[dmm]

How about a model like this:
Suppose one supercomputer center (SC) gave 100% boost to one research center (RC). Then, to account for overloading of the SC, SEIV could have the boost decrease by 10% for every additional RC supported, up to a max support of 10 (by which point the support of the overloaded SC has decreased so much that it is as if only one RC were being supported).

1 SC with 1 RC = 2.0x1 = 2.0 research

2 SC with 2 RC = 2.0x2 = 4.0 research
1 SC with 2 RC = 1.9x2 = 3.8 research

3 SC with 3 RC = 2.0x3 = 6.0 research
1 SC with 3 RC = 1.8x3 = 5.4 research
...
5 SC with 5 RC = 2.0x5 = 10.0 research
1 SC with 5 RC = 1.6x5 = 8.0 research
...
10 SC with 10 RC = 2.0x10 = 20.0 research
1 SC with 10 RC = 1.1x10 = 11.0 research

Of course, this model could be modified, if MM wanted to make SCs more/less powerful. For example, perhaps a level 1 SC could only support 5 RCs (going down by 20% each time), level 2 could support 10 as above, level 3 could support 20, etc. Or, for example, the basic boost could be 150% rather than 100%, so that building a SC would be advantageous even if it were only supporting one RC.

The programming for this would be very easy if all RCs were the same. In general, though, some RCs might be level 1, some level 2, some level 3, etc., and obviously, it would be best to assign one's SCs to high-level RCs, which would complicate the programming some.