Watching the Mythos debate play out. Some see a watershed moment, others see hype. Fair argument to have, but if you're responsible for cyber, product, or technology, it's not the argument worth your time.
The reality is that AI capability keeps getting better. So whether Mythos is the inflection point or just another step matters less than whether your org can do the boring work at speed.
Time from patch availability to deployed in production.
Percentage of deploys with automated tests, phased rollout, and automatic rollback.
Time from an upstream fix in an open source dependency to running in prod.
These look like cyber metrics. They're really a measure of whether your organization can change at the speed this work now requires.
Cyber resilience needs this. Agentic AI needs this. Any product team trying to ship real value needs this. The orgs that can't do the boring work at speed won't be able to keep up.
Spend less time arguing about the model. Measure the maturity of your deployment pipeline.
Animation of the double slit experiment, to be used in the upcoming ACM TechTalk on July 17.
The talk will be based on the recently released "Building Quantum Software" book. The animation was created with help from Gemini.
#QuantumComputing#manim#visualization#Gemini
Quantum computing parallelism ~ SIMP = single instruction multiple pairs
The description of the effect of a single qubit gate as a single instruction changing all amplitudes in the state in pairs is as precise as possible. The pair processing is the butterfly pattern, and it happens simulataneously on a real quantum computer, logically speaking (the amplitudes in the state are our representation).
This is the parallelism (compared to classical equivalents) that can give a quantum advantage. What about controlled gates? The same pair processing occurs, but for only half, or a quarter, or a smaller fraction of the pairs. Theorically you can say that a controlled gate is a multi-qubit gate that is an instruction acting on groups (4, or 8, or 16, etc) of amplitudes, but in fact those are just a lot of multiplications by zero that can be avoided by just working on fewer pairs.
Conflating parallel universes with computational parallelism is the source of widespread confusion.
#quantumcomputing
#parallelism
Nature creates patterns with tiny probabilities.
You may recognize the pattern in the image as the intensities of light in a double slit experiment. This simulation uses 12 qubits, allowing for 4,098 locations to be measured. The probability of the location with the highest intensity is 0.002.
#QuantumComputing
Creating 128 dots with 91 quantum instructions. The quantum matrix.
This is a larger version of a video I posted earlier. This time I'm showing the encoding of the identity function with 7-qubit inputs and outputs. This can give a better idea of how quantum computing works.
- The state of a quantum bit system consists of a large number of complex numbers, which can be visualized as pixels in an image.
- A quantum computing instruction (gate) changes all these numbers (pixels) instantly.
- The state is not visible, but as its programmers, we know what it is, so we can visualize it.
- Measuring/reading a quantum state gives just the location of a pixel. If a pixel is more intense, its location is measured more often.
- Creating a desired state efficiently requires taming the immense power of quantum gates with interference while avoiding too much entanglement.
- Classically, one can create the 128 dots in the graph of the identity function with a for loop with 128 iterations.
- The quantum implementation uses 91 gates (the effect of each gate is shown in the video).
In quantum computing, the 0s and 1s are coming from measurements, they are not manipulated directly. This is the quantum version of the matrix: complex numbers (color pixels) are changed, and we only see their intensity by repeating the same computation.
#quantumcomputing
#visualization
#thematrix
It's Elementary: Meet Quantum Computing
A brief and elementary introduction to quantum computing for first-year college students and motivated high-school graduates.
Part 1: https://t.co/apgO2CQax3
Part 2: https://t.co/Gz0ezsowG9
#QuantumComputing#Elementary#Visual #Introduction
New benchmarks for FFT implementations from an amazing high-school student after PhastFT added support for single-precision.
https://t.co/DSXkm2eg4h
#FastFourierTransform#QuantumComputing