decompress

function to run diverge algorithms on the input. You can choose the diverge algorithm with m.decompress (trivial, paramlike,graphlike,classic,classiclg,ccll), c can be a list (multiple divergences) and also handles the bottleneck actions (define a new input, and return it later). Always returns: g,compressed version,new input

Arguments:

• g: a grap object
• m: a constant variable. created for example by getm()
• c: the abstraction factor

Returns

• g: the decompressed grap value
• com: latent space vector
• inn2: input for the decompressor