[R] Fitting a distribution involving a cusum

[Dr Andrew Wilson]

Sorry if this is a naive question, but I only started using R last
week.

I would like to fit a distribution to data which includes a cumulative
sum:

P(r) = \sum^{V}_{i=r} D e^{-n/i} i^{-d}

This equation is used to fit rank-frequency data: P(r) is the frequency of
an entity with rank r; V is the number of unique entity types; and D, n
and d are constants to be estimated.

I've found out how to fit nonlinear distributions generally (using nls)
but I don't know how to include the cumulative sum in the formula.  Could
anyone help me?

Many thanks,
Andrew Wilson