boennecd / pedmod Goto Github PK

R package with quasi-Monte Carlo methods to estimate mixed models commonly used for random effect structures from pedigrees.

R 3.44% TeX 0.27% C 0.17% C++ 8.98% Fortran 0.20% HTML 86.95%

generalized-linear-mixed-models probit threshold-model mixed-models pedigree importance-sampling quasi-monte-carlo graph-partitioning

pedmod's People

Contributors

Stargazers

Watchers

pedmod's Issues

Show warning when the requested number of threads cannot be set.

There needs to be warning on these two lines when the requested number of threads cannot be set:

pedmod/src/r-api.cpp

Line 179 in 051fc85

n_threads = std::min(n_threads, terms_ptr->max_threads);

pedmod/src/r-api.cpp

Line 253 in 051fc85

n_threads = std::min(n_threads, terms_ptr->max_threads);

Support weights

Implement support for case weights to:

support weights when performing non-parametric bootstrap.
support cases where there are only categorical variables and, thus, the potential that multiple log marginal likelihood terms are identical.

README should state that the unshared random effect variance is fixed

It might not be immidealty clear from these lines

pedmod/README.Rmd

Lines 43 to 50 in 051fc85

    
           The $C_{il}$s are known scale/correlation matrices where each of the $l$'th  
        
           types correspond to a type of effect. A  
        
           typical example is that $C_{il}$ is two times the kinship matrix in which  
        
           case we call: 
        
           $$ 
        
           \frac{\sigma_l^2}{1 + \sum_{k = 1}^K\sigma_k^2} 
        
           $$

that

there is an unshared random effect.
the scale parameter / variance for the unshared random effect is fixed to be one.

boennecd / pedmod Goto Github PK

pedmod's People

Contributors

Stargazers

Watchers

pedmod's Issues

Show warning when the requested number of threads cannot be set.

Support weights

README should state that the unshared random effect variance is fixed

Recommend Projects

React

Vue.js

Typescript

TensorFlow

Django

Laravel

D3

Recommend Topics

javascript

web

server

Machine learning

Visualization

Game

Recommend Org

Facebook

Microsoft

Google

Alibaba

D3

Tencent

	The $C_{il}$s are known scale/correlation matrices where each of the $l$'th
	types correspond to a type of effect. A
	typical example is that $C_{il}$ is two times the kinship matrix in which
	case we call:

	$$
	\frac{\sigma_l^2}{1 + \sum_{k = 1}^K\sigma_k^2}
	$$