How can I save/load a .NET array to a .mat file?

8 views (last 30 days)
George on 9 Apr 2013
Answered: TB on 19 Nov 2013
How can I save/load a .NET array to a .MAT file? I always get errors when trying to load the data after it has been saved e.g.
>> dArray = NET.createArray('System.Double', 2);
>> dArray(1) = 1.1;
>> dArray(2) = 2.2;
>> save dArray.mat dArray
>> load('dArray.mat')
Warning: Cannot load an object of class 'Double[]':
No matching constructor signature found.

Answers (2)

Eric on 9 Apr 2013
It's perhaps kind of a pain, but you could cast the array to a Matlab double array prior to saving and then cast it back to a .NET array after loading. You could do something like
mat_Array = double(dArray);
save dArray.mat mat_Arry
dArray = NET.convertArray(mat_Array);
Alternatively, you could try to install System.Double into the Global Assembly Cache. See for information on how to do that. That might then allow Matlab to find the constructor for the System.Double[] object, but I'm not sure.
Good luck.
George on 10 Apr 2013
Thanks Eric. I am already using NET.addAssembly() and it works perfectly saving/loading a SINGLE object, but it doesn't work for an ARRAY of the same object type.
It looks like MATLAB is confused by the '[]' notation for the arrays and is trying to locate a class called e.g. 'System.Double[]' instead of an array of 'System.Double' objects.

Sign in to comment.

TB on 19 Nov 2013
Can you put your .Net array inside a scalar wrapper object? Something like
Public Class ArrayWrapper
Public CustomArray() As CustomClass
End Class
and then try to save an instance of the ArrayWrapper class? Not sure if it will work for you, but I have used a similar work around for a different problem in the past.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!