Computer Science

Droplet-based microfluidic systems are a promising platform forlab-on-a-chip (LoC) applications. These systems can also be used toenhance LoC applications with integrated droplet control information or for data transmission scenarios in the context of molecular communication. For both use-cases the detection and characterisation of droplets in small microfluidic channels is crucial. So far, only complex lab setups with restricted capabilities have been presented as detection devices. We present a new low-cost and portable droplet detector. The device is used to confidently distinguish between individual droplets in a droplet-based microfluidic system. Using on-off keying a 16-bit sequence is successfully transmittedfor the first time with such a setup. Furthermore, the devices capabilities to characterise droplets regarding colour and size are demonstrated. Such an application of a spectral sensor in a microfluidic system presents new possibilities, such as colour-coded data transmission or analysis of droplet content.

Wang et al. claim [J. Appl. Phys. 125, 054504 (2019)] that a current-carrying wire interacting with a magnetic core represents a memristor. Here, we demonstrate that this claim is false. We first show that such memristor "discovery" is based on incorrect physics, which does not even capture basic properties of magnetic core materials, such as their magnetic hysteresis. Moreover, the predictions of Wang et al.'s model contradict the experimental curves presented in their paper. Additionally, the theoretical pinched hysteresis loops presented by Wang et al. can not be reproduced if their model is used, and there are serious flaws in their "negative memristor" emulator design. Finally, a simple gedanken experiment shows that the proposed Φ -memristor would fail the memristor test we recently suggested in J. Phys. D: Appl. Phys. 52, 01LT01 (2019). The device "discovered" by Wang et al. is just an inductor with memory.

In his paper "If it's pinched it's a memristor" [Semicond. Sci. Technol. 29, 104001 (2014)] L. Chua claims to extend the notion of memristor to all two-terminal resistive devices that show a hysteresis loop pinched at the origin. He also states that memcapacitors and meminductors can be defined by a trivial replacement of symbols in the memristor relations, and, therefore, there should be a correspondence between the hysteresis curves of different types of memory elements. This leads the author to the erroneous conclusion that charge-voltage curves of any memcapacitive devices should be pinched at the origin. The purpose of this Comment is to correct the wrong statements in Chua's paper, as well as to highlight some other inconsistencies in his reasoning. We also provide experimental evidence of a memcapacitive device showing non-pinched hysteresis.

Artificial neural networks are notoriously power- and time-consuming when implemented on conventional von Neumann computing systems. Consequently, recent years have seen an emergence of research in machine learning hardware that strives to bring memory and computing closer together. A popular approach is to realise artificial neural networks in hardware by implementing their synaptic weights using memristive devices. However, various device- and system-level non-idealities usually prevent these physical implementations from achieving high inference accuracy. We suggest applying a well-known concept in computer science -- committee machines -- in the context of memristor-based neural networks. Using simulations and experimental data from three different types of memristive devices, we show that committee machines employing ensemble averaging can successfully increase inference accuracy in physically implemented neural networks that suffer from faulty devices, device-to-device variability, random telegraph noise and line resistance. Importantly, we demonstrate that the accuracy can be improved even without increasing the total number of memristors.

Compact device models play a significant role in connecting device technology and circuit design. BSIM-CMG and BSIM-IMG are industry standard compact models suited for the FinFET and UTBB technologies, respectively. Its surface potential based modeling framework and symmetry preserving properties make them suitable for both analog/RF and digital design. In the era of artificial intelligence / deep learning, compact models further enhanced our ability to explore RRAM and other NVM-based neuromorphic circuits. We have demonstrated simulation of RRAM neuromorphic circuits with Verilog-A based compact model at NCKU. Further abstraction with macromodels is performed to enable larger scale machine learning simulation.

Memristor, memory resistor, is an emerging technology for computational memory. Number of different memristor models are available based on the physical experiments. To use memristor as a computational memory element, one should know how the internal state modulates in time when driven by current or voltage. In this paper, we examine three widely used models and make a comparison of how internal state in these models changes with respect to input current or voltage. In Strukov model, internal state changes linearly with the input current. However, the linearity of internal state modulation in Yang model can be controlled. On the other hand, Pickett model shows non linear variation in internal state with the input current.

Biological regulatory networks depend upon chemical interactions to process information. Engineering such molecular computing systems is a major challenge for synthetic biology and related fields. The chemical reaction network (CRN) model idealizes chemical interactions, allowing rigorous reasoning about the computational power of chemical kinetics. Here we focus on function computation with CRNs, where we think of the initial concentrations of some species as the input and the equilibrium concentration of another species as the output. Specifically, we are concerned with CRNs that are rate-independent (the computation must be correct independent of the reaction rate law) and composable ( f∘g can be computed by concatenating the CRNs computing f and g ). Rate independence and composability are important engineering desiderata, permitting implementations that violate mass-action kinetics, or even "well-mixedness", and allowing the systematic construction of complex computation via modular design. We show that to construct composable rate-independent CRNs, it is necessary and sufficient to ensure that the output species of a module is not a reactant in any reaction within the module. We then exactly characterize the functions computable by such CRNs as superadditive, positive-continuous, and piecewise rational linear. Thus composability severely limits rate-independent computation unless more sophisticated input/output encodings are used.

Hyphae within the mycelia of the ascomycetous fungi are compartmentalised by septa. Each septum has a pore that allows for inter-compartmental and inter-hyphal streaming of cytosol and even organelles. The compartments, however, have special organelles, Woronin bodies, that can plug the pores. When the pores are blocked, no flow of cytoplasm takes place. Inspired by the controllable compartmentalisation within the mycelium of the ascomycetous fungi we designed two-dimensional fungal automata. A fungal automaton is a cellular automaton where communication between neighbouring cells can be blocked on demand. We demonstrate computational universality of the fungal automata by implementing sandpile cellular automata circuits there. We reduce the Monotone Circuit Value Problem to the Fungal Automaton Prediction Problem. We construct families of wires, cross-overs and gates to prove that the fungal automata are P-complete.

Actin filaments are conductive to ionic currents, mechanical and voltage solitons. These travelling localisations can be utilised in making the actin network executing specific computing circuits. The propagation of localisations on a single actin filament is experimentally unfeasible, therefore we propose a `relaxed' version of the computing on actin networks by considering excitation waves propagating on actin bundles. We show that by using an arbitrary arrangement of electrodes it is possible to implement two-inputs-one-output circuits. Frequencies of the Boolean gates' detection in actin network match an overall distribution of gates discovered in living substrates.

The thermodynamic binding networks (TBN) model is a tool for studying engineered molecular systems. The TBN model allows one to reason about their behavior through a simplified abstraction that ignores details about molecular composition, focusing on two key determinants of a system's energetics common to any chemical substrate: how many molecular bonds are formed, and how many separate complexes exist in the system. We formulate as an integer program the NP-hard problem of computing stable (a.k.a., minimum energy) configurations of a TBN: those configurations that maximize the number of bonds and complexes. We provide open-source software solving this integer program. We give empirical evidence that this approach enables dramatically faster computation of TBN stable configurations than previous approaches based on SAT solvers. Furthermore, unlike SAT-based approaches, our integer programming formulation can reason about TBNs in which some molecules have unbounded counts. These improvements in turn allow us to efficiently automate verification of desired properties of practical TBNs. Finally, we show that the TBN has a natural representation with a unique Hilbert basis describing the "fundamental components" out of which locally minimal energy configurations are composed. This characterization helps verify correctness of not only stable configurations, but entire "kinetic pathways" in a TBN.