- allow jagged arrays of covariance keys - simplify
0.4
- Fixes 5 value error if covariance keys have multiple observations - fixes scaling covariance cross terms in addition to diagonals
0.3.3
- Fixes 4, ValueError if just one observation
0.3.2
- fixes 2: figure out nobs automatically - fixes 3: sparse matrices - adds multiprocessing pool - uses pvlib instead of proprietary solar utils - travis build
0.3.1
- Fixes 1 works with Pint's ureg.wraps() now - use indices for positional args - don't use inspect.argspec - test Jacobian estimates using AlgoPy and numdifftools - update docs with more examples, show Jacobian erros vs AlgoPy
0.3
- specify covariance keys, group args and returns - covariance is dimensionless fraction - more complex examples, test with Python extension and with multiple input arguments and returns