topic Re: Dot product error? in Bifrost Forum

Dot product error?

mcw0 — Wed, 15 Feb 2023 18:47:24 GMT

I'm comparing the 3 vectors. I want to know how closely 2 vectors match the first. But my dot values are 0! Please, what did I do wrong? This has to work today. 😓

Re: Dot product error?

mcw0 — Wed, 15 Feb 2023 18:56:50 GMT

As an experiment, I created new vectors and the dot product worked. But as soon as I plug in my earlier normalized vectors to the new dot node, I get a zero. All input types are the same, Math::float3. What's going on?

Re: Dot product error?

morten.bojsen-hansen — Wed, 15 Feb 2023 20:03:43 GMT

Please check what the watchpoint (to print_array1) is saying.

Re: Dot product error?

mcw0 — Wed, 15 Feb 2023 20:21:46 GMT

So the print_array is returning incorrect values? But only on some and not all? How do I fix this? I need to see the values to debug my graph. If I have to resort to the watchpoint...OUCH!

Re: Dot product error?

Grahame_Fuller — Wed, 15 Feb 2023 20:26:20 GMT

It looks like the final result (0.77) is correct but for some reason the first print_array node is showing an incorrect value for index 4. The result of normalizing (1, 0, 10) is (0.0995037, 0, 0.995037) not (.1, 0, 0) but if that incorrect value had actually been used for the dot product then the result would have been 0.071 not 0.77.

gray

Re: Dot product error?

mcw0 — Wed, 15 Feb 2023 20:31:55 GMT

Good catch! I missed that. So bottom line, I can't rely on the print_array results.

Betreff: Dot product error?

Roland.Reyer — Thu, 16 Feb 2023 07:37:10 GMT

Please get the latest version of the PRINT_PACK.

There was a bug in the display of min/max.

Also, print_table is the better choice if you need to compare arrays.

Re: Dot product error?

Roland.Reyer — Thu, 16 Feb 2023 16:59:38 GMT

When the floats are converted to string I truncate the decimal places to a number that you can set in the print node.

When you set "Float Precision" to 10.000 you would see 5 decimal places.