My journey in creating SuperCAI — A superconductor simulation model to advance quantum computing, superconductor, and mechanical industries
After being “super” interested and doing a bunch of research on quantum hardware, I found that the concept of conductors was important, and critical to my quantum computing proposition. As I was completing even more AI projects, it got me thinking: Could AI possibly help to develop quantum computers?
What part of quantum computers would this artificial intelligence algorithm be used for?
Then THAT got me to realize the answer: – Superconductors!! 🥳🥳🥳
But then I started asking the how. Artificial intelligence has been used for “super” things before, including artificial intelligence algorithms that could test for aging and lifetime in supercapacitor cells. However, supercapacitors and superconductors are not the same thing. I knew I had my work cut out for me, and I knew that this journey would be exceedingly confusing, but I wanted to successfully complete this project.
Before I had even started my focus in AI, I already knew what I wanted to do.
I learned even more about the math and intuition behind AI. To do so,
- I Learned a bit on more linear algebra and calculus (fortunately, I was also talking calc in school);
- Did image style transfer using CNNs Built a NLP + RNN and LSTM algorithm for sentiment analysis;
- Built a neural network to predict credit card fraud;
- Developed Bayesian NP nonparametric time-series;
- Did a lot of research and development in deep learning and artificial intelligence as well as other neuromorphic technologies;
- Got super fascinated by the brain
and though, it does not seem like a lot of items, they are hours and hours of stuff.
Interesting stuff, but still 100+ hours.
Then I stumbled upon the realm of superconductors, and their intersection with artificial intelligence.
Superconductors
Superconductors – Let us take a small journey in conductivity because conductivity leads us to superconductivity.
Superconductors are materials that are very conductive, with a capital r. They are far more conductive than semiconductors.
Conductivity describes the ability of a material to allow energy pass through it with a certain amount of resistance, which is a measure of opposition to the energy being passed. Though when we hear conductivity, we always think electricity, conductors can cOnDuCt, a variety of energy types:
- Kinetic Energy, like heat
- Sound
- Electricity, of course.
Conduction is one of the most commonly taught and learned energy transfer types, and one of three of the fundamental heat transfer methods (the other two being convection and radiation). Below are some animated images of conduction, feel free to find and view more on the internet.
Conduction!
More conduction
More conduction!
For heat, conduction is described as
Where q is local heat flux density (how fast heat goes), –k is the measure of conductivity (how conductive it is), and ∇T (read: nabla T) is the temperature gradient, which describes the direction and change in temperature, much like how its cousin delta describes change.
For electrolytic conductivity (conduction of electricity), the formula is
Where R = resistance, A = material area (how much space does the material take up on the inside), l = distance, and ρ (read: rho), is the specific resistance. Specific resistance, on a high level, is the level of conductivity based on a unit of the material when there is an applied voltage (when it gets shocked with ⚡️).
There are two types of conduction: transient and steady state.
Steady state conduction deals heavily with heat transfer (which it is typically applied to). It basically means that the conduction is at equilibrium, or its unchanging, even when heat is being passed through the material. The distribution of heat just tends to remain unchanged, which defines steady state conduction.
Transient conduction is literally the opposite. So much so that it is sometimes referred to as non-steady-state conduction. It essentially means that there are spatial temperature changes due to the heat transfer at different times during the conduction process.
Oftentimes the conductive material is at equilibrium, but what happens is that an environmental change or ambient temperature difference, or really any type of interference thrusts the process into transience. This means that transient conduction makes temperature depend on material position and time.
So, let’s get back to what a superconductor is.
Now that you know all about conductors, you should understand this.
A superconductor is a material that is superconductive 🤯!!
Just kidding, I am going to tell you what superconductivity is right now!
Superconductors are materials that can pass energy through them with 0 resistance. In fact, in science, super often means no resistance, like how a superfluid is a fluid material with no resistance to flow (resistance to flow = a property called viscosity).
Anyway, superconductors are really, really, important. Le us run through a number of exciting reasons why.
Have you heard of a hyperloop? Well, after chatting with the amazing Meit Shah [see my notes linked on his name or reach out via LinkedIn Okezue Bell] on magnetic levitation (among other cool hyperloop things 😎), guess what could power them? If you guessed Superconductors, you are right!
Superconductor properties allow for magnets to levitate on them!
In fact, if we model superconductors properly, we can make the hyperloop travel with 0 friction between the “wheels” and the track 🚅💨.
How about Google’s quantum computer?
It runs off of superconducting electronic circuits.
If superconductors are modeled well, there could be huge advancements in reducing negative effects of quantum decoherence and other issues by using new superconductive materials.
What about qubits?
Transmons are some of the best qubits around, and they are used in superconducting quantum computing, and they are superconductive!!
Electromagnets?
Make strong ones with superconductors.
SQUIDs 🦑?
Superconductors.
WAIT, wait, wait, wait…not that type of squid!!!
SQUIDs. Superconducting Quantum Interference Devices, I mean this type of squid. They are super powerful electromagnet systems that can detect even extremely weakened magnetic field signals.
Motors?
More efficient with superconductors.
High speed magnetic levitation trains?
More efficient with superconductors.
More efficient transmission and voltage transfer for communication? Ha!
More efficient with superconductors.
How about a really, good generator for the next storage devices
More efficient with superconductors.
These illustrate just a small gallery of things that superconductors can accelerate.
Yeah… big flex with the superconductors.
So clearly superconductors have a lot of uses. But why model them with AI?
Because superconductors are expensive and time-soaking to test. Therefore, by using AI, we’re able to cut down trial costs. Not only that, but superconductors have problems. They act weird in certain environments.
For example, they can do all of these somewhat negative (-) things:
- Be chemically unstable (exploding, volatile behaviour [but not in a this-material-evaporates-or-sublimates-type-way])
- Requires a lot of energy and $$ to keep them cold and at low-critical temperatures, which has proven to be a problem in quantum computing
- They have no (well, limited) malleability. They are more brittle than 🥜-brittle (now I’m hungry…). Without ductility, superconductors cannot be easily customized or shaped to build different products
- Lots of trial and error
However, with computational chemistry and AI, I aim to stop this. My product uses state transformations to convert superconductor electromagnetic simulations to classical computational electromagnetic (CEM) problems.
A more common example:
conductivity = MattisBardeenSuperconductorConductivity(gap_energy, conductivity_1)result = conductivity.evaluate(temperature, frequency)
print(f”sigma = {result}”). In addition, my algorithm can calculate degradation and lifetime in superconductors and superconductor coils (which can be up to 100,000). Wait, instead of taking this route of explanation, I’ll just show you some graphs, and use bullets 🧠
My Algorithm(s) / Project
After reading this paper by Kam Hamidieh: A Data-Driven Statistical Model for Predicting the Critical Temperature of a Superconductor https://arxiv.org/abs/1803.10260
I was able to comprehend temperature predictions in R, too.
This project was able to:
- Determine superconductor material properties
- Simulate superconducting qubits
- Minimize free energy in superconducting units
- Determine temperature
- Finding critical temperature
- Determine frequencies
- Conduct Electromagnetic Simulations
- Determine aging and lifetime of superconductors
- Help model superconductor integration for quantum computing system
- Model more interesting uses!!
Here are some photos from the simulations:
Temperature and frequency graphical simulations
Lifetime, Frequency, and complex electromagnetic simulation
Transmon superconducting qubit simulations
Summary.
AI is probably the most important opportunity to learn about superconductors and to get close to the potential of their applications. What I have shared is by no means comprehensive but nonetheless some really interesting stuff, and I definitely recommend looking into my website for more. My SuperCAI website is https://conductai.netlify.app (now called ConductAI). Also feel free to share your work and ideas, invite me to speak at your conference or institution; we can learn from each other! Hey if anything here, created more questions than answers, that is great…it is supposed to!
•Okezue Bell is an artificial intelligence and biology/biotechnology developer. Okezue has been recently sponsored by companies to build brain-controlled prosthetics, and generate electricity using cells. He is also engaged in the cellular agriculture space, where he is working with American companies such as Aleph Farms and Perfect Day to build new products. Find him at Okezuebell.com or via LinkedIn
