Performing queries#

There are a total of three steps in fully specifying a query:

  1. Definition (concrete type and constructor arguments) of the query itself.

  2. Input column(s) with which it is filled with.

  3. Associated selection(s) at which it is performed.

Template
auto q = df.get(query::output<DEF>(ARGS...))
             .fill(COLS...)
             .at(SELS...);

Defining a query#

Call dataflow::get() specifying the definition and constructor arguments of the query.

using h1d = qty::boost::histogram::histogram<double>;
using linax = qty::boost::histogram::axis::regular;

auto q1 = df.get(query::output<h1d>(linax(100, 0.0, 1.0)));

Filling a query#

A defined query can be fill()ed with input columns:

auto q1x = q.fill(x);

A query can be filled multiple times, as long as the dimensionality of each fill is valid:

auto q1xy = df.get(query::output<h1d>(linax(10, 0.0, 1.0)))(x)(y);

using h2d = qty::boost::histogram::histogram<double, double>;
auto q2xy =
    df.get(query::output<h2d>(linax(10, 0.0, 1.0), linax(10, 0.0, 1.0)))(x, y);

Booking a query#

Booking a query at a particular selection fully instantiates the lazy query:

auto all = df.filter(column::constant(true));
auto q1x_all = q1x.at(all);

A given query can be booked at multiple selections:

auto [q1x_a, q1x_b] = q1x.at(cut_a, cut_b);

Or, multiple different queries can be booked from a selection:

auto [q1x_c, q2xy_c] = cut.book(q1x, q2xy);

Accessing results#

Access the result of any query to trigger the dataset traversal for all.

auto h1x_a = q1x_a.result(); // takes a while
auto h2xy_c = q2xy_c.result(); // instantaneous