Tetrad-7.1.3-1

Issue fixes for 7.1.3-1

1. Added CStaR as an non-experimental algorithm. Note that the CStaR and Stability Selection tables only appear if logging is enabled, or else in the terminal/console if Tetrad is launched from the command line or run through causal-cmd.
2. Consolidated BOSS1/2 using a parameter.
3. Made a version of BOSS called BOSS-MB that calculates causes and effects of a variables or graphs over Markov blanket variables plus target. Allows for multiple targets. Added this as an experimental algorithm.
4. Did some work on IMaGES to allow it to specify time series and do boostrapping.
5. Pulled the Zhang-Shen score and the Kim et al. scores (for linear Gaussian high dimensional search) back into the current code.
6. Renamed MBFS (Markov Blanket Fan Search) to "PC-MB" for clarity.
7. Fixed a bug in DagToPag that was generating some reversed edges.
8. Made the FCI complete final orientation ruleset the default throughout Tetrad.
9. Turned on PAG edge coloring for DAG to PAG in GUi.
10. Changed the display of PAG edge coloring; now solid edges are visible, dashed invislble, thick edges definitely direct, think edges not definitely direct.
11. Added cancel buttons to the Random Graph popup in the GUI.
12. Fixed the Paths dialog to get rid of duplicates and rendering issues.
13. Added notation for latent variables to the Paths dialog (putting the latents in parentheses) to make it easy to find latent confounders for PAGs.
14. Fixed the Fruchterman-Reingold layout so that it doesn't hang.
15. Fixed the Bayes PM parameter setter in the GUI. Made 3-valued the default.
16. In Graph displays, put Graph/Edges selector into a tab on the right of the display.
17. For bootstrapping, fixed some counting bugs.
18. For Bayes IM, made random initialization the default.
19. Changed default coefficient range to (-1, 1) throughout.
20. Added some more dialogs to the simulation box to make sure user want to simulate a new model &c.
21. For linear, Gaussian models, made EBIC with gamma 0.8 the default score.
22. Added Kim et al. scores as an optional linear, Gaussian scores, following GIC theory of Kim et al.
23. Some updates to data loading.
24. Some fixes to the data editor.
25. Rescaled the scatterplot, q-q plot, and histogram.
26. Using Zhang's algorithm for PAG to MAG.
27. Putting linear Fisher model back. Still useful for simulating data from large LG models.
    1, Fixed time series simulator in simulation box.
28. Added Erdos-Renyi DAG graph type.
29. Added method to generate times series model automatically within various search algorithms. (Specify a maximum lag, automatically generates forbidden knowledge.)
30. Fixed several bugs in bootstrapping; it appears to work correctly now for all algorithm, except for the time series/bootstrapping combo for IMaGES.
31. Fixed IMaGES so that it can take an arbitrary score.
32. Turned caching back on in the Teyssier scorer (and off in the SEM BIC score), to accommodate different score types.
33. Made several adjustments to the Markov checker.
34. Refactored Knowledge, fixing several bugs.
35. Added some additional statistics to the algorithm comparison and StatsListEditor--notably some PAG comparison to the true DAG statistics (which appear in the GUI when you compare an estimated PAG back to the true DAG with latents).
36. Added an intron optimization to BOSS.
37. Switched from floats to doubles for scores.
38. Fixed the edge comparison dialog so that now it tells the truth.
39. Some cleanup/correction to the algorithm Comparison class.
40. (For coders.) Added DAG to PAG method in SearchGraphUtils for convenience.
41. Added some bidirected edge stats.
42. Added a note in contributors section for SEI contributors.
43. Added code for required edges for GRaSP and BOSS.
44. Added some context menus in Search and Simulation boxes where possible for saving graph images, showing graph properties, etc., directly without having to make a graph box first.
45. (For coders.) Got rid of IKnowledge interface and renamed Knowledge2 to Knowledge throughout (since it was the only remaining implementation of Knowledge.
46. Added some new graph stats to the Graph Properties dialog.
47. Got rid of unnecessary graph displays in the comparison editors. (These were slowing things down.)
48. Got rid of some non-working graph layouts.
49. Fixed edges so that once again for X-->Y and Xo->Y edges the heads of these edges are to the right. Fixed several corresponding such bugs throughout the code.
50. Added an API to set graph types (DAG, PAG, unspecified). When random graphs are generated, they are stamped as DAGs, and then algorithms find PAGs, they are stamped as PAGs.
51. Fixed several errant alternate code notations for < or >.
52. Switched to using the Gamma.logGamma function in the Apache library.
53. Fixed several unit tests.
54. Changed bookmarking in TeyssierScorer. Now you can bookmark once and then to go that same bookmark repeatedly after that.
55. Added function to compute edge probabilities.
56. Added edge probability column to edge table.
57. Fixed label for edge probabilities.
58. Fixed edge coloring.
59. Fixed missing edge properties.
60. Added a menu item to the Graph menu in relevant components to set the graph type to PAG (and therefore do PAG coloring).
61. Set font for logging text area to monospaced to accommodate CStaR.
62. Added edge numbering to bootstrap edge probability table.
63. Added path statistics to Stats List Comparison--ancestor F1, semidirected path F1, no semidirected path F1. Can compare arbitrary graphs, with varying interpretations.
64. Added edge marking statistics to Stats List Comparison--arrow precision, tail (-->) precision, % bidirected edges implying latent confounders, number of edges of various sorts.
65. Updated manual to include recent details.
66. Added Poisson Prior linear, Gaussian score, which uses the Poisson distribution as a prior.
67. Added Zhang Shen Bound linear, Gaussian score, which adapts theory by Zhang and Shen.
68. Added knowledge sensitivity to SP and SP-FCI.
69. Added choice of DAG/CPDAG/PAG for comparison graph for Edge Comparison and Misclassifications Comparison as in Stats List Editor.
70. Added row numbers to Edge Probability Table for bootstrapping.
71. Added menu items in Graph box to save and load graphs in PCALG PAG format.
72. Fixed saving and loading of graphs with additional bootstrap edge probability information in text format.
73. Re-added the parallelized option for the algcomparison API.
74. Set FGES to be single-thread by default; to to multithreading, the flag "parallelized" needs to be set to true (in causal cmd, "--parallelized".
75. Added a seed parameter to the bootstrap API, so that successive bootstrap runs (even with different algorithms) will use the same datasets (if a seed is set).
76. Added a menu item to the Graph box that allows the PAG designation for graphs to turned on or off. When graphs are marked as PAGs, the PAG graph coloring is applied.
77. Fixed the graph comparison type choosers for the edgewise, misclassification, and stats list comparisons so that they remember the last choice you made.
78. In the stats list comparison, moved the new path statistics to the bottom of the list for backward compatibility.
79. Fixed the loading of text graphs with bootstrap information so that they will be compatible with the causal-cmd revision.
80. Made all random graph dialogs cancelable.
81. Updated the date on the manual.