This talk will outline three revolutions that happened in the past decades in the development of Earth system models that are used to perform weather and climate predictions. The quiet revolution has leveraged better observations and more compute power to allow for constant improvements of prediction quality of the last decades, the digital revolution has enabled us to perform km-scale simulations on modern supercomputers that further increase the quality of our models, and the machine learning revolution has now shown that machine learned weather models are often competitive with physics based weather models for many forecast scores while being easier, smaller and cheaper. This talk will summarize the past developments, explain current challenges and opportunities, and outline how the future of Earth system modelling will look like. In particular, regarding machine-learned foundation models in a physical domain such as Earth system science.

TBA

In my talk, I will present a novel approach to magnetic resonance microscopy that exploits nitrogen-vacancy (NV) centers in diamond for optically detected magnetic resonance (ODMR). The fusion of optical microscopy and nuclear magnetic resonance (NMR) spectroscopy bypasses the conventional reliance on k-space sampling and magnetic field gradients for spatial encoding of NMR signals, enabling real-space magnetic resonance imaging (MRI). We demonstrate the capabilities of our widefield optical NMR microscopy technique by imaging NMR signals within a model microstructure, achieving a spatial resolution of approximately 10 μm over an area of ~235 × 150 μm². Each camera pixel captures a complete NMR spectrum, providing comprehensive information on signal amplitude, phase, local magnetic field strengths, and gradients. The integration of optical microscopy and NMR opens up new possibilities for a wide range of applications in the physical and life sciences, which I will discuss in the last part of my talk. These applications include imaging metabolic activity in single cells or tissue slices, analyzing battery materials, and facilitating high-throughput NMR analysis.

Title: Circuits of Microfluidic Memristors: Computing with Aqueous Electrolytes Speaker: René van Roij, Institute for Theoretical Physics, Utrecht University, The Netherlands Abstract: In this online talk we will discuss recent advances in our understanding of the physics of cone-shaped microfluidic channels under static and pulsatile voltage- and pressure drops. On the basis of Poisson-Nernst-Planck-Stokes equations for transport of aqueous electrolytes through channels carrying a surface charge, we will provide a theoretical explanation for the experimentally observed diode-like current rectification of these channels. At steady electric driving this rectification involves salt depletion or accumulation in the channel depending on the sign of the applied voltage [1], and this effect also explains the observed pressure-sensitivity of the electric conductance. An extension towards an applied AC voltage predicts these channels to be tunable between diodes at low frequencies ωτ<<1, memristors (resistors with memory) at intermediate frequencies ωτ ~ 1, and Ohmic resistors at high frequency ωτ>>1 , with a characteristic (memory retention) time τ proportional to the square of the channel length [2]. We predict that Hodgkin- Huxley-inspired iontronic circuits of short (fast) and long (slow) conical channels yield neuromorphic responses akin to (trains of) action potentials [2] and several other neuronic spiking modes [3]. Next, we show theoretically and experimentally that a tapered microfluidic channel filled with an aqueous nearly close-packed dispersion of colloidal charged spheres is a much stronger memristor than the channel with only surface charges on the channel wall [4]. Upon applying a train of four positive (negative) voltage pulses, each pulse representing a binary “1” (“0”), we map the hexadecimal number represented by this train on an analog channel conductance, which offers opportunities for reservoir computing -we give a proof of principle for the case of recognizing hand-written digits [4]. Finally we will also discuss recent and ongoing work on iontronic information processing. We exploit the mobility of the medium (water) by considering simultaneously applied pulsatile pressure and voltage signals to increase the bandwidth [5]. Finally, the versatile ionic nature of the charge carriers allows for Langmuir-like ionic exchange reaction kinetics on the channel surface [6]. We show that this can give rise to direct iontronic analogues of synaptic long-term potentiation and coincidence detection of electric and chemical signals [7], which are both ingredients for brain-like (Hebbian) learning. References: [1] W.Q. Boon, T. Veenstra, M. Dijkstra, and R. van Roij, Pressure-sensitive ion conduction in a conical channel: optimal pressure and geometry, Physics of Fluids 34, 101701 (2022). [2] T.M. Kamsma, W.Q. Boon, T. ter Rele, C. Spitoni, and R. van Roij, Iontronic Neuromorphic Signaling with Conical Microfluidic Memristors, Phys. Rev. Lett. 130, 268401 (2023). [3] T.M Kamsma, E. A. Rossing, C. Spitoni, and R. van Roij, Advanced iontronic spiking modes with multiscale diffusive dynamics in a fluidic circuit, Neuromorph. Comput. Eng. 4 024003 (2024). [4] T.M. Kamsma, J. Kim, K. Kim, W.Q. Boon, C. Spitoni, J. Park, and R. van Roij, Brain-inspired computing with fluidic iontronic nanochannels, PNAS 121, e23202242121 (2024). [5] A. Barnaveli, T.M. Kamsma, W.Q. Boon, and R. van Roij, Pressure-gated microfluidic memristor for pulsatile information processing, arXiv:2404.15006. [6] W.Q. Boon. M. Dijkstra, and R. van Roij, Coulombic Surface-Ion Interactions Induce Nonlinear and Chemistry-Specific Charging Kinetics, Phys. Rev. Lett. 130, 058001 (2023). [7] T.M. Kamsma, M. Klop, W.Q. Boon, C. Spitoni, and R. van Roij, arXiv:2406.03195