Implicit conversion from system array to ilnumerics array doesnt retain double precision

Question

When I assign a system array of doubles to an ilnumerics double array, the values are rounded off to nearest integer. This happens particularly for only large arrays.

Is there any way in ILnumerics to specify up to how many decimals the rounding should occur?

The following screenshot shows the problem . Sample_pulsedata is double array of length 1860 which I am assigning to sample_ydata.

Answer 1

The elements are not really rounded. The effect rather comes from the way the elements are displayed in Visual Studios data tips. ILNumerics tries to find a common scale factor which allows to display all elements in an array aligned.

In your example - presumably - there exist large values at higher indices, which are not shown currently (scroll down in order to find them). These elements cause the scale factor to be 1/10^4. This is indicated in the first line, index [0]: '(:;:) 1e+004'. The 32.57 therefore must get rounded to 33 in order to fit into the 4 digits after the decimal point. '4' is a fixed value in ILNumerics and cannot easily get changed.

The actual values of the array elements are not affected, of course. You can use the Watch window to show only the interesting part of the array, without the rounding effect:

sample_ydata["0:13"]

Or, even better, use the ILNumerics Array Visualizer in order to visualize your data graphically. This not only gives a nice overview of the whole array but also prevents from such artefacts as you encountered.