JBBollen

I have problems with rounding floating point numbers to 2 decimals using cnmat.round. The results are quite unpredictable: with an input of the standard Max representation of 6 decimals, results higher than 16.025998 or so (!!!) are sometimes rounded, at other times not... Although sprintf or the multiply/divide trick will not give me the same "nearest" results I'm after (since I need "round-up" or round-down"), I put sprintf in the patch below to show that it behaves similarly erratic. I was wondering if anybody could reproduce what I am experiencing?

I have problems with rounding floating point numbers to 2 decimals using cnmat.round. The results are quite unpredictable: with an input of the standard Max representation of 6 decimals, results higher than 16.025998 or so (!!!) are sometimes rounded, at other times not... Although sprintf or the multiply/divide trick will not give me the same "nearest" results I'm after (since I need "round-up" or round-down"), I put sprintf in the  patch below to show that it behaves similarly erratic. I was wondering if anybody could reproduce what I am experiencing?
Thanks,
JB

Strange behaviour of rounded floating point output.