Test failure due to Eigen version change

[ihnorton]

      Start 11: UKFTractography_2T_fw_Test
11/29 Test #11: UKFTractography_2T_fw_Test .............................   Passed    0.24 sec
      Start 12: UKFTractography_2T_fw_Test_Compare
12/29 Test #12: UKFTractography_2T_fw_Test_Compare .....................***Failed    0.11 sec

Reverting to the prior version of Eigen "fixes" the issue.

Looking at the resulting files, the output is identical for the first ~50 points (out of 59), and then diverges slightly. Will investigate.

[ihnorton]

I looked at this a bit more. The trajectory is diverging at step 139 in the 2T_fw_Test. Unsure of the reason now, as it happens in both Eigen versions.

[ihnorton]

I accidentally replicated this on Mac while refactoring with a few parameters unset:

diff --git a/tmp/d_refactored_branch b/tmp/d_master
index 04458b6..8f2fc7f 100644
--- a/tmp/d_b
+++ b/tmp/d_m
@@ -1,10 +1,10 @@
-(Tractography) $1 = {
+(Tractography) $2 = {
   _ukf = size=0 {}
-  _model = 0x0000000110851fa0
+  _signal_data = 0x70706f74732d2d18
+  _model = 0x000000010ef2d300
   _filter_model_type = _2T_FW
   _output_file = "/Users/inorton/work/git/ukftractography-build/Testing/Temporary/2T_fw_fiber.vtk"
   _output_file_with_second_tensor = ""
-  _signal_data = 0x0000000000000000
   _record_fa = false
   _record_nmse = true
   _record_trace = false
@@ -19,10 +19,10 @@
   _store_glyphs = false
   _branches_only = false
   _is_branching = false
-  _p0 = 0
+  _p0 = 0.01
   _sigma_signal = 0
-  _sigma_mask = 0
-  _min_radius = 0
+  _sigma_mask = 0.5
+  _min_radius = 0.87
   _max_length = 834
   _full_brain = false
   _noddi = false

I think it's probably unrelated, but putting up as a reminder to closely check the contents of the object on Windows.