multi_index | Minimal boost : :multi_index implementation | Build Tool library

I am in a situation where I am forced to use a std::vector container as my underlying data structure. I'm trying to exploit boost::multi_index::sequenced<> to shadow the vector offset, and provide a mechanism for richer data queries. I really don't want to unnecessarily copy all the data from one container to another.

In the example code snippet below I have a class BarInterface that manages the insertion and removal of elements to the Bar::foos container. Initially I tried to store references to the vector elements as elements of the typedef boost::multi_index_container, but these are not stable against insertions.

What I'd like to do is have a custom key extractor that is a function of the _bar container and the boost::multi_index::sequenced<> index. So, for example to get the name of an element, I'd find the sequence offset x, and then _bar.foos[x].name. So, I'm really just using boost::multi_index as a proxy for richer queries against a vector of variable size.

Say I have some boost graph

What I want to to: Let's say I have data in form of a 1D array. After fitting that data, (scipy.optimize.curve_fit), this will be reduces to a skaler/0D array. So far so good. That is the easy part.

The problem is, that the data is not actually in 1D, but (n+1)D. So I will have to iterate over the whole array over all axis but one, take a 1D slice, fit that slice and write this into a new array with n dimensions. For the sake of simplicity I used the sum function instead of fitting in this example code.

I have a piece of code which iterates over a three-dimensional array and writes into each cell a value based on the indices and the current value itself:

I have multi index container like below

I've a dataframe with the following structure:

What I need to do is to "fill" the time_hour gaps with the nearest previous value (time) according to their company and product_type. In this case, for Adidas, a row for the 04:00 is missing so it'll need to be filled with 4, the value from 03:00 sales.

I know how to do it in the case of using a datetime as the unique index but this multi_index setting is something I couldn't solve for the moment.

I have a three-dim Series like this

In my code I need to have a functionality to iterate over all elements and check if there some element already exists possibly as soon as possible, so my choice fell on boost multi index container where I can use vector and unordered_set interface for my class Animal at the same time. The problem is that I am not able to find some element through unordered_set interface since I replaced key from std::string to std::array and adjusted the code, and I don't know what I am doing wrong ?

code: https://wandbox.org/permlink/dnCaEzYVdXkTFBGo

Is it possible to use lambda for hashing in hashed__unique interface for boost::multi_index? See this example: https://godbolt.org/z/1voof3

I also saw this: How to use lambda function as hash function in unordered_map? where the answer says:

You need to pass lambda object to unordered_map constructor since lambda types are not default constructible.

and I'm not sure is it even possible to do for the given example on godbolt.