Lets mash up two far-out topics, Artificial Life and Quantum Computing, and we have Quantum Artificial Life! [1]
Whoa!
This fall IBM (who else?) reports an actual implementation of a form or artificial life, running on a quantum computer.
OK, it’s a 1 qbit genotype and 1 qbit phenotype, but it’s still the essence of one flavor of “artificial life”.
The researchers indicate that the term “quantum computing” has come to include a range of approaches. For that matter, the term “artificial life” has been applied to quite a few approaches, not even considering things like bio-inspired robots.
This particular group is interested in what they call “quantum biomimetics” which is concerned with “quantum algorithms based on the imitation of biological processes”.
“microscopic quantum systems can efficiently encode quantum features and biological behaviours, usually associated with living systems and natural selection.” ([1], p. 1)
Aside from the raw, “so there!”, of implementing something on QC, this flavor of AL happens in the face of quantum weirdness, such as entanglement and quantum states. Parents and descendants are linked via entanglement, and I don’t think that time works quite the same here, either.
Basically, they use quantum states and operations to implement a “genotype”, a corresponding “phenotype”, the lifetime of a phenotype (i.e., the environment interacting with the phenotype over time), and replication with mutation of the genotype into a new generation. Together, these are all the basic features of Darwinian evolution.
The implementation, of course, has nothing at all to do with actual biology. And, of course, I haven’t a clue about the details.
“The first step for this implementation is to express each of the building blocks of the previous paragraph in terms of the quantum gates available in the superconducting circuit architecture of IBM cloud quantum computer “ ([1]. p.2)
One evaluation of this exercise is to compare the QAL with classical computation of AL. This appears to be tricky.
They report that the QC model matches the probabliity distributions of the equivalent classical computation fairly well. However, there are quantum weirdnesses inside, so it isn’t really the same model, and the correspondence is perfect.
In any case, a big part of the point is that this sort of Darwinian adaptation is the heart of a powerful machine learning approach. To the degree that this QC version works, it may lead to practical quantum algorithms.
“We believe that the presented results and vision, both in theory and experiments, should hoist this innovative research line as one of the leading banners in the future of quantum technologies “ ([1], p. 8)
- U. Alvarez-Rodriguez, M. Sanz, L. Lamata, and E. Solano, Quantum Artificial Life in an IBM Quantum Computer. Scientific Reports, 8 (1):14793, 2018/10/04 2018. https://doi.org/10.1038/s41598-018-33125-3
Robert McGrath's Blog 