Is the phase noise at the outputs of the LTC6951 correlated?? Meaning if there was no delay between 2 outputs, would the phase noise be correlated. Would there still be correlation but shifted in time if one of the outputs was delayed with respect to the other??

The PLL in band, VCO curve phase noise and VCO wide band are shared noises among all outputs. These will be correlated noises for outputs whose clock edges that occur on the same LTC6951 Pdivider cycle. Each output has a different output driver, so this portion of the wide band noise will not be correlate (just the wide band noise from the shared VCO).

Here are some examples

If you have two outputs running at frequencies slower than the LTC6951 Pdivider and their delays are programmed such that their edges occur at 90 degrees (for example) offset from each other, then these will 'not' be correlated.

If you have two outputs running at frequencies slower than the LTC6951 Pdivider and their delays are programmed such that their edges occur at the same instance (integer multiples of 360 degrees of output frequency) then these will be correlated.

If you have two outputs running at frequencies equal to the LTC6951 Pdivider, then these will be correlated.