Impact: Advanced Computing

Humans have always guessed when forecasting the weather and climate change is making educated guesses harder to do. To date, around 5% of high-performance computing resources are dedicated to weather forecasting globally. This requires a lot of energy and time, creating many harmful emissions.

At the same time, our weather is being more unpredictable and violent because of climate change. Today’s computers are too slow to analyze all of the world’s weather data and make a prediction before an extreme event, like a hurricane.

Over time, computers have improved their handling of many variables, like pressure, humidity and cloudiness, but these gains have also increased their energy usage. It’s a double-edged sword.

Advanced computers could have two positive effects here. They could have the capability to handle even more weather variables to predict patterns while minimizing the energy and time needed to calculate weather forecasts and climate models.

Fossil fuels are commonly used as they are easy to store and deliver to where electricity is needed. However, they produce harmful emissions, so we use different electricity sources, like wind, solar and geothermal.

This energy is often far from where the energy is consumed, so it needs to be transported on a long, complex and reliable grid. Yet, these grids must balance energy loads which vary depending on how many people are using it at one time.

It becomes even more complicated if every vehicle needs to be refueled with electricity, which limits the adoption of electric mobility.

Traditional computers struggle to calculate the balance between supply and demand at different times of day, which risks brownouts.

Advanced computing could help utility providers optimise energy delivery by understanding. how to anticipate future demand for energy. This scheduling could allow millions of people can freely recharge their vehicles without impacting real-time energy availability.

Humans are living longer and encountering more complex and untreatable diseases. To date, drug discovery has helped humans live longer. Computers today can create medicines that stick or bind, to the protein causing an illness to stop it.

However, this research is limited by under-sampling and approximation. Researchers tend to focus on rare events and then make educated guesses when designing drugs to target and bind to the specific proteins that cause disease.

Also, it’s getting harder and harder to find new drugs because we have discovered so many treatments so far. The improvements to existing medicines are incremental.

Current technology is unable to design drugs on the tiniest scale, so they cannot be accurate enough to attack the illness. It’s like a key that’s too big for a lock.

With advanced computing, the hope is to create the right configuration for the medicine to bind to the protein on the smallest scale and then run many simulations very quickly until we find a design that works.

One simulation will produce the right design and then researchers can create medicines that better target illness.

The most important ingredient in farming is ammonia, which is a nitrogen-based fertilizer. These materials increase how much food a single farm can produce. Fertilizers allow us to use just 15% of the world’s total land area to feed 8 billion people, without it we’d need to use half of all ice-free continents.

However, the process of making fertilizer, Haber-Bosch, is energy and creates lots of greenhouse gases. A key part of the Haber-Bosch process is catalysts but existing computers aren’t powerful to design better catalysts so fertilizer production requires high temperatures, which consume a lot of energy.

High temperatures lower how much fertilizer is produced, so manufacturers use high pressure to boost yields. However, high pressure is also energy-intensive.

Advanced computers could simulate the reaction on a molecular level and design better catalysts, which improve the efficiency of the Haber process. These catalysts can work at lower temperatures and lower pressure, which has the double benefit of lower energy usage and better yields.

US banks use FICO scores to assess the creditworthiness of people seeking a loan, whether it’s for a home, their education or starting a business.

Typically, those with low credit scores face the highest rates for debt and these higher costs make it harder for the poorest to use debt effectively. In some circumstances, this can adversely impact minority groups.

Today we use algorithms to assess credit risk, yet classic computers can’t handle the volume of data that could be used to provide more precise analysis. Therefore, many people who could get more affordable credit currently miss out.

Using more sophisticated algorithms could help retain machine learning and create better credit risk models and more accurate credit scoring. An improvement of 1% could reduce global defaults by $17 billion.

However, better algorithms can only be created with more powerful computers that can process more data. Advanced computers could bridge this gap and improve credit scoring accuracy.

People trade more than $13 trillion worth of so-called over-the-counter derivatives. These are options and futures contracts for buying and selling anything from fruit and vegetables to company stocks and government debt.

Sometimes, there are minuscule errors in the data, which become amplified by the huge volume of trading activity. These can snowball and have large financial ramifications.

Moreover, it can take days to accurately risk assess certain financial instruments so banks must hold large capital buffers. The 20 largest global financial institutions carry $800 billion as a buffer which has a capital cost of $80 billion annually.

More accurately priced derivates would reduce portfolio losses, while quicker risk assessments would free up billions of dollars currently used as capital buffers. However, both require lots of computing power and time, which is energy intensive.

Advanced computing could be deployed to calculate expected prices and variance using historical data, which improves the accuracy of the financial data and reduces computing resources to improve energy usage.

The theory behind producing safe and mighty supplies of clean energy, such as nuclear fusion, must overcome practical engineering challenges, such as plasma stability, to encourage its use.

Today’s computers lack the power to solve difficult problems, like how to control plasma, which is created when you apply extremely high temperatures to solids and liquids like fuel. Plasma is essentially a superhot, untamed gas that reacts with most things it touches.

Advanced computing could address these hurdles and unleash the potential of sustainable, safe and abundant energy, like fusion. It is a step change from current approaches. Powerful computers could accurately predict how particles move in very hot environments, which can keep power plants safe from damage.