zhouyan / vsmc Goto Github PK

The current code involves much dynamic allocations which can be avoided

It would be very useful to have a file showing a complete example of how vSMC would be used to filter a dataset. An "introduction" page in the documentation would also be very welcome. All of this significantly reduces the learning curve for using the library.

Since I am not very good with cmake, I maybe asking too simple question but, may I ask for more information on what packages should be available in the system (like ubuntu). As far as my cmake go it generates makefile with no problem and OpenMP as available backend but when I compile the examples I get errors like these (and many more):

common.hpp:167:15: error: expected nested-name-specifier before ‘distribution_type’ using distribution_type = Name##Distribution

aligned_memory.hpp:359:1: error: expected unqualified-id before ‘using’
using Vector = typename std::conditionalstd::is_scalar<T::value,

Should I have done anything specific before starting to compile?

in sampler: ess, resample, should always have the same size,
monitor index and monitor record should also always have the same size

Hi Yan, do you have the supplementary material for CESS?

Now std::size_t is used exclusively. This should be replace with proper size_type in future. Since vSMC depends on vDist, it is important that the length of possible internal arrays are something vDist can handle.

For now it shall not be a big problem, as one is not likely to create samplers with particle number larger than an int

Currently some interface like wreight return reference to an eigen vector. Others like record of monitor get std::vector. They shall all be eigen vector in future