gleiss / saturation-visualization Goto Github PK

remove each node which is never active and is not transitive parent of any node which is active at some point.

Would be great to parse --proof_extra full to display age and weight of a selected clause in the graph.

remove each node which is either a theory axiom or only derived from theory axioms.

Our nodes should correspond to clauses, together with the following information:

• whether it occured as a final preprocessing clause
• the point at which it was added to passive
• the point at which it was added to active if such a point exists

We want to change the Node-class accordingly:

• Keep the fields number, clause, inference_rule, parents and children
• Add fields passiveTime and activeTime, denoting the time when the clause was added to passive/active
• Add field isFromPreprocessing for those clauses occuring in the final preprocessing
• modify sanity check accordingly

currently width changes depending on width of search results

The right column of the tool currently contains two blocks of functionality:

1. functionality to get a new graph and
2. functionality to find/select certain nodes.

We want to add a third block, which displays information (name, number and inference) about the currently selected node (if exactly one node is selected).

(Later this third block will also get functionality to change how the node is displayed)

remove each node which is not in nodes and doesn't have a transitive parent node which is in nodes.

remove each node which is not in nodes and is not a transitive parents of a node in nodes.

The proof output is a series of events, where each event has one of the following forms:

• [PP] final: Clause
• [SA] passive: Clause
• [SA] active: Clause

where Clause is of the form

• id, the clause formula, some statistics of the form (a : b : c), the inference and the parent ids, e.g. 67. ~$less(sK0,0) (0:3:1) [cnf transformation 57].

We want to parse these events into the modified Node-class.
To compute passiveTime and activeTime, we should use a counter which tracks the number of already parsed passive and active lines / events.

Makes it possible to click through the search result.

Would be great to display the name of the uploaded file as I use multiple tabs with different vampire outputs to compare!

We use different visual representations for different nodes, e.g. for hidden nodes, nodes from preprocessing, theory axioms, ...

It would be good if we have an enum of all the possible representations (hidden, preprocessing, theoryAxiom, passive, active, maybe other stuff in the future).

We then can refactor the main functionality of get_layout into two sub-methods:

• computeRepresentation(node, history_state), which returns for each node the appropriate representation
• formatNode(node, representation) / formatEdge(node, representation), which returns the json-encoding for a node and a given representation

Similar to the already existing "select parents"-button, but for selecting children.

The children-ids of a node can be obtained using the already implemented children()-method from the Dag-class.

The user should be able to highlight certain nodes using a checkbox, which is located in the third block of the right column.
Highlighted nodes could be colored e.g. red or orange.

Currently get_layout uses the node-id to determine whether it should display the node or not.
We want to change this to:

• if the passive-time is smaller or equal the history_state, display the node
• depending on whether the active-time (if it exists) is smaller or equal the history_state, display the node as passive node or as active node (passive nodes should be in the background, I expect that we are mainly interested in the active nodes, and only look at the passive nodes from time to time)

Add a button to the find/select block, which selects all common consequences of a set of selected nodes.

Given a set of ids, one can get the id of common consequences using the findCommonConsequences()-method already implemented in the file search.py.