• PDF / 1,050,981 Bytes
• 9 Pages / 585 x 783 pts Page_size
• 85 Downloads / 215 Views

DOWNLOAD

REPORT

---

Introduction Since the mid-1990s, Pixar Animation Studios has created and released 12 full-length animated movies.1 Aficionados of computer-generated imagery (CGI) are aware that several of their movies have pushed the boundaries of what can be realistically animated. Examples include vegetation (“A Bug’s Life”), fabric (“Toy Story 2”), hair (“Monsters, Inc.”), water (“Finding Nemo”), surfaces with layers of paint and dirt (“Cars”), food (“Ratatouille”), and rust and decay (“WALL-E”).2 Interestingly, as processor speeds and the number of processors used by Pixar have increased, the time needed to render the individual frames of its movies has not decreased. This is because the animators exploit increased computational power to render more complex physics and more complex scenes in each image. One of their employees referred to this as the “Law of Constancy of Pain.”3 With continued enhancements to their algorithms and software, Pixar is moving toward the goal of enabling fast, realistic rendering of any physical phenomenon with a minimum of effort by the animator. Their algorithm developers are adept at knowing what low-level expensive details can be discarded while still producing images that “look right.” An important part of this process is comparing the animated images to real life, whether it be live video of swimming fish or human actors and their facial expressions. As fans of Pixar films know, the company’s success (nearly two dozen Academy Awards and an average gross of $600 million per film) is not simply due to their animation prowess, but to

their ability to use animation as a tool to tell an entertaining story that appeals to children and adults alike. Some parallels to computational materials science and, in particular, to the growing use of many-body potentials in atomistic modeling, are evident. Over the 30 years covered in this issue of MRS Bulletin, and leveraging the same increases in computational power available to Pixar, the scope and fidelity of atomistic materials modeling has grown by leaps and bounds. This is true both of the length and timescales accessible to simulation, as well as the complexity of the underlying physics encoded in a growing suite of empirical potentials. A key motivation for developers of new potentials has been to enable more accurate modeling of specific classes of materials such as metals, ceramics, oxides, or carbon nanotubes, often by including many-body effects. Part of the art and acumen needed in the development process is to know what physics and chemistry to include to capture the desired physical effects but also what can be excluded to enhance computational efficiency. Simulations using empirical potentials have to do more than just “look right;” quantitative accuracy is required for comparisons with experiments or to more expensive and smaller-scale quantum calculations. For simulators, the analog of the “Law of Constancy of Pain” is that while computing power has grown over time, the amount of wall-clock time available to an individual researcher on large compu