memory efficient grids during variogram computation

[basaks]

For large problems this will conserve memory during variogram computation.

[rth]

The code update looks good to me; thanks!

Not so sure about the .travis.yml changes though. The Miniconda-latest-Linux-x86_64.sh script points to,

Miniconda2-4.0.5-Linux-x86_64.sh	25.9M	2016-03-29 19:32:26	42dac45eee5e58f05f37399adda45e85

which is fairly old. To get the latest versions, I think we should use Miniconda3-latest-Linux-x86_64.sh (or Miniconda2-latest-Linux-x86_64.sh if we really want PY2).

+1 for using the conda virtual environement. We had this setup before it was removed in PR #17 as for some reason the system python was used even after activating the virtual environment, as far as I remember. Just mentioning it in case it matters for this PR (I don't think it does)...

[basaks]

@rth thank you. Very useful and you attention to detail is amazing :)
Please merge when you are happy.

[basaks]

To get the latest versions, I think we should use Miniconda3-latest-Linux-x86_64.sh (or Miniconda2-latest-Linux-x86_64.sh if we really want PY2).

I think we can create python3.x envs using mimiconda2.sh.
See how I created py2.7 env using it: https://travis-ci.org/bsmurphy/PyKrige/jobs/204137080#L370
and also py3.5 here: https://travis-ci.org/bsmurphy/PyKrige/jobs/204137078#L383.

I did not know anaconda2 could do python3.x envs before I created this PR :)

[rth]

Thanks, everything looks good!

I did not know anaconda2 could do python3.x envs before I created this PR :)

Yeah, that is very practical for testing PY2/PY3 versions...