[lmi-commits] [lmi] master 1a54e49 1/2: Clarify non-MEC solves

[Greg Chicares]

branch: master
commit 1a54e490ed35a6a2e56aa852d4d23c596477db50
Author: Gregory W. Chicares <gchicares@sbcglobal.net>
Commit: Gregory W. Chicares <gchicares@sbcglobal.net>

    Clarify non-MEC solves
    
    Changed the objective function to
      is_mec ? -1.0 : 1.0
    which expresses the intention clearly, as opposed to the old
      0.5 - is_mec
    which was just too cute. Improved documentation.
---
 ihs_avsolve.cpp | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/ihs_avsolve.cpp b/ihs_avsolve.cpp
index 4e04d98..27f3da5 100644
--- a/ihs_avsolve.cpp
+++ b/ihs_avsolve.cpp
@@ -125,6 +125,11 @@ class SolveHelper
 /// contract ever becomes a MEC. The result is naturally a boundary
 /// value, so it is not interesting to compare it to any solve input.
 ///
+/// Non-MEC solves for loan or withdrawal are untested because they
+/// probably make no sense--see:
+///   https://lists.nongnu.org/archive/html/lmi/2009-02/msg00028.html
+/// so perhaps they should be prohibited.
+///
 /// Non-MEC solves seem acceptably fast despite this two-valued step
 /// function. Other options considered include:
 ///  - Use MEC duration. This inserts a monotone segment into the
@@ -231,7 +236,7 @@ double AccountValue::SolveTest(double a_CandidateValue)
 
     if(mce_solve_for_non_mec == SolveTarget_)
         {
-        return 0.5 - InvariantValues().IsMec;
+        return InvariantValues().IsMec ? -1.0 : 1.0;
         }
 
     return value - SolveTargetCsv_;