Parallel Computing for Data Science

60 CHAPTER 3. SCHEDULING

# re t u rn va l u e :

# R matrix , showing a d j u s t e d R−s quared va l u e s ,

# in d e x e d by p r e d i c t o r s e t

mcapr <− function ( x , y , k , ncore s , r e v e r s e=F ,

dyn=F , chunk=1) {

require ( p a r a l l e l )

# add 1 s column t o X

x <− cbind ( 1 , x )

# f i n d X’X, X’Y

xpx <− crossprod ( x , x )

xpy <− crossprod ( x , y )

# ge n er a t e matri x o f p r e d i c t o r s u b s e t s

allcombs <− genall c ombs ( ncol ( x)−1 ,k )

ncombs <− length ( allcombs )

# s e t up t a s k i n d i c e s

t a s k s <− i f ( ! r e v e r s e ) seq ( 1 , ncombs , chunk ) else

seq ( ncombs ,1 , −chunk )

out <− mclapply ( tasks ,