search

pele-python / mcpele

Monte Carlo and parallel tempering routines built on the pele foundation
Other
20 stars 5 forks source link

Histogram has no error bars on bin height #12

Closed kjs73 closed 10 years ago

kjs73 commented 10 years ago

Suppose the histogram has N entries. Suppose that bin i has n_i entries. The fraction of entries in bin i is f_i = n_i / N The error bar on the height (histogram is assumed normalised such that sum of all bin areas is unity) is then (from the expected error of the estimated mean of the binomial distribution to be in bin i): sqrt( f_i *(1-f_i)/ N ) / delta where delta is the bin width: delta = (xmax - xmin)/N

Since this can be computed in the end, basically without extra cost or bookkeeping, we could consider adding it.

smcantab commented 10 years ago

Sure, definitely worth adding it. On 5 Jul 2014 08:48, "kjs73" notifications@github.com wrote:

Suppose the histogram has N entries. Suppose that bin i has n_i entries. The fraction of entries in bin i is f_i = n_i / N The error bar on the height (histogram is assumed normalised such that sum of all bin areas is unity) is then (from the expected error of the estimated mean of the binomial distribution to be in bin i): sqrt( f_i *(1-f_i)/ N ) / delta where delta is the bin width: delta = (xmax - xmin)/N

Since this can be computed in the end, basically without extra cost or bookkeeping, we could consider adding it.

— Reply to this email directly or view it on GitHub https://github.com/pele-python/mcpele/issues/12.