TM1 Forum

Hi All,

If we see the value in rules tracer, last six digits it is 565012, where as the value in cube viewer under element 2014 last six digits is 57100. please see the below screenshot If I replace my periods dimension and replace it in columns area in cube viewer, the value changes again. Please see the below screenshot. Can some please share info, why it is showing different values when the dimension position is changed?

Regards,
Deepak Jain

I have to say that the values look odd for a measure called Margin%.
Aside from that, you're not giving much information here. You've cleared so much in the screenshots that they're not giving much more information than no screenshots.
Are there any rules applied against the 2014 Plan? If so, are the rules calculating on a consolidated level? Any other info that you can share?

Michel

I suspect you are seeing slight differences due to the representation of values in TM1 as IEEE-754 doubles.

If you enter your values into the calculator here:

http://babbage.cs.qc.cuny.edu/IEEE-754.old/Decimal.html

And look at the double precision area you will see that the mantissa (bits 51-0) only differs at the last couple of bits:

1 .0100000001110010111101001110000110011110100010010101
1 .0100000001110010111101001110000110011110100010010010

Floating point is an interesting beast. You can get a different number by adding up the same numbers in a different order. So I think all you are seeing here is floating point error down in the last bits of the mantissa.

You might think it is a "large" error but relative to the size of your numbers it is not.
If you calculate the percentage difference for your last example you will find that it is around 0.00000000000005%.

Hi Michel,

It is calculated based on
It might not be business interest to check the value at that level.
It is my curiosity to know why this thing is happening.

If any more information is required, please let me know.

Regards,
Deepak Jain

Thanks Nyble. I will look out.

Was just looking to understand, when I am replacing the Title Dimension with Column dimension the value is changing.
It should not happen, if it is the scenario as suggested.

Regards,
Deepak Jain

deepakjain2020 wrote: It is my curiosity to know why this thing is happening.

I would go for Dr. Nybble's explanation. TM1 is a multidimensional system,in which there are multiple paths to calculate a consolidated value. Given the floating point precision rolling up one way to a total could lead to marginally different results than when rolling up using a different path. TM1 is 'calculating on demand'. The optimal path to come the resulting value is more or less determined dynamically.

Michel

Thanks Dr. Nybble & Michel for sharing your knowledge