forked from inakleinbottle/exponentiating_formulae
-
Notifications
You must be signed in to change notification settings - Fork 0
/
forward_transformer.cpp
127 lines (103 loc) · 4.57 KB
/
forward_transformer.cpp
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
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
#include "SHOW.h"
#include "environment.h"
// the receiving environment
constexpr DEG WIDTHOUT = 3;
constexpr DEG DEPTHOUT = 2;
// the depth of the shuffles producing the channels
constexpr DEG INOUTDEPTH = 2;
// the incoming stream with enough accuracy to determine the
// required integrals of the projections
constexpr DEG WIDTHIN = 2;
constexpr DEG DEPTHIN = (DEPTHOUT * INOUTDEPTH);
// the input and output environments
using IN = Environment<WIDTHIN, DEPTHIN>;
using OUT = Environment<WIDTHOUT, DEPTHOUT>;
// steady state requires inputs limited to the same depth as the output
using SHORT_LIE = IN::LIE_<DEPTHOUT>;
// limiting the degree of nonlinearity in the functions on paths
using SHORT_SHUFFLE = IN::SHUFFLE_TENSOR_<INOUTDEPTH>;
int forward_transformer()
{
auto k = IN::K;
IN in;
const auto& sbasis = in.sbasis;
std::cout << "Creating the two generic input log signatures \"before\" and \"during\" truncated to level " << DEPTHIN << "\n\n";
IN::LIE logsig_before;
add_equals_short(logsig_before, in.generic_vector<SHORT_LIE>(1000));
SHOW(logsig_before);
IN::TENSOR tensor_logsig_before = in.maps_.l2t(logsig_before);
IN::TENSOR sig_before = exp(tensor_logsig_before);
SHOW(sig_before);
IN::LIE logsig_during;
add_equals_short(logsig_during, in.generic_vector<SHORT_LIE>(2000));
SHOW(logsig_during);
IN::TENSOR tensor_logsig_during = in.maps_.l2t(logsig_during);
IN::TENSOR sig_during = exp(tensor_logsig_during);
SHOW(sig_during);
//IN::LIE logsig_after;
//add_equals_short(logsig_after, in.generic_vector<SHORT_LIE>(3000));
//SHOW(logsig_after);
//IN::TENSOR tensor_logsig_after = in.maps_.l2t(logsig_after);
//IN::TENSOR sig_after = exp(tensor_logsig_after);
//SHOW(sig_after);
std::cout << "Concatenating \"before\" and \"during\" to form \"sig\" and then \"logsig\" truncated to level " << INOUTDEPTH << "\n\n";
IN::TENSOR sig = sig_before * sig_during;// *sig_after;
SHOW(sig);
SHOW(antipode(sig) * sig);
IN::TENSOR tensor_logsig = log(sig);
IN::LIE logsig = in.maps_.t2l(tensor_logsig);
SHOW(logsig);
OUT out;
IN::SHUFFLE_TENSOR generic_basic_shuffles[OUT::WIDTH];
const DEG in_shuffle_tensor_width = SHORT_SHUFFLE::BASIS::start_of_degree(INOUTDEPTH + 1) - SHORT_SHUFFLE::BASIS::start_of_degree(0);
// now populate a vector of shuffles that gives the OUT path
std::cout << "Creating the weights: " << OUT::WIDTH << " generic truncated input shuffles truncated to level " << INOUTDEPTH << "\n\n";
{
int count = 0;
for (auto& sh : generic_basic_shuffles) {
SHOW(count);
add_equals_short(sh, in.generic_vector<SHORT_SHUFFLE>(count));
count += in_shuffle_tensor_width;
SHOW(sh);
}
std::cout << "Created " << count << " generic shuffle weight polynomials \n";
}
// now populate the tensor over the vector of shuffle coordinates
// to get the signature of the OUT path (a grouplike element).
std::map<OUT::TENSOR::BASIS::KEY, IN::SHUFFLE_TENSOR> result;
const OUT::TENSOR::BASIS obasis;
auto& obegin = obasis.begin();
auto& oend = obasis.end();
const IN::SHUFFLE_TENSOR::BASIS ibasis;
auto& ibegin = ibasis.begin();
auto& iend = ibasis.end();
auto tkey = obegin;
// the first entry in the tensor is the polynomial that is the constant 1
if (tkey != oend) {
result[tkey] = IN::SHUFFLE_TENSOR(IN::poly_t(1));
tkey = obasis.nextkey(tkey);
}
// the increment of the path (so constant terms get set to zero!!!)
if (tkey != oend) {
for (auto basic_shuffle : generic_basic_shuffles) {
basic_shuffle[ibegin] -= basic_shuffle[ibegin];
result[tkey] = basic_shuffle;
tkey = obasis.nextkey(tkey);
}
}
for (; tkey != oend; tkey = obasis.nextkey(tkey)) {
auto letter = tkey.lparent();// first letter of key as a key
auto rest = tkey.rparent(); // remainder of key as a key
result[tkey] = half_shuffle_multiply(result[letter], result[rest]);
}
OUT::TENSOR ans = apply1<OUT, IN>(result, sig);
SHOW(antipode(ans) * ans);
OUT::TENSOR ans_before_during = apply1<OUT, IN>(result, sig_before * sig_during);
SHOW(antipode(ans_before_during) * ans_before_during);
OUT::TENSOR ans_before = apply1<OUT, IN>(result, sig_before);
SHOW(antipode(ans_before) * ans_before);
OUT::TENSOR ans_during = antipode(ans_before) * ans_before_during;
SHOW(antipode(ans_during) * ans_during);
std::cout <<out.maps_.t2l(log(ans_during)) << "\n\n";
return 0;
}