Talks

Quantum control is valuable for various quantum technologies such as high-fidelity gates for universal quantum computing, adaptive quantum-enhanced metrology, and ultra-cold atom manipulation. Although supervised machine learning and reinforcement learning are widely used for optimizing control parameters in classical systems, quantum control for parameter optimization is mainly pursued via gradient-based greedy algorithms. Although the quantum fitness landscape is often compatible for greedy algorithms, sometimes greedy algorithms yield poor results, especially for large-dimensional quantum systems. We employ differential evolution algorithms to circumvent the stagnation problem of non-convex optimization, and we average over the objective function to improve quantum control fidelity for noisy systems. To reduce computational cost, we introduce heuristics for early termination of runs and for adaptive selection of search subspaces. Our implementation is massively parallel and vectorized to reduce run time even further. We demonstrate our methods with two examples, namely quantum phase estimation and quantum gate design, for which we achieve superior fidelity and scalability than obtained using greedy algorithms.

Many tasks in quantum information rely on accurate knowledge of a system's Hamiltonian, including calibrating control, characterizing devices, and verifying quantum simulators. In this talk, we pose the problem of learning Hamiltonians as an instance of parameter estimation. We then solve this problem with Bayesian inference, and describe how rejection and particle filtering provide efficient numerical algorithms for learning Hamiltonians. Finally, we discuss how filtering can be combined with quantum resources to verify quantum systems beyond the reach of classical simulators. More information on this topic is available at http://www.cgranade.com/research/talks/qml/2016/

Quantum machine learning algorithms usually translate a machine learning methods into an algorithm that can exploit the advantages of quantum information processing. One approach is to tackle methods that rely on matrix inversion with the quantum linear system of equations routine. We give such a quantum algorithm based on unregularised linear regression. Opposed to closely related work from Wiebe, Braun and Lloyd [PRL 109 (2012)] our scheme focuses on a classification task and uses a different combination of core routines that allows us to process non-sparse inputs, and significantly improves the dependence on the condition number. The second part of the talk presents an idea that transcends the reproduction of classical results. Instead of considering a single trained classifier, practicioners often use ensembles of models to make predictions more robust and accurate. Under certain conditions, having infinite ensembles can lead to good results. We introduce a quantum sampling scheme that uses the parallelism inherent to a quantum computer in order to sample from 'exponentially large' ensembles that are not explicitely trained.

Supervised Machine Learning is one of the key problems that arises in modern big data tasks. In this talk, I will first describe several different classical algorithmic paradigms for classification and then contrast them with quantum algorithmic constructs. In particular, we will look at classical methods such as the nearest neighbor rule, optimization based algorithms (e.g. SVMs), Bayesian inference based techniques (e.g. Bayes point machine) and provide a unifying framework so that we can get a deeper understanding about the quantum versions of the methods.

1. Beatrice Bonga, "Closed universes and the CMB",

Abstract:
Cosmic microwave background (CMB) observations put strong constraints on the spatial curvature via estimation of the parameter $\Omega_k$. This is done assuming a nearly scale-invariant primordial power spectrum. However, we found that the inflationary dynamics is modified due to the presence of spatial curvature leading to corrections to the primordial power spectrum. When evolved to the surface of last scattering, the resulting temperature anisotropy spectrum shows deficit of power at low multipoles ($\ell<20$). This may partially explain the observed $3 \sigma$ anomaly of power suppression for $\ell <30$. Since the curvature effects are limited to low multipoles, the estimation of cosmological parameters remains robust under inclusion of positive spatial curvature.

(This talk is based on http://arxiv.org/abs/1605.07556)

2. Keir Rogers, "Spin-SILC: CMB polarisation component separation for next-generation experiments",

Abstract:

B-mode polarisation is a powerful cosmological observable which in principle allows the detection of a stochastic background of gravitational waves predicted by inflation, and gives strong constraints on the neutrino sector using the weak gravitational lensing of the cosmic microwave background (CMB). Astrophysical foregrounds present a formidable obstacle in extracting these signatures of new physics from CMB polarisation data. Indeed, recent forecasts for post-2020 CMB experiments predict one sigma constraints on, for example, the tensor-to-scalar ratio of about 10^-4 and the sum of neutrino masses of about 30 meV. However, these constraints are predicated on highly-accurate foreground and noise removal. I will present the first component separation method specifically developed for this task and tested on the latest-release Planck data. The method, Spin-SILC, is an internal linear combination algorithm that uses spin wavelets to fully analyse the spin polarisation signal P = Q + iU, where Q and U are the measured Stokes parameters. This allows all the information in the measured signal to be used in extracting the cosmological background. Furthermore, Spin-SILC is the first method to simultaneously and efficiently perform component separation and the E-B decomposition necessary for cosmological analyses thanks to the construction of the spin wavelets we use. Spin-SILC also uses the morphological information of the foregrounds and CMB to better localise the cleaning algorithm. This is because the wavelets we use are additionally directional, and, when convolved with signals on the sphere, can separate the filamentary structures which are characteristic of both the CMB and foregrounds. I will present the results of applying these novelties to Planck data and discuss further how Spin-SILC can also mitigate the E-B leakage problem of future CMB experiments.