Mian-Chowla sequence, A051788

[Jack Brennen]

all at abouthugo.de wrote:
> 
> For the Mian-Chowla sequence A005282 a conjectured list of differences
> not covered is:
> %S A080200  33 88 98 99 105 106 112 126 130 132 134 150 152 154 156 162
> 163 165 170 176 184 188 198 205 214 215 216 217 220 222 228
> 234 235 240 246
> It was created using a little Fortran program available at:
> http://www.randomwalk.de/sequences/a080200.txt
> which checked the elements of A005282 up to 4*10^6.
> 

I checked it out to the 2429th element of the sequence, which is
167041391.  Nothing was found to contradict any of the terms
conjectured above.  The last difference I found which is less than 1000
is the difference between elements number 1346 and 1347, which are
31979676 and 31980141, with difference 465.

> Any ideas how to prove that differences not occurring early in the
> sequences
> A005282 or A051788 (which is just a variation of A005282 starting with
> 1,3,..
> instead of 1,2,..)  will not occur further up?
> 

None at all.  :(