Error when taking max value of Eigen tensor

[JamieAl]

I am trying to get the max value of an Eigen tensor, something equivalent to max(max(max(T))) in MATLAB. The thing is I got a case that works for me:

static const int nx = 4; 
static const int ny = 4;
static const int nz = 4;

Eigen::Tensor<double, 3> MaxTest(nx,ny,nz); 
MaxTest.setZero();  

Eigen::Tensor<double, 0> AbsMaxAsTensor = MaxTest.abs().maximum();
double Max = AbsMaxAsTensor(0);
std::cout << "Maximum = " << Max << std::endl;

However, now when I try to apply this in a function as the following,

double vMaxArr[10];
		
Eigen::Tensor<double, 0> AbsMaxAsTensor = Test.maximum();
vMaxArr[0] = AbsMaxAsTensor(0);

where Test is passed to the function as Eigen::Tensor<std::complex<double>, 3>& Test
Then I get the error:

 error: cannot convert ‘Eigen::TensorEvaluator<Eigen::Tensor<double, 0>, Eigen::DefaultDevice>::EvaluatorPointerType’ {aka ‘double*’} to ‘Eigen::TensorReductionEvaluatorBase<const Eigen::TensorReductionOp<Eigen::internal::MaxReducer<std::complex<double>, 0>, const Eigen::DimensionList<long int, 3>, const Eigen::Tensor<std::complex<double>, 3>, Eigen::MakePointer>, Eigen::DefaultDevice>::EvaluatorPointerType’ {aka ‘std::complex<double>*’}
  152 |     return m_rightImpl.evalSubExprsIfNeeded(m_leftImpl.data());
      |                                                              ^
      |                                                              |
      |                                                              Eigen::TensorEvaluator<Eigen::Tensor<double, 0>, Eigen::DefaultDevice>::EvaluatorPointerType {aka double*}

I am not sure what the issue is?

[mbozzi]

Looks like you forgot to call abs in line 3

Maybe it should read

Eigen::Tensor<double, 0> AbsMaxAsTensor = Test.abs().maximum();
vMaxArr[0] = AbsMaxAsTensor(0);

[JamieAl]

@mbozzi
But I am trying to just get the maximum value without the absolute value if possible.

[lastchance]

where Test is passed to the function as Eigen::Tensor<std::complex<double>, 3>& Test

You can’t take the maximum of a complex number - it would make no sense.

It does make sense to take the maximum of a real number, which is what you would get by applying abs() first.

[JamieAl]

@lastchance
Thanks! I am confusing myself A LOT because I am using MATLAB to contrast and compare.