User talk:Pcotton
From FinMath
Contents |
[edit] Quick things I think would be useful
Assistance appreciated!
- Improve the speed and accuracy of Code:tenpin#pairwiseImpliedGauss.m
- Improve the performance of Code:normal_util#mat2corr.m which in my humble opinion isn't too crash hot. In particular it often gives worse results than Code:normal_util#corr2factor.m which ought not to be the case.
- Collecting interesting data sets for Code:tenpin/data (including actual bowling data). Also non-bowling data in a similar format. Any set of multivariate binary random variables will do so long as they have non-trivial dependence.
- Write embedded test for Code:matlab_util#File_IO cell2csv.m using textscan.
- Include a decent hash table class (perhaps this one or this).
- Create new and challenging tests for funcompose.m.
- Fix logical bug in funcombine.m and more carefully deal with limits on the length of input/output argument lists (see also at.m).
[edit] Some Wiki things I may get around to
- Autogenerate pdf's from pages called Paper:Something using this or some other Latex template.
- Better implementation of fmuse.m with performance, stability improvements (some files corrupted at present - could use rsync) and functionality (tricks to accomodate private function directories, classes, offline usage).
- Protocols for test scripts which modify a page upon failure (so that one can automatically be alerted of code breaks by placing a watch on such a page).
- Support for R
[edit] Some broad things I'm working on and could use help with
- Overview and taxonomy of credit models and more comprehensive list of references
- Bizarre, interesting and/or realistic examples of "fake" bespoke CDO data.
- Estimation of 3-Copulas.
[edit] Some things on the backburner
- Implementation of multivariate Copula examples such as described in "Simulating exchangeable multivariate archimedean Copulas and it applications". Wu, VAldez and Sherris (2006)
- Implementation of Albanese et al Dynamic credit correlation modeling paper.
