The first version was posted 7/29/00.

12/30/00 10AM:  There were two serious errors in the order2guts.m code
that were detected via Jinill Kim and Sunghyun Kim independently
obtaining a second-order solution to the sessionEG.m example.  Those
bugs are now fixed, and the fact that results match for for two
independent solutions on a non-trivial single-state example suggests
that the code is reliable for single-state models.  It is not
thoroughly tested for multi-state models, however, so users are
cautioned.  Anyone who finds anomalies, particularly anyone who finds
results from gensys2.m conflicting with results obtained
independently, should let me (sims@princeton.edu) know.

10/8 10:18PM:  Previous fix of the "pick" bug was incorrect.  Now it is
fixed to the extent that it gives correct answers in the analytically
solvable version of the one-sector growth model.

9/29 9:50PM:  Significant bug in gensys2.m, pointed out by Jinill Kim,
corrected.  The bug made results using a "pick" matrix incorrect.

8/31/00 4PM:  Clarifying changes to the comments in gensys2.m . Again no
change in the code itself.

7/30/00 7:45PM: Another error in the comments to gensys2.m fixed.  Again no
change in the code itself.

7/30/00 7:45PM: Another error in the comments to gensys2.m fixed.  Again no
change in the code itself.

An error in handling dimensions that makes the program stop when the number of
states and number of shocks is different was corrected in the version of
order2guts.m posted 7/30/00, 1:30PM.

An error in the comments to gensys2.m that made matlab help give the wrong
calling sequence was corrected in the version posted 7/30/00, 1:30PM.