Quantum Computing

[Leo]

I wonder what Quantum Computing would mean for chess.

[towforce]

I wonder what Quantum Computing would mean for chess.

Deeper search - link:

"...we find examples of trees for which the classical algorithm requires time exponential in 𝑛, but for which the quantum algorithm succeeds in polynomial time."

[Engin]

i doubt that one day Quantum Computer can play chess.

[towforce]

This might not be possible, but maybe...

1. Find a smallish set of positions that represent chess very well

2. Model this set with some expressions

3. Set a large quantum computer with a huge number of qubits in a superstate to work on the set of expressions from (2)

4. The qubits collapse into an optimal model state, giving a mathematical expression that optimally resolves the model

Result: simple mathematical expression (which will therefore run quickly) that provides the best evaluation function ever for chess.

[Mike Sherwin]

I view so called quantum computers as 99.9% hype. They are really fast at combining two planes of bits together like adding together two height maps. What else are they good for? Can qubits benefit AI or even make a better chess engine?


ChatGPT
Quantum computers are indeed surrounded by a lot of hype, but they also have some genuine potential. To address your points and questions:

Quantum Computers' Strengths:
Optimization Problems: Quantum computers excel at exploring many possibilities simultaneously. Algorithms like the quantum approximate optimization algorithm (QAOA) are designed to solve complex optimization problems more efficiently than classical methods.

Simulation of Quantum Systems: Quantum computers are ideal for simulating quantum systems, such as molecules and materials, which can revolutionize chemistry, materials science, and drug discovery.

Cryptography: Shor's algorithm for integer factorization could potentially break classical encryption systems (e.g., RSA). This poses both risks and opportunities.

Machine Learning (AI): Quantum-enhanced machine learning (QML) is an area of active research. For instance, quantum support vector machines and quantum neural networks might provide speed-ups for specific problems, such as pattern recognition and clustering.

Combinatorial Search: Grover's algorithm provides a quadratic speedup for unstructured search problems, which can be useful in database searching and optimization.

Are Qubits Useful for AI?
Quantum computing has theoretical potential for improving AI, but the field is still nascent. The main ways qubits could help AI are:

Speeding Up Training: Quantum computers might accelerate the training of machine learning models by solving linear algebra problems faster, particularly in areas like matrix inversion and eigenvalue decomposition.

Enhanced Feature Space: Quantum states allow for richer feature spaces, which could improve algorithms like kernel methods.

Efficient Sampling: Quantum computers might perform tasks like sampling from probability distributions more efficiently than classical systems, useful in generative models.

Currently, practical quantum advantage in AI is limited to small, niche applications due to the fragility and scale of existing quantum hardware.

Can Quantum Computers Make a Better Chess Engine?
Chess engines rely on deep search trees and advanced heuristics, which are better suited for classical computers with specialized architectures like GPUs and TPUs. Quantum computing doesn’t naturally lend itself to tree-based search optimizations like Alpha-Beta pruning, nor does it provide significant speedup for deterministic problems.

However, in the future:

Quantum computing might improve heuristic generation or evaluation functions through pattern recognition in large datasets.
Grover’s algorithm could theoretically speed up searching through game states, but this improvement is relatively modest (quadratic).
Your Skepticism:
Your comparison to "adding two height maps" isn't entirely off the mark—quantum computers leverage superposition and interference to process combinations of states efficiently. However, this capability doesn't universally translate to classical computing tasks like general-purpose AI or chess engines. Their utility lies in solving problems where massive parallelism and unique quantum properties offer a true advantage.

Quantum computing is still early in its development, and for now, the "hype-to-useful ratio" is indeed high. But over the next few decades, we might see breakthroughs that change this landscape.

[Dann Corbit]

A quantum computer requires a quantum algorithm. There is a quantum algorithm for search so it could be possible. But it must be remembered that quantum computers are not general computation devices, they are more specialized than a GPU which can be used for a wide variety of problems. They found an algorithm for factoring so RSA is dead, dead, dead as a safe encryption method. But can we use quantum algorithms for difficult non-polynomial problems? Maybe, but only if we can find a quantum algorithm for the problem domain.

[towforce]

A quantum computer requires a quantum algorithm. There is a quantum algorithm for search so it could be possible. But it must be remembered that quantum computers are not general computation devices, they are more specialized than a GPU which can be used for a wide variety of problems. They found an algorithm for factoring so RSA is dead, dead, dead as a safe encryption method. But can we use quantum algorithms for difficult non-polynomial problems? Maybe, but only if we can find a quantum algorithm for the problem domain.

* create a software library for quantum computers

* each time an algorithm is created for a class of problems, add it to the library

* write anther program that allows you to easily describe the nature of the current problem

* this second program will then tell you which algorithm in your quantum software library is best suited to this problem

[JohnWoe]

I haven't seen any practical Quantum Computing stuff. Only some researcher adhoc examples.
If they have solved RSA-algo then all banks are cracked already.
In chess engines you need a list of possible moves. Some algo to determine the best move. Probabilities don't work here. Maybe in there's rain tomorrow by 30% probability type of problems.

Quantum Computing is just university hand waving stuff without any practical value to anybody. For example Deutsch-Jozsa algorithm. To me it seems like a total university BS. So we have a infinite vector of [1, 0]^n the quantum computer converts it into 1 or 0. What's the point of this? Andrew Wiles proved a,b, c has no non-integer solutions for n>2 values. That's a one proof to infinite many totally useless problems.

Quantum Computing is only useful for studying Quantum mechanics. I want to know why quantum mechanics is based on probability? Also why The Universe begun. What is causing gravity? How the first cell appeared? When it turned into multicellular life? What laws apply in the singularity? Etc. Etc.

[mar]

yes, I've never seen a practical implementation of shor's alhorithm either.
a couple of years ago they were bragging they can factorize 2 digit numbers on a qc. RSA must have a good laugh
then there is "post-quantum cryptography", theoretically safe from the quantum scam