Grant: Discovery Grants Program - Individual
University of Waterloo
$20,000.00
Issues in high-dimensional quasi-monte carlo sampling
FY 2012-13
Summary
Suppose a financial analyst wants to price a complex product. The Monte Carlo method can be used for that purpose. It uses random sampling in order to ""simulate"" possible financial scenarios, and for each, the corresponding value of the product under study. By repeating this process several times, a sample of possible values for the product is created, which can then be used for inference, e.g., its mean gives an estimator for the product's price. Quasi-Monte Carlo methods aim at improving this final estimation phase by replacing the random sampling inherent in Monte Carlo by a more structured form of sampling. This improved sampling is based on the use of constructions that attempt to place points in a domain in a very uniform way. These methods have gained a lot of attention in the last 10 to 15 years, as they have proven to be useful on difficult high-dimensional problems in finance, e.g., involving the simulation of several financial assets over long periods of time. More precisely, they can provide estimators with a smaller error than those obtained by applying Monte Carlo, using the same amount of computational effort. Because of this success, these methods are now used in applications that are becoming increasingly complex.
This research proposal focuses on developing new ideas to enhance the performance of quasi-Monte Carlo methods, especially with respect to high-dimensional problems such as those arising in finance, computer graphics, and machine-learning. This will be achieved as follows. Firstly, I plan to work on new quasi-Monte Carlo constructions that work well in high dimensions. Secondly, I wish to study the combination of ideas typically used to improve the quality of estimators derived from Monte Carlo methods with techniques that are specifically suited to enhance the performance of quasi-Monte Carlo methods. Finally, I plan to research quasi-Monte Carlo versions of Markov chain Monte Carlo and sequential Monte Carlo methods, which can be used to perform more sophisticated forms of inference than what is typically done with standard Monte Carlo.
Some of the work achieved while pursuing these objectives will result in the development of computer code that will be made available for free on the internet, so that the advances made in this proposal can be shared easily with practitioners and other researchers.
Program
Discovery Grants Program - Individual
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
238959-2010
Name nom
Lemieux, Christiane
Department département
Statistics and Actuarial Science
Organization id
33
Institution établissement
University of Waterloo
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2012
Competition year année de concours
2010
Award amount
20000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1205
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Mathematical sciences
Area of application fr
Sciences mathématiques
Research subject code
2714
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Mathematics of computing
Research subject fr
Mathématiques de l'informatique
Partie
3
Nb partie
5
Application title
Issues in high-dimensional quasi-monte carlo sampling
Keyword
NSERC
Application summary
Suppose a financial analyst wants to price a complex product. The Monte Carlo method can be used for that purpose. It uses random sampling in order to ""simulate"" possible financial scenarios, and for each, the corresponding value of the product under study. By repeating this process several times, a sample of possible values for the product is created, which can then be used for inference, e.g., its mean gives an estimator for the product's price. Quasi-Monte Carlo methods aim at improving this final estimation phase by replacing the random sampling inherent in Monte Carlo by a more structured form of sampling. This improved sampling is based on the use of constructions that attempt to place points in a domain in a very uniform way. These methods have gained a lot of attention in the last 10 to 15 years, as they have proven to be useful on difficult high-dimensional problems in finance, e.g., involving the simulation of several financial assets over long periods of time. More precisely, they can provide estimators with a smaller error than those obtained by applying Monte Carlo, using the same amount of computational effort. Because of this success, these methods are now used in applications that are becoming increasingly complex.
This research proposal focuses on developing new ideas to enhance the performance of quasi-Monte Carlo methods, especially with respect to high-dimensional problems such as those arising in finance, computer graphics, and machine-learning. This will be achieved as follows. Firstly, I plan to work on new quasi-Monte Carlo constructions that work well in high dimensions. Secondly, I wish to study the combination of ideas typically used to improve the quality of estimators derived from Monte Carlo methods with techniques that are specifically suited to enhance the performance of quasi-Monte Carlo methods. Finally, I plan to research quasi-Monte Carlo versions of Markov chain Monte Carlo and sequential Monte Carlo methods, which can be used to perform more sophisticated forms of inference than what is typically done with standard Monte Carlo.
Some of the work achieved while pursuing these objectives will result in the development of computer code that will be made available for free on the internet, so that the advances made in this proposal can be shared easily with practitioners and other researchers.