Physics of Information Group at IBM Research

Physics of Information / Quantum Information Group at IBM Research Yorktown

We are in the midst of an information revolution, so much so that even lay people know the basic facts about information—how it can be encoded in bits 0 and 1, stored, retrieved, transmitted, and processed using logic gates like AND and NOT. This revolution is based on our ability to treat information in an abstract way, largely independent of its physical embodiment, which may be as diverse as a hole in a punch card, a voltage in a wire, or the magnetization of a speck of iron oxide. The field of Information Physics treats ways in which it nevertheless fruitful to reintegrate physical laws and principles into the science of information. These include:

Thermodynamics: Processing information consumes energy and generates waste heat, and the amount turns out to depend both on hardware and the nature of the logic operations being performed. The founder of our group, the late Rolf Landauer, during his long career at IBM Research, continually emphasized the connection between information processing and physics, and discovered the connection between logical irreversibility and heat generation now known as Landauer's principle.

Quantum effects: Quantum phenomena like entanglement and interference were neglected in the classical theory of information processing developed by Shannon, Turing, von Neumann and their contemporaries. In retrospect this was a mistake. Including quantum effects, and indeed abstracting them away from any particular physical embodiment, leads to a more coherent and powerful theory of information processing, as well as making possible information-processing feats unachievable with conventional “classical” information, notably quantum cryptography and quantum computational speedups. In place of bits the new quantum information theory has qubits, which are capable of entanglement and superposition, and interact with one another via quantum gates.

Fault-tolerance: Any real physical information processing apparatus, whether man-made or biological, is subject to errors. To make computing systems scalable in the presence of errors, a fault-tolerant architecture is required. This old problem has become acute in the case of quantum computers, where a considerable gap remains to be closed between experimentally available error rates and the thresholds at which fault tolerant architectures would take hold.

Physical Complexity: How can various mathematical notions of complexity, such as time/space complexity, parallel complexity, and algorithmic information, be used to characterize the complexity of physical states, phase transitions, and the behavior of systems at and away from thermal equilibrium. Are there physical systems or dynamics that are uncomputable in the mathematical sense?

Physical Authentication: Can our understanding of the computational complexity be used to authenticate physical objects and evolutions as genuine, rather than forged or simulated?

© Physics of Information Group at IBM Research