MARIUS BULIGA |
---|

(home) (research subjects) | ||

Chemlambda v2 | ||

Github repository | ||

Demos | ||

Chemlambda for the people - slides | ||

(arXiv) (DOI)- Molecular computers, text version (2018) of the original, open issue for comments at Github | ||

(DOI)- The Chemlambda collection of simulations (2017) | ||

(GitHub)- The library of chemlambda molecules (2016), instructions to use here | ||

(GitHub)- Turing machines, chemlambda style (2015) | ||

(html)- Molecular computers (2015) | ||

(DOI) (journal)- Build a molecular computer. Journal of Brief Ideas (2015) | ||

Nodes | ||

See this visual tutorial to understand how the nodes and the edges orientations are represented in the simulations, for example this one involving kinesin like graphs. | ||

Beta family of graph rewrites | ||

DIST family of graph rewrites | ||

Pruning family of graph rewrites | ||

alternative to FO-T, FOE-T are FO2-T, FOE2-T, FO3-T, FOE3-T | ||

Comb family of graph rewrites | ||

Conflicts between rewrites are possible. One of them is between the Beta rewrite and A-FO or A-FOE. This conflict, by itself, does not affect the reduction of molecules (i.e. chemlambda graphs) which represent untyped lambda terms, because of the particular shape of those graphs. See arXiv:1305.5786, section 3, for the algorithm of representing a lambda term as a chemlambda molecule. (Just use the chemlambda v2 node notations for the nodes A, L, FI, FO, T instead of those from GLC from the article, the algorithm is the same.) | ||

Another conflict is between the FO-FOE rewrite and any *-FO rewrite from the DIST family. This conflict has two effects: it produces a supplimentary fanout node which may trigger more DIST rewrites, and it produces a supplimentary FI node which may not meet later a FOE node. | ||

Here are the two possible evolutions of a particular molecule where this happens. The conflict is between a L-FO rewrite and a FO-FOE rewrite. If we pick the L-FO rewrite first then we end with 3 copies of a small molecule (which represents the identity in lambda calculus). | ||

If we pick the FO-FOE rewrite first then we end with 4 copies of that small molecule, two of them connected to a FI node. Ther are more possible paths to the end result than those shown, depending on the order of application of FI-FOE and L-FO rewrites, but all end into the same place. | ||

There are several techniques to compensate for this phenomenon, from careful redesigning the molecule (an example. Or another example). The last one can be further modified to this simulation of a molecule inspired from the factorial of 5. | ||

My favorite technique, which has the right chemical blend, is: amplify, let the reaction end, sample for result. We can always anihilate the unwanted FI nodes by replacing FROUT nodes with a sufficiently long chain of FOE nodes, and FRIN nodes, if they are present, with the same number of FI nodes. | ||

content licensed (CC BY 4.0) unless otherwise specified | ||