The use of computational tools enables scientists to study systems that are too big or dangerous to experiment on in the real world. These “grand challenges” are driving scientific research and innovation in many disciplines.

Scientific models can be written down in a variety of notations including mathematical equations, diagrams and plain language. However, a model that can be executed by a computer is the most precise and unambiguous.

Theoreticians

Since electronic computers became available in the 1940s, scientists have used them to explore new territory. Experimenters sifted through large data sets for statistical patterns, and theorists used them to calculate the equations of mathematical models.

With the rise of powerful supercomputers, theorists are increasingly using computer simulation to reach places where experiment or theory cannot. Aeronautical engineers, for instance, have eliminated the need to test wings in wind tunnels by simulating the flow of air around them; astronomers use computers to predict the results of galaxy collisions; and chemists simulate how proteins fold into their native shapes.

For many scientists, it is difficult to make a distinction between “model” and “method.” A “program,” after all, is just a set of instructions for the computer, and programmers are often more concerned with obtaining fast results than with how those results were obtained. The resulting frenzy to be ever faster is beginning to give way to a more sober attitude that emphasizes reproducibility, verification and reliability as the prime values of science.

Experimenters

Using the computer to Role of Computation in Science speed up the traditional work of experimenters was a revolutionary development. But something even more remarkable happened: scientists began to use computers in new ways, designing models and simulations that allowed them to reach conclusions where previous experiments had failed.

These models can be qualitative, like biochemical pathways in a cell, or quantitative: the kind of numbers that are compared to data from observations. They can also be very large, requiring "midscale" machines with hundreds of thousands of parallel cores (like today's supercomputers).

Using the same computer to perform multiple calculations on different variables and to generate multiple results can help eliminate bias from differences in gender or other factors that might affect outcomes. However, the frenzy to become faster is gradually giving way to a more sober attitude that reinstates reliability and verifiability as the prime values of science.

Computational Designers

CAD software has digitalized drafting, making it easier for designers to produce error-free designs. But with recent advancements in processing power and technology, computational design is enabling more holistic tools to incorporate multiscale modeling, the analysis of big data for hard-to-see correlations, and better atomistic approaches for materials-by-design.

Computational strategies, like parametric and generative design, can break down a project into multiple parts and generate infinite possibilities. Applied to architecture, these processes can overcome geometric limits and provide unique solutions for a design project.

For architects, Computational Design represents a powerful tool for accelerating project delivery and establishing competitive advantage. But the most important step to implementing Computational Design is educating staff and building momentum with your team. Communicate your strategy through lunch-and-learns, an intranet or monthly update memos to create visibility and get the team excited about Computational Design. This will help make the shift smoother and ensure successful outcomes. The future is in Computational Design.

Grand Challenges

As computational modeling, simulation and data analysis become increasingly prevalent across science fields, scientists are addressing grand challenges of enormous importance that impact the world in positive ways. These include transforming our energy system, improving human health, advancing fusion and renewable energy technologies, and understanding and responding to climate change.

Scientific computation is a broad field that includes the development of algorithms and high performance computing to solve complex scientific problems. For example, aeronautical engineers use computational models to predict airflows around aircraft wings and thereby eliminate the need for wind tunnel testing, while chemists model chemical reactions to understand their physical properties, and biologists analyze DNA models and genomes in order to better understand the nature of life.

To address these and other science problems, NIST scientists need access to advanced computational resources. Research in the areas of algorithm development, high performance computing and scientific simulation helps make breakthrough discoveries that push science forward.