* Bug #9840 fixed - Default G tolerance for lsqrsolve() was too large 27/12427/4
Paul BIGNIER [Wed, 4 Sep 2013 13:57:11 +0000 (15:57 +0200)]
Change-Id: I956f6d3e41a37c7e21b8d6195c592bf3619a5480

scilab/CHANGES_5.5.X
scilab/modules/optimization/sci_gateway/fortran/intlsqrsolve.f

index 439263c..e64015d 100644 (file)
@@ -91,6 +91,8 @@ Scilab Bug Fixes
 
 * Bug #9701 fixed - optim with "qn" option was failing for large problems.
 
+* Bug #9840 fixed - Default G tolerance in lsqrsolve was too large.
+
 * Bug #10012 fixed - lmisolver and lmitool called with no input now produce errors.
 
 * Bug #10175 fixed - Clearer example added for sp2adj to adj2sp conversion (and backwards conversion).
index adb43dd..130f1e9 100644 (file)
@@ -91,7 +91,7 @@ c     checking variable tol (number 5)
       else
          ftol=1.d-8
          xtol=1.d-8
-         gtol=1.d-5
+         gtol=1.d-8
          maxfev=1000
          epsfcn=0
          factor=100.0d0
@@ -219,5 +219,3 @@ c
  999  continue
       return
       end
-
-