Skip to content

Latest commit

 

History

History
73 lines (43 loc) · 4.67 KB

syntax.md

File metadata and controls

73 lines (43 loc) · 4.67 KB
Raw

Adage Syntax

Here is a reference guide for the Adage functions. These are implemented in the DAG base class and the HistosDAG derived class

Root Operators

template<class O> void Initial(O op)

Inbound root lambda. Allowed O lambdas are of the form:

  • std::function<void(Node*,NodePath*)> (implement in your code as auto op = [/*captures*/](Node *N, NodePath *bins){ /*body*/ }, with no return value since it is void)
  • std::function<void(NodePath*,Node*)>
  • std::function<void(NodePath*)>
  • std::function<void(Node*)>
  • std::function<void()> (no arguments)

We will refer to this set of lambdas as the "standard" lambdas. Use the Node pointer to access the current control node in the traversal, and the NodePath pointer to access the current DAG path of nodes to this control node. See the respective header files and tutorials for how to use these objects in practice.

template<class O> void Final(O op)

Outbound root lambda. Allowed O lambdas are standard (as listed above).

Leaf Operators

template<class O> void Payload(O op)

Payload operator, staged on the leaf node. Since there is no difference between inbound and outbound lambdas on the leaf node, only one staging template function is provided. This Payload function is meant to be used on HistosDAG to act on the stored Histos objects.

Allowed O lambdas are of the form:

  • std::function<void(Histos*)> (implement in your code as auto op = [/*captures*/](Histos *H){ /*body*/ }; this is the most common use case)
  • std::function<void(NodePath*)>
  • std::function<void(Histos*,NodePath*)>
  • std::function<void(NodePath*,Histos*)>
  • std::function<void()>

Similar to the standard set, all of these return void. The first lambda includes a Histos pointer; this is the most common use case, where you want to act on the Histos object on every multi-dimensional bin. If you need the bin path, include the NodePath pointer (though you can also access binning information from the Histos object itself).

template<class O> void LeafOp(O op)

Low-level function to stage a standard lambda O on the leaf operator. It is likely you will find Payload much more useful. LeafOp is wrapped by Payload via a standard lambda function which calls HistosDAG::GetHistos(NodePath*) to access the Histos object in the current multi-dimensional bin specified by the NodePath pointer.

Subloops

template<class O1, class O2> void Subloop(std::vector<TString> layers, O1 opBefore, O2 opAfter)

template<class O> void Subloop(std::vector<TString> layers, O opBefore)

void Subloop(std::vector<TString> layers)

These functions will make a subloop by creating a control node to control the layers listed in std::vector<TString> layers. The layers are iterated through in the specified order. The first Subloop template function is the full function, and the other two are "overloads", so you do not have to specify all (or any) of the lambdas. The inbound lambda will be opBefore and the outbound lambda will be opAfter.

The allowed O1,O2,O lambdas are the standard set, since at the subloop control node you only need access to the current node and/or its node path.

template<class O> void BeforeSubloop(std::vector<TString> layers, O op)

template<class O> void AfterSubloop(std::vector<TString> layers, O op)

These functions allow for an alternative specification of subloops. Again, O is a standard lambda.

Conditional Controls

It can be useful to add conditional controls to subloop control nodes, to allow or disallow the iteration through the subloop. See Adage documentation for more information.

Multi-Payloads

template<class O1, class O2, class O3> void MultiPayload(std::vector<TString> layers, O1 opPayload, O2 opBefore, O3 opAfter)

template<class O1, class O2> void MultiPayload(std::vector<TString> layers, O1 opPayload, O2 opBefore)

template<class O1> void MultiPayload(std::vector<TString> layers, O1 opPayload)

These three (overloaded) functions allow for the creation of multiple subloops, each with their own payload. The inbound and outbound lambdas, opBefore and opAfter are standard lambdas, and the payload lambda opPayload must have the same type you would use with the Payload staging function.

template<class O1, class O2, class O3> void MultiLeafOp(std::vector<TString> layers, O1 opLeaf, O2 opBefore, O3 opAfter)

This is the low-level function wrapped by MultiPayload; it is not recommended to use this function directly. Behind the scenes, it will build the requested subloop, and then will call Payload(opLeaf) in the inbound lambda (along with opBefore), which overwrites any staged payload.