No abstract

Introduction: A series of modelling reports that quantify the effect of non pharmaceutical interventions (NPIs) on the spread of the SARS-CoV-2 virus have been made available prior to external scientific peer-review. The aim of this study was to investigate the method used by the Imperial College COVID-19 Research Team (ICCRT) for estimation of NPI effects from the system theoretical viewpoint of model identifiability. Methods: An input-sensitivity analysis was performed by running the original software code of the systems model that was devised to estimate the impact of NPIs on the reproduction number of the SARS-CoV-2 infection and presented online by ICCRT in Report 13 on March 30 2020. An empirical investigation was complemented by an analysis of practical parameter identifiability, using an estimation theoretical framework. Results: Despite being simplistic with few free parameters, the system model was found to suffer from severe input sensitivities. Our analysis indicated that the model lacks practical parameter identifiability from data. The analysis also showed that this limitation is fundamental, and not something readily resolved should the model be driven with data of higher reliability. Discussion: Reports based on system models have been instrumental to policymaking during the SARS-CoV-2 pandemic. With much at stake during all phases of a pandemic, we conclude that it is crucial to thoroughly scrutinise any SARS-CoV-2 effect analysis or prediction model prior to considering its use as decision support in policymaking. The enclosed example illustrates what such a review might reveal.

The role of non-pharmaceutical interventions (NPIs) on the spread of SARS-CoV-2 has drawn significant attention, both scientific and political. Particularly, an article by the Imperial College COVID-19 Response Team (ICCRT), published online in Nature on June 8, 2020, evaluates the efficiency of 5 NPIs. Based on mortality data up to early May, it concludes that only one of the interventions, lockdown, has been efficient in 10 out of 11 studied European countries. We show, via simulations using the ICCRT model code, that conclusions regarding the effectiveness of individual NPIs are not justified. Our analysis focuses on the 11th country, Sweden, an outlier in that no lockdown was effectuated. The new simulations show that estimated NPI efficiencies across all 11 countries change drastically unless the model is adapted to give the Swedish data special treatment. While stated otherwise in the Nature article, such adaptation has been done in the model code reproducing its results: An ungrounded country-specific parameter said to have been introduced in all 11 countries, is in the code only activated for Sweden. This parameter de facto provides a new NPI category, only present in Sweden, and with an impact comparable to that of a lockdown. While the considered NPIs have unarguably contributed to reduce virus spread, our analysis reveals that their individual efficiency cannot be reliably quantified by the ICCRT model, provided mortality data up to early May.

ControlSystems.jl enables the powerful features of the Julia language to be leveraged for control design and analysis. The toolbox provides types for state-space, transfer-function, and timedelay models, together with algorithms for design and analysis. Julia's mathematically-oriented syntax is convenient for implementing control algorithms, and its just-in-time compilation gives performance on par with C. The multiple-dispatch paradigm makes it easy to combine the algorithms with powerful tools from Julia's ecosystem, such as automatic differentiation, arbitrary-precision arithmetic, GPU arrays, and probability distributions. We demonstrate how these features allow complex problems to be solved with little effort. I. INTRODUCTIONThe Julia programming language [1] has, over the last couple of years, revolutionized technical computing. It is a high-level language with mathematically-oriented syntax and semantics, but still achieves execution speeds comparable to C by relying on just-in-time compilation. Julia is free, open source and due to its suitability for technical computing, already has a rich ecosystem with high-quality packages for applied mathematics. Julia's use of multiple-dispatch and duck typing simplifies code reuse and composability across packages, making it possible to achieve complex functionality with little effort. The power of the Julia language has been demonstrated in numerous applications [2].ControlSystems.jl [3] provides a large set of algorithms for control design and analysis, enabling the control community to leverage the power of the Julia language and its ecosystem. For example, ControlSystems.jl makes it easy to tune controller parameters using automatic differentiation and optimize performance subject to uncertainty by propagating probability distributions.The ControlSystems.jl toolbox supports common functionality such as creating linear time-invariant (LTI) systems using either a state-space representation or as transfer functions with either polynomials or zeropole-gain representations. Systems with delays are supported, with the time-response capabilities provided

The increased use of cloud and other large scale datacenter IT services and the associated power usage has put the spotlight on more energy-efficient datacenter management. In this paper, a simple model was developed to represent the heat rejection system and energy usage in a small DC setup. The model was then controlled by a reinforcement learning agent that handles both the load balancing of the IT workload, as well as cooling system setpoints. The main contribution is the holistic approach to datacenter control where both facility metrics, IT hardware metric and cloud service logs are used as inputs. The application of reinforcement learning in the proposed holistic setup is feasible and achieves results that outperform standard algorithms. The paper presents both the simplified DC model and the reinforcement learning agent in detail and discusses how this work can be extended towards a richer datacenter model. CCS CONCEPTS• Computer systems organization → Sensors and actuators; • Hardware → Enterprise level and data centers power issues; Temperature control; • Computing methodologies → Reinforcement learning; Modeling and simulation.