Hyper-quicksort: energy efficient sorting via the Templar framework for Template Method Hyper-heuristics

Jeremiah Swan; Nathan John Burles

Hyper-quicksort: energy efficient sorting via the Templar framework for Template Method Hyper-heuristics

Computer Science

Research output: Contribution to conference › Other

Abstract

Scalability remains an issue for program synthesis:
- We don’t yet know how to generate sizeable algorithms from scratch.
- Generative approaches such as GP still work best at the scale of expressions (though some recent promising results).
- Formal approaches require a strong mathematical background.
- ... but human ingenuity already provides a vast repertoire of specialized algorithms, usually with known asymptotic behaviour.
Given these limitations, how can we best use generative hyper-heuristics to improve upon human-designed algorithms?

Original language	English
Publication status	Published - 2015
Event	39th CREST Open Workshop: Measuring, Testing and Optimising Computational Energy Consumption - UCL, London, United Kingdom Duration: 23 Feb 2015 → 24 Feb 2015

Workshop

Workshop	39th CREST Open Workshop: Measuring, Testing and Optimising Computational Energy Consumption
Country/Territory	United Kingdom
City	London
Period	23/02/15 → 24/02/15

Access to Document

hyper-quicksort-energy-efficient-sorting-via-the-templar-framework-for-template-method-hyper-heuristicsAccepted author manuscript, 414 KB

Cite this

@conference{5a18f200cce043c58e2ba1556d84f3bf,

title = "Hyper-quicksort: energy efficient sorting via the Templar framework for Template Method Hyper-heuristics",

abstract = "Scalability remains an issue for program synthesis: - We don{\textquoteright}t yet know how to generate sizeable algorithms from scratch. - Generative approaches such as GP still work best at the scale of expressions (though some recent promising results). - Formal approaches require a strong mathematical background. - ... but human ingenuity already provides a vast repertoire of specialized algorithms, usually with known asymptotic behaviour. Given these limitations, how can we best use generative hyper-heuristics to improve upon human-designed algorithms?",

author = "Jeremiah Swan and Burles, \{Nathan John\}",

year = "2015",

language = "English",

note = "39th CREST Open Workshop: Measuring, Testing and Optimising Computational Energy Consumption ; Conference date: 23-02-2015 Through 24-02-2015",

}

TY - CONF

T1 - Hyper-quicksort: energy efficient sorting via the Templar framework for Template Method Hyper-heuristics

AU - Swan, Jeremiah

AU - Burles, Nathan John

PY - 2015

Y1 - 2015

N2 - Scalability remains an issue for program synthesis: - We don’t yet know how to generate sizeable algorithms from scratch. - Generative approaches such as GP still work best at the scale of expressions (though some recent promising results). - Formal approaches require a strong mathematical background. - ... but human ingenuity already provides a vast repertoire of specialized algorithms, usually with known asymptotic behaviour. Given these limitations, how can we best use generative hyper-heuristics to improve upon human-designed algorithms?

AB - Scalability remains an issue for program synthesis: - We don’t yet know how to generate sizeable algorithms from scratch. - Generative approaches such as GP still work best at the scale of expressions (though some recent promising results). - Formal approaches require a strong mathematical background. - ... but human ingenuity already provides a vast repertoire of specialized algorithms, usually with known asymptotic behaviour. Given these limitations, how can we best use generative hyper-heuristics to improve upon human-designed algorithms?

M3 - Other

T2 - 39th CREST Open Workshop: Measuring, Testing and Optimising Computational Energy Consumption

Y2 - 23 February 2015 through 24 February 2015

ER -