What is Non-equilibrium Dynamics?

this post focuses mainly on Non-equilibrium Dynamics and Dissipative Structure, and how non-living systems under specific external source input can exhibit living system like behaviors and movements.

Research and development efforts in so-called artificial intelligence has increased dramatically. However, designing AI with energy efficiency is becoming an important priority. It is not yet clear how this should be done. One possible inspiration is to study the physics of self-organizing systems, both non-living and living, as guidance for future designs with functional intelligence. Irreversible processes at non-equilibrium can drive a system to self-organize and exhibit characteristics shown in systems known as dissipative structures. Our research explores the characteristics of experiments that use electrically conductive beads in an applied electric field. The setup resembles a primitive dissipative structure that can be interpreted as a possible bridge between behaviors in non-living and living systems. Using video and electrical measurements, we investigate the transient and steady-state behavior of self-organizing worm-like behavior under a range of initial and driving conditions. The non-equilibrium processes of a non-living system exhibiting characteristics that also exist within a living system are a possible way to explore biomimetic structures that exhibit intelligence.

So, what is Dissipative Structure exactly?

A dissipative system is a thermodynamically open system which is operating out of, and often far from, thermodynamic equilibrium in an environment with which it exchanges energy and matter. A dissipative structure is a dissipative system that has a dynamical régime that is in some sense in a reproducible steady state. This reproducible steady state may be reached by natural evolution of the system, by artifice, or by a combination of these two. In 1977, Chemist Ilya Prigogine won the Nobel Chemistry Prize for the Dissipative Structure theory.

Methodology

The experimental setup will first be the replicate version of the setup from Kondepudi et al.’s two papers, and if necessary, further modification will be applied.

The experimental setup consists of a High Voltage aluminum rod electrode with needle-like tip at the end, a 0-30kV High Voltage Output Source, silicon oil of 50 cst, a 15cm diameter petri dish, a 9cm diameter stainless steel ring as the ground electrode, a complete circuit (which will be further explained), a High Voltage probe, a regular multimeter, a digital multimeter, a laptop, an iPad, and a duo voltage supply to act as the HV supply power source and voltage output adjustment knob for the HV output. (See Figure 1 for theoretical model diagram.)

The circuit consists of three resistors, one of 10 KΩ, and two of 100 MΩ. One end of the circuit is connected in series with the ground, and the other end is connected to the stainless steel ring. A digital multimeter is connected to the 10k ohm resistor to measure the voltage across the resistor. We can later use the equation I = V/R to get the current across the circuit. A clamp stand is placed next to the circuit, and two clamps are used for holding the aluminum electrode rod and the HV probe. Both the aluminum rod and the HV probe are insulated with HV electric tape and teflon tape. Foam plastic pieces are placed on both sides of the stand so that an acrylic transparent piece can be balanced on the pieces. A 15 cm Petri dish is placed on the acrylic piece, and a camera is placed beneath the acrylic piece. Here we are using iPad as the video taking facility. The stainless steel ring will be placed at the center of the petri dish. Stainless steel beads of various sizes are used during the experiment, and they are placed inside of the ground ring. Silicon oil is applied to the petri dish to the level that it just submerges the beads. The oil is used to reduce the friction between the beads and the surface of the petri dish and to insulate the beads.

From our preliminary results, it seems that several interesting aspects have occurred:

1. The noise information is valuable because it is clear that there is a sudden noise increase after the tree formation. (See Figure 2.)Python code will be written to do Fourier Transformation on existing data, and this is a good method for measuring noises that are of low frequency. Furthermore, Spectrum Analyzer will be used directly to observe the peaks from the frequency spectrum, and since the Spectrum Analyzer we have can scan over a wide range, it can be used to detect noises from the higher frequency range.

2. The motion of the beads suggests that the whole system is quite complicated. There are several modes in the dynamic system. Some groups of beads will naturally form a tree-like structure after some time duration, while others will continue being dynamic and keep on oscillating back and forth as the electron shuttling motion. Further research will need to be done to have the ability to create computational models and simulations.

3. More systematic manipulations should be done to test the self-recovery ability of the beads/patterns. The setup will potentially be modified so that systematic force will be applied to the ring to see if there is any quantifiable relationships between the level of perturbation and the recovery time and results. We are planning to make or purchase a malleable and bendable stainless steel ring that can allow us to apply constant force to make modifications to the shape of the ring. A non-conducting stabilizer will also be needed to fix the ring in place. The force can also be applied by an adjustable scaler so that the force can be interpreted as the length that is extended out by the scaler.

4. In Kondepudi et al.’s papers, it is mentioned that the entropy production keeps increasing as the tree structure tries to seek for the most optimal path. From our preliminary data, we observed something different. (See Figure 3.) The red line is the fit for the data points that were collected. The function for the fit shows Y = 2.79*e-5.97x +6.25, which is an exponential decay. More experiments will need to be done to understand the reason why this happened, and potentially a different theory can be proposed.

5. One observation in Kondepudi et al.’s papers has been proven: 1). When no beads are in the ground ring, the current measured across the circuit is proven to be increasing almost quadratically with the increase of the input High Voltage. The plots are shown below in Figure 4, Figure 5, and Figure 6. It is reasonable to conclude that there is a typo in Kondepudi et al.’s 2017 paper on the equation of the fitting line because the experimental and fitting data from Figure 5 matches with our experimental data and fit, but when we used their written equation to plot, the two results varied by a significant amount. From Figure 6, it is clear to see that the smaller the height, the higher the voltage, although the amount of increase tend to get smaller when the height gets closer and closer to 0. More quantifiable systematic experiments need to be conducted to find out the quantitative relationship between the height increase and the amount of current decrease difference.

6. There is a positive linear relation between the current and the number of strands of beads with the same N. We conducted the experiment with bead of size 3/16” in diameter. For N = 4, data from number of strands from 1 to 9 were collected. Each run lasted about 60 seconds. Similar runs were done with N = 6, number of strands from 1 to 7, and N = 8, number of strands from 1 to 6. (The number of strands varied due to the spacing limit in the ground ring). The results are shown as in Figure 7 (experimental data and fits) and Figure 8 (fitting extended to more numbers of strands when space allowed). From Figure 8, it is clear to see the linear relation between current and the number of strands. What is more interesting is that it seems as the N of beads goes up, the difference of the increase is minimized. Therefore, we have the hypothesis that the increase in current will only appear when the total length of the strand is < than the radius of the ring, and once the length exceeds the radius length, minimal or no effects will be shown on the current.

7. By using the Spectrum Analyzer, we observed that there are constant frequency peaks that are at the range of several hundreds MHz after tree formation, which suggests that the beads are still not in wire-like connection even after tree formation. The very high frequency shuttling in the tree structure would be a very interesting aspect to study as this potentially can be contributing to the fact that the beads exhibit behaviors that are characteristic on living systems.

Figure 1. A schematic diagram of the basic set up. A voltage in the range 5–30 kV is applied to this system between the center source electrode and the ground ring electrode. The dotted lines indicate a conical spray of ions form the source to the ring electrode. The solid line shows the current flow from the source electrode to the tip of the tree. Kondepudi et al.

Figure 2: Plot of current through circuit for beads of size 3/16” in diameter and N = 60

Clear noise fluctuation increase before and after tree formation(the current jump)

Figure 3: plot and fit for N = 364 beads of 1/32” diameter under the time span of approximately 35 seconds after the tree formation.

Electronics

The digital multimeter, FLUKE 8842A, that is used has three frequencies when it comes to recording: 10 Hz, 20Hz, and 100Hz. The 100Hz mode will mostly be used for that the accuracy of the results will be largely increased with more data points.

The Dual DC voltage supply that is being used is of the model Hewlett.Packard 6205C. It has two separate parts that can output voltage. The first DC voltage component is set at 24V as the turn-on voltage for the HV supply. The second part is used to adjust the voltage output of the HV supply. 0-10V for the second part covers the full range (0-30kV) for the HV supply.

The HV supply uses a 15 pin D-type connector at the back which has the function of power input, current and voltage threshold, voltage adjustment, etc. The output connector is compatible with Amphenol HV connectors.

A spectrum analyzer will also be used to analyze the noise occurred during data collection.

Potentially a noise reduction controller needs to be purchased and applied to the system so that irrelevant noises (e.g. WiFi signal) can be eliminated and valuable frequency noises can be directly detected by the spectrum analyzer.