You also need to know how to use the complex, automated ovens, how to manage the massive inventory of ingredients (data), and how to work with a team of other chefs without burning the place down. This is computational research training. It's the recipe book and the apprenticeship that turns a home cook into a master of the industrial kitchen. For a long time, this training was like a secret club, mostly available in wealthy countries with their own industrial kitchens. But what happens when brilliant, hungry chefs in other parts of the world want to learn, but they don't have a kitchen to practice in, or a master chef to teach them? That is the question this paper explores.

---

The Pilot: A Cooking Class in a World Without Ovens

James Afful and his team of collaborators decided to test a new way to teach these "industrial kitchen" skills to students and early-career researchers in Africa. They called their experiment the Center for Computational Research and Education (CCRE). Their goal wasn't to prove they could train thousands of people at once; instead, they wanted to see if a specific, step-by-step model could work at all when resources are tight.

Think of their model like a leaky bucket designed to catch the most motivated learners. First, they threw a wide net using social media (LinkedIn, WhatsApp, Instagram) to find anyone interested. They found 24 applicants—a good crowd of hungry learners. But you can't teach everyone deeply with limited resources, so they filtered the group. 12 people were invited to take a mandatory "pre-cooking" course online. This was like a test to see who had the patience and internet connection to follow a recipe without a chef standing right next to them.

Only 6 of those 12 managed to finish all the required modules. These six became the final "pilot cohort." They were a diverse group: mostly undergraduates and early-career researchers from fields like engineering, physics, and biology. Crucially, five out of the six had never touched a supercomputer before. They were like people who had read about cooking but had never actually held a spatula.

The Journey: From Theory to the Real Thing

The program was designed in stages, moving from theory to practice:

1. The Prep: The six survivors took a four-module online course to learn the basics, like how to talk to a computer using text commands (Linux) and how to organize files.
2. The Mentorship: Once they passed the prep, they got paired with mentors. This wasn't just a teacher lecturing; it was a guide helping them solve real problems.
3. The Project: The real magic happened here. The six students split into two teams and built actual tools using real supercomputing power.
   • Team 1 built a "surrogate model" for a nuclear reactor safety test. Instead of running a slow, expensive simulation every time, they trained a machine-learning model to predict safety outcomes instantly. They even built a colorful, interactive website (using a tool called Streamlit) so others could use it.
   • Team 2 created an AI tool to spot brain tumors in MRI scans. They taught a computer to look at images and highlight exactly where the tumor was, using a technique called Grad-CAM to show its "thinking."

The Results: High Hopes, Real Hurdles

When the program ended, the researchers asked the participants how it went. The results were glowing. The students gave the program an average rating of 4.20 out of 5. Every single person who answered said they would definitely recommend it to a friend. They felt they had learned practical skills, connected with a community, and gained confidence.

However, the paper is very careful not to call this a "miracle cure." While the students felt great, the structural problems didn't magically disappear.

• The Internet: Three out of five students reported that their internet connection was shaky, causing them to miss parts of the class.
• The Big Barriers: Even after the training, the students still reported that they lacked access to supercomputers, didn't have local experts to mentor them, and couldn't afford the cost of cloud computing.

The paper suggests that while this training program was a fantastic "spark," it couldn't fix the "fuel" problem. You can teach someone how to drive a race car in a simulator, but if they don't have a car, a road, or a mechanic, they can't actually race. The training gave them the skills and the confidence, but the infrastructure (the computers, the mentors, the money) is still missing in their local environments.

What This Means for the Future

The authors conclude that you can't just run a one-off workshop and expect to solve the problem. Instead, they propose a four-pillar framework for the future:

1. Access: Give people a chance to touch real computers, not just read about them.
2. Preparation: Use structured online lessons to get everyone on the same page.
3. Mentored Practice: Keep the human connection; mentors are essential for keeping learners from giving up.
4. Sustainability: This is the big one. The program needs to build a community where past students become mentors for new ones, and where universities and companies partner up to provide long-term access to computing power.

In short, this paper suggests that the recipe for success in African computational research isn't just about teaching the skills; it's about building a whole ecosystem where those skills can actually be used. The pilot proved the recipe works, but the kitchen still needs to be built.