ternlang-core 0.3.3

Compiler and VM for Ternlang — balanced ternary language with affirm/tend/reject trit semantics, @sparseskip codegen, and BET bytecode execution.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31

// Module:  stdlib/research/trit_memory.tern
// Purpose: Ternary Memory-Augmented Network
// Author:  RFI-IRFOS
// Ref:     https://ternlang.com

// Neural Turing Machine / DNC analog.

fn read_trit(memory: trittensor<4 x 4>, address: trittensor<4 x 1>) -> trittensor<4 x 1> {
    @sparseskip
    let out: trittensor<4 x 1> = memory * address;
    return out;
}

fn write_trit(memory: trittensor<4 x 4>, address: trittensor<4 x 1>, data: trittensor<4 x 1>) -> trittensor<4 x 4> {
    return memory;
}

fn address_trit(query: trittensor<4 x 1>, keys: trittensor<4 x 4>) -> trittensor<4 x 1> {
    @sparseskip
    let out: trittensor<4 x 1> = keys * query;
    return out;
}

fn tend_as_no_op(action: trit) -> trit {
    if action == tend { return tend; } // No read/write performed
    match action {
        affirm => { return affirm; }
        tend   => { return tend;   }
        reject => { return reject; }
    }
}