error using compute rdf

Dear Dr. Axel,
I now get which version you mean I should download. I downloaded the tarball. I have also unzipped the tarball but there is no read_me file in it. I am using windows. the only file present in the unzipped tarball is the examples file. How do you recommend I should proceed?
I sincerely appreciate your help.

Dear Dr. Axel,
I now get which version you mean I should download. I downloaded the
tarball. I have also unzipped the tarball but there is no read_me file in
it. I am using windows. the only file present in the unzipped tarball is
the examples file. How do you recommend I should proceed?

​LAMMPS is distributed as source code that needs to be compiled.
pre-compiled windows (or linux, etc) binaries are occasionally provided as
a courtesy by different people, but since none of the current core LAMMPS
developers uses windows, there is no incentive to build and compile windows
binaries for every patch level (or support one of the canonical windows
compiler toolchains like visual c++).​ in fact, the windows binaries are
built on linux using a cross-compiler toolchain.

you have multiple options:
- compile LAMMPS from source on windows. general compilation instructions
are in the LAMMPS manual. but they require you set up a unix-like
environment and know what you are doing.
- install a linux system inside a virtual machine and compile/install
LAMMPS on linux there.
- use your windows version of LAMMPS, but then collect your trajectory as
is, and then switch the pair style against one where you can set the
cutoff easily (e.g. lj/cut), and then do your g(r) computation using the
rerun command after the fact
- use a standalone postprocessing tool that computes g(r) (actually a good
practice to write one yourself).

axel.