Test failed (16 31)

[LinYuChi]

Extended from the RobOptim forum

OS: Ubuntu 14.04
Compiler: gcc 4.8.4
Eigen 3.2.0
branch: master
commit hash: ff97690

16(Problem) Output:

Running 2 test cases...
log4cxx: Could not open file [/home/yuchi/RobOptim/roboptim-core/tests/log4cxx.xml].
[4,2]((100,0), (0,0.01), (10,0), (0,-0.4))
[4,2]((1,0), (0,0.5), (1,0), (0,-2))
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-5, 5), (-10, inf)
  Arguments scaling: 0.01, 0.5
  Arguments names: x₀, x₁
  Number of constraints: 3
  Constraint 0
      Numeric linear function:
        A = [2,2]((100,0), (0,0.01))
        B = [2](200,-30)
      Bounds: (-inf, inf), (-inf, inf)
      Scaling: 1, 100
      Initial value: [2](200, -30)
  Constraint 1
      Numeric linear function:
        A = [2,2]((10,0), (0,-0.4))
        B = [2](2,3)
      Bounds: (0, inf), (0, inf)
      Scaling: 10, 10
      Initial value: [2](2, 3)
  Constraint 2
      a * b, a + b (not differentiable)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](0, 0)
  Starting point: [2](0,0)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-inf, inf), (-inf, inf)
  Arguments scaling: 1, 1
  No constraints.
  No starting point.
  Infinity value (for all functions): inf
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-5, 5), (-10, inf)
  Arguments scaling: 0.01, 0.5
  Arguments names: x₀, x₁
  No constraints.
  Starting point: [2](0,0)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[2](0,0)
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-5, 5), (-10, inf)
  Arguments scaling: 0.01, 0.5
  Arguments names: x₀, x₁
  No constraints.
  Starting point: [2](200,30)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[2](195,0)
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-inf, 5), (-inf, inf)
  Arguments scaling: 1, 1
  Arguments names: x₀, x₁
  Number of constraints: 2
  Constraint 0
      Constant function
        offset = [2](3,-1)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](3, -1)
  Constraint 1
      Numeric linear function:
        A = [2,2]((10,0), (0,-0.4))
        B = [2](2,3)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](2, 3)
  Starting point: [2](0,0)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[6](0,0,0,0,0,0)
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-inf, 5), (-inf, inf)
  Arguments scaling: 1, 1
  Arguments names: x₀, x₁
  Number of constraints: 2
  Constraint 0
      Constant function
        offset = [2](3,-1)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](3, -1)
  Constraint 1
      Numeric linear function:
        A = [2,2]((10,0), (0,-0.4))
        B = [2](2,3)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](2002, -17) (constraint not satisfied)
  Starting point: [2](200,50)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[6](195,0,0,0,1960,-5)

log4cxx: Could not open file [/home/yuchi/RobOptim/roboptim-core/tests/log4cxx.xml].
[4,2]((100,0), (0,0.01), (10,0), (0,-0.4))
[4,2]((1,0), (0,0.5), (0.1,0), (0,-0.2))
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-5, 5), (-10, inf)
  Arguments scaling: 0.01, 0.5
  Arguments names: x₀, x₁
  Number of constraints: 3
  Constraint 0
      Numeric linear function:
        A = [2,2]((100,0), (0,0.01))
        B = [2](200,-30)
      Bounds: (-inf, inf), (-inf, inf)
      Scaling: 1, 100
      Initial value: [2](200, -30)
  Constraint 1
      Numeric linear function:
        A = [2,2]((10,0), (0,-0.4))
        B = [2](2,3)
      Bounds: (0, inf), (0, inf)
      Scaling: 10, 10
      Initial value: [2](2, 3)
  Constraint 2
      a * b, a + b (not differentiable)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](0, 0)
  Starting point: [2](0,0)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-inf, inf), (-inf, inf)
  Arguments scaling: 1, 1
  No constraints.
  No starting point.
  Infinity value (for all functions): inf
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-5, 5), (-10, inf)
  Arguments scaling: 0.01, 0.5
  Arguments names: x₀, x₁
  No constraints.
  Starting point: [2](0,0)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[2](0,0)
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-5, 5), (-10, inf)
  Arguments scaling: 0.01, 0.5
  Arguments names: x₀, x₁
  No constraints.
  Starting point: [2](200,30)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[2](195,0)
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-inf, 5), (-inf, inf)
  Arguments scaling: 1, 1
  Arguments names: x₀, x₁
  Number of constraints: 2
  Constraint 0
      Constant function
        offset = [2](3,-1)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](3, -1)
  Constraint 1
      Numeric linear function:
        A = [2,2]((10,0), (0,-0.4))
        B = [2](2,3)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](2, 3)
  Starting point: [2](0,0)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[6](0,0,0,0,0,0)
Problem:
  Constant function
    offset = [2](0,0)
  Objective scaling: 1, 1
  Arguments bounds: (-inf, 5), (-inf, inf)
  Arguments scaling: 1, 1
  Arguments names: x₀, x₁
  Number of constraints: 2
  Constraint 0
      Constant function
        offset = [2](3,-1)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](3, -1)
  Constraint 1
      Numeric linear function:
        A = [2,2]((10,0), (0,-0.4))
        B = [2](2,3)
      Bounds: (-12, 42), (-12, 42)
      Scaling: 0.1, 10
      Initial value: [2](2002, -17) (constraint not satisfied)
  Starting point: [2](200,50)
  Starting value: [2](0,0)
  Infinity value (for all functions): inf
[6](195,0,0,0,1960,-5)

/home/yuchi/RobOptim/roboptim-core/tests/problem.cc(263): error in "problem<N8roboptim17EigenMatrixSparseE>": check output->match_pattern () failed. Mismatch at position 66
...0.1,0...
...1,0),...

*** 1 failure detected in test suite "Master Test Suite"

31(Problem) Output:

Running 2 test cases...
log4cxx: Could not open file [/home/yuchi/RobOptim/roboptim-core/tests/log4cxx.xml].
x = [2](0,0)
λ = [4](0,0,0,1)
LICQ conditions: not satisfied
  rank: 0
  max_rank: 1
KKT conditions: satisfied
  ∇L(x*,λ*) = [2](0,0)
  λ = [4](0,0,0,1)
  Constraint violation: 0
  Complementary slackness: 0
  Dual feasible: true
Null gradient condition: not satisfied
  - Constraint: x * x (differentiable function)
    Null gradient indices: 0
x = [2](0,3)
λ = [4](1,1,-1,1)
LICQ conditions: not satisfied
  rank: 1
  max_rank: 2
KKT conditions: not satisfied
  ∇L(x*,λ*) = [2](1,1)
  λ = [4](1,1,-1,1)
  Constraint violation: 0
  Complementary slackness: 0
  Dual feasible: true
Null gradient condition: satisfied
x = [2](-1,1)
λ = [4](1,1,-1,1)
LICQ conditions: satisfied
  rank: 1
  max_rank: 1
KKT conditions: not satisfied
  ∇L(x*,λ*) = [2](1,1)
  λ = [4](1,1,-1,1)
  Constraint violation: 0
  Complementary slackness: 0
  Dual feasible: true
Null gradient condition: satisfied
Problem:
  x * x (differentiable function)
  Objective scaling: 1
  Arguments bounds: (-1, 3), (-1, 3)
  Arguments scaling: 1, 1
  Arguments names: x₀, x₁
  Number of constraints: 1
  Constraint 0
      x * x (differentiable function)
      Bounds: (0, 9)
      Scaling: 1
      Initial value: [1](0)
  Starting point: [2](0,0)
  Starting value: [1](0)
  Infinity value (for all functions): inf

log4cxx: Could not open file [/home/yuchi/RobOptim/roboptim-core/tests/log4cxx.xml].
/home/yuchi/RobOptim/roboptim-core/tests/result-analyzer.cc(180): error in "result_analyzer<N8roboptim17EigenMatrixSparseE>": check ngd failed
x = [2](0,0)
λ = [4](0,0,0,1)
LICQ conditions: not satisfied
  rank: 0
  max_rank: 1
KKT conditions: satisfied
  ∇L(x*,λ*) = [2](0,0)
  λ = [4](0,0,0,1)
  Constraint violation: 0
  Complementary slackness: 0
  Dual feasible: true
Null gradient condition: not satisfied
  - Constraint: x * x (differentiable function)
    Null gradient indices: 0
x = [2](0,3)
λ = [4](1,1,-1,1)
LICQ conditions: not satisfied
  rank: 1
  max_rank: 2
KKT conditions: not satisfied
  ∇L(x*,λ*) = [2](1,1)
  λ = [4](1,1,-1,1)
  Constraint violation: 0
  Complementary slackness: 0
  Dual feasible: true
Null gradient condition: not satisfied
  - Constraint: x * x (differentiable function)
    Null gradient indices: 0
x = [2](-1,1)
λ = [4](1,1,-1,1)
LICQ conditions: satisfied
  rank: 1
  max_rank: 1
KKT conditions: not satisfied
  ∇L(x*,λ*) = [2](1,1)
  λ = [4](1,1,-1,1)
  Constraint violation: 0
  Complementary slackness: 0
  Dual feasible: true
Null gradient condition: satisfied
Problem:
  x * x (differentiable function)
  Objective scaling: 1
  Arguments bounds: (-1, 3), (-1, 3)
  Arguments scaling: 1, 1
  Arguments names: x₀, x₁
  Number of constraints: 1
  Constraint 0
      x * x (differentiable function)
      Bounds: (0, 9)
      Scaling: 1
      Initial value: [1](0)
  Starting point: [2](0,0)
  Starting value: [1](0)
  Infinity value (for all functions): inf

/home/yuchi/RobOptim/roboptim-core/tests/result-analyzer.cc(213): error in "result_analyzer<N8roboptim17EigenMatrixSparseE>": check output->match_pattern () failed. Mismatch at position 617
...not s...
...satis...

*** 2 failures detected in test suite "Master Test Suite"

[LinYuChi]

I upgraded eigen to 3.2.1 and get the same result.

Also, I installed Eigen 3.2.94 standalone. If you know how to substitute the eigen include path in cmake, I would like to have another test. The part finding eigen include path is wrapped in a macro, and I do not know how to manually specify the include path in this way. Thank you.

[bchretien]

The error happens with sparse matrices, and older Eigen version had several bugs related to such matrices (cf. this for instance). In order to use your recent Eigen version, can you try setting PKG_CONFIG_PATH accordingly? (cf. man pkg-config).

[LinYuChi]

My installation of Eigen 3.2.94 has some problem, so I finally use the Eigen 3.2.5 debian package. Now I can pass test for core, trajectory and Ipopt-plugin. However, I would suggest set a Eigen version requirement in cmake to prevent other users from facing this problem again.

[bchretien]

Indeed, thanks for the feedback!