** Instruction of qmulti1.c

0. NOTE

Little error handling is implemented in the program. Thus it may
cause erroneous results; please use the program wisely. You may
freely modify or rewrite the program so that you can accomplish your purpose. 

The program may not converge in some situations, where
you can obtain its convergence by modifying the starting value
of the iterative procedures.

1. How to compile?

gcc -o qmulti qmulti1.c -lm

2. How to use?

./qmulti population_size cell_total < size_index_filename,

where "population_size" implies the size of the corresponding population
to be assessed. "cell_total" implies
the total number of cells (J). "size_index_filename"
implies the name of the file that contains size indices.

For example,

./qmulti 4000 3420 < argus.f

produces the result of Hoshino (2008, Section 5). By changing 4000
for a suitable number, you can estimate population size indices.

3. How to describe size index information?

Just write s_i in the i-th line of a file. See an example file
(argus.f) provided in this site.
The last line of such a file is supposed to be of
the largest cell; namely, the last line should be other than 0.

4. Referece

Hoshino, N. (2008) The quasi-multinomial distribution as a tool
for disclosure risk assessment. To appear in the Journal of Official
Statistics.

---
Nobuaki Hoshino