viterbi_trellis crf::viterbi_scorer::viterbi(const sequence& seq)
{
// we only need the scores for the states as the transition scores, set
// up during construction, will never change between sequences
scorer_.state_scores(*model_, seq);
viterbi_trellis table{seq.size(), model_->num_labels()};
// initialize first column of trellis. We use the original state() and
// trans() matrices because we are working in the log domain.
for (label_id lbl{0}; lbl < model_->num_labels(); ++lbl)
table.probability(0, lbl, scorer_.state(0, lbl));
// compute remaining columns of trellis using recursive formulation
for (uint64_t t = 1; t < seq.size(); ++t)
{
for (label_id lbl{0}; lbl < model_->num_labels(); ++lbl)
{
double max_score = std::numeric_limits<double>::lowest();
for (label_id in{0}; in < model_->num_labels(); ++in)
{
auto score = table.probability(t - 1, in)
+ scorer_.trans(in, lbl);
if (score > max_score)
{
max_score = score;
table.previous_tag(t, lbl, in);
}
}
table.probability(t, lbl, max_score + scorer_.state(t, lbl));
}
}
return table;
}
// is end of a chunk (IOB1)?
int evaluation::is_end_of_chunk_iob1(int human_model, int i, sequence & seq,
string b_tag, string i_tag) {
if (human_model == 1) {
if (seq[i].label == atoi(b_tag.c_str())) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1].label != atoi(i_tag.c_str())) {
return 1;
} else {
return 0;
}
}
} else if (seq[i].label == atoi(i_tag.c_str())) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1].label != atoi(i_tag.c_str())) {
return 1;
} else {
return 0;
}
}
} else {
return 0;
}
} else if (human_model == 2) {
if (seq[i].model_label == atoi(b_tag.c_str())) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1].model_label != atoi(i_tag.c_str())) {
return 1;
} else {
return 0;
}
}
} else if (seq[i].model_label == atoi(i_tag.c_str())) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1].model_label != atoi(i_tag.c_str())) {
return 1;
} else {
return 0;
}
}
} else {
return 0;
}
} else {
return 0;
}
}
// is end of a chunk (IOB1)?
int is_end_of_chunk_iob1(int human_model, int i, sequence & seq,
string b_tag, string i_tag) {
if (human_model == 1) {
if (seq[i][seq[i].size() - 2] == b_tag) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1][seq[i + 1].size() - 2] != i_tag) {
return 1;
} else {
return 0;
}
}
} else if (seq[i][seq[i].size() - 2] == i_tag) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1][seq[i + 1].size() - 2] != i_tag) {
return 1;
} else {
return 0;
}
}
} else {
return 0;
}
} else if (human_model == 2) {
if (seq[i][seq[i].size() - 1] == b_tag) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1][seq[i + 1].size() - 1] != i_tag) {
return 1;
} else {
return 0;
}
}
} else if (seq[i][seq[i].size() - 1] == i_tag) {
if (i >= seq.size() - 1) {
return 1;
} else {
if (seq[i + 1][seq[i + 1].size() - 1] != i_tag) {
return 1;
} else {
return 0;
}
}
} else {
return 0;
}
} else {
return 0;
}
}
sequence<R> operator*( const sequence<T1> &X, const sequence<T2> &Y )
{
if( X.size()==0 || Y.size()==0 )
return sequence<R>();
vec<R> vec = conv( X.buffer(), Y.buffer() );
int t1 = X.t1() + Y.t1();
return sequence<R>( vec, t1 );
}
void generate_vietoris_sequence(sequence& a) {
if (a.empty())
return;
a[0] = rdm(0, MAX_VAL);
if (a.size() < 2)
return;
a[1] = rdm(0, a[0]);
for (int k = 2; k < a.size(); k++) {
a[k] = rdm(0, (1.0 * k - 1) / k) * a[k - 1];
}
}
void generate_vietoris_sequence(sequence& a) {
if (a.empty())
return;
a[0] = 1;
if (a.size() < 2)
return;
a[1] = .5;
for (int k = 2; k < a.size(); k++) {
a[k] = a[k - 1] * (k - 1) / k;
}
}
/*
* Helper for element and docuemnt constructors to insert sequence
* of atomic values. Returns true if node was actually inserted.
* In this case left pointer is changed to the last inserted indirection.
* In any case at_vals sequence is cleared.
*/
static inline bool
process_atomic_values(xptr& left, const xptr& parent, sequence& at_vals) {
if (at_vals.size() > 0)
{
executor_globals::tmp_op_str_buf.clear();
tuple_cell tcc;
sequence::iterator it = at_vals.begin();
do {
tcc = tuple_cell::make_sure_light_atomic((*it).cells[0]);
tcc = cast(tcc, xs_string);
executor_globals::tmp_op_str_buf.append(tcc);
it++;
}
while (it != at_vals.end());
at_vals.clear();
if(executor_globals::tmp_op_str_buf.get_size() > 0) {
insert_text(indirectionDereferenceCP(left),
XNULL,
indirectionDereferenceCP(parent),
text_source_strbuf(&(executor_globals::tmp_op_str_buf)));
left = get_last_mo_inderection();
return true;
}
}
return false;
}
void crf::tagger::tag(sequence& seq)
{
auto trellis = scorer_.viterbi(seq);
auto lbls = util::range(label_id{0},
label_id(static_cast<uint32_t>(num_labels_ - 1)));
auto last_lbl = functional::argmax(
lbls.begin(), lbls.end(), [&](label_id lbl)
{
return trellis.probability(seq.size() - 1, lbl);
});
seq[seq.size() - 1].label(*last_lbl);
for (uint64_t t = seq.size() - 1; t > 0; t--)
seq[t - 1].label(trellis.previous_tag(t, seq[t].label()));
}
orthonomial(const sequence &a, const sequence &b, const sequence &c) : n(c.size() - 1), a(a), b(b), c(c) {
/* Ze względów bezpieczeństwa. */
this->a.push_back(0);
this->b.push_back(0);
this->c.push_back(0);
this->c.push_back(0);
}
// is matching chunk (IOE2)?
int evaluation::is_matching_chunk_ioe2(int i, sequence & seq, string i_tag, string e_tag) {
if (!is_start_of_chunk_ioe2(1, i, seq, i_tag, e_tag) ||
!is_start_of_chunk_ioe2(2, i, seq, i_tag, e_tag)) {
return 0;
}
int len = seq.size();
int j = i, k = i;
while (j < len) {
if (is_end_of_chunk_ioe2(1, j, seq, i_tag, e_tag)) {
break;
} else {
j++;
}
}
while (k < len) {
if (is_end_of_chunk_ioe2(2, k, seq, i_tag, e_tag)) {
break;
} else {
k++;
}
}
return (j == k);
}
void perceptron::tag(sequence& seq) const
{
for (uint64_t t = 0; t < seq.size(); ++t)
{
analyzer_.analyze(seq, t);
seq[t].label(model_.best_class(seq[t].features()));
seq[t].tag(analyzer_.tag(seq[t].label()));
}
}
RObject
AORB::object_key_to_object(sequence<octet>& object_key)
{
RString str = new String((const char*)object_key.data(), object_key.size(), NormalSST | CCAscii);
ObjectKey objkey(str);
if (objkey.isLocal() == true)
return objkey.getLocalObject();
// return Skelleton here
return Nil;
}
complex inner_prod(const sequence<T> &X, const sequence<T> &Y )
{
// If any vector is empty
if( X.size() == 0 || Y.size() == 0 )
return 0;
// Overlapping interval
int ta = max(X.t1(),Y.t1());
int tb = min(X.t2(),Y.t2());
// If they do not overlap
if( ta > tb )
return 0;
// They do overlap
complex r = 0;
for( int t = ta; t <= tb; t++ )
r += inner_prod( X(t), Y(t) );
return r;
}
// compares the given two instances of ss
double cm_assembly_ssq3::compare(ss const &__first, ss const &__second) const {
sequence<cchb_dssp> const seq1(__first.get_sequence()), seq2(__second.get_sequence());
if(seq1.size() != seq2.size()) {
throw math::compare_error(get_identifier() + ": Sequence length differ, sequence1.length=" +
std::to_string(seq1.size()) + ", sequence2.length=" + std::to_string(seq2.size()));
} // if
size_t c_correct(0), h_correct(0), e_correct(0); // initialize
for(size_t pos(0); pos < seq1.size(); ++pos) { // works for both sequences b/c of same length
char const sequence1_ss(seq1[pos].get_identifier_char());
char const sequence2_ss(seq2[pos].get_identifier_char());
if(sequence1_ss == sequence2_ss) { // actually it's only important that they are the same, not which one
if(sequence1_ss == 'C') {
++c_correct;
} // if
else if(sequence1_ss == 'H') {
++h_correct;
} // else if
else if(sequence1_ss == 'E') {
++e_correct;
} // else if
} // if
} // for
DEBUG << get_identifier() << ": c_correct=" << c_correct << " h_correct=" << h_correct
<< " e_correct=" << e_correct << " seq_len=" << seq1.size();
return ((double)c_correct + h_correct + e_correct) / seq1.size();
} // compare()
// counting matching chunks (IOE2)
int count_matching_chunks_ioe2(sequence & seq, string i_tag, string e_tag) {
int count = 0;
for (int i = 0; i < seq.size(); i++) {
if (is_start_of_chunk_ioe2(1, i, seq, i_tag, e_tag)) {
if (is_matching_chunk_ioe2(i, seq, i_tag, e_tag)) {
count++;
}
}
}
return count;
}
sequence<decltype(T()*S())> element_prod( const sequence<T>& X, const sequence<S>& Y )
{
typedef decltype(T()*S()) R;
// If any vector is empty
if( X.size() == 0 || Y.size() == 0 )
return sequence<R>();
// Overlapping interval
int ta = max(X.t1(),Y.t1());
int tb = min(X.t2(),Y.t2());
// If they do not overlap
if( ta > tb )
return sequence<R>();
// They do overlap
vec<R> v = element_prod(
X.buffer()( range( ta-X.t1(), tb-X.t1()+1 ) ),
Y.buffer()( range( ta-Y.t1(), tb-Y.t1()+1 ) ) );
return sequence<R>( v, ta );
}
// counting number of chunks (IOE2)
int count_chunks_ioe2(int human_model, sequence & seq, string i_tag, string e_tag) {
int count = 0;
for (int i = 0; i < seq.size(); i++) {
if (human_model == 1 && is_start_of_chunk_ioe2(1, i, seq, i_tag, e_tag)) {
count++;
}
if (human_model == 2 && is_start_of_chunk_ioe2(2, i, seq, i_tag, e_tag)) {
count++;
}
}
return count;
}
void convert_IOB2_IOB1(int is_cap, int is_last, sequence & seq) {
map<int, int> fixedlabels;
map<int, int>::iterator obsrit;
char begin = is_cap ? 'B' : 'b';
char inside = is_cap ? 'I' : 'i';
string INSIDE = is_cap ? "I" : "i";
string str1, str2, istr, newlabel;
int i, len = seq.size();
for (i = 1; i < len; i++) {
int col1 = is_last ? seq[i - 1].size() - 1 : seq[i - 1].size() - 2;
int col2 = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
str1 = seq[i - 1][col1];
str2 = seq[i][col2];
if (str2[0] != begin) {
continue;
}
istr = INSIDE;
istr += strtail(str2);
if (str1 == str2 || str1 == istr) {
fixedlabels.insert(pair<int, int>(i, i));
}
}
for (i = 0; i < len; i++) {
obsrit = fixedlabels.find(i);
if (obsrit != fixedlabels.end()) {
continue;
}
int col = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
if (seq[i][col][0] == begin) {
// B- or b- => I- or i-
newlabel = INSIDE;
newlabel += strtail(seq[i][col]);
seq[i][col] = newlabel;
}
}
}
void convert_IOE1_IOE2(int is_cap, int is_last, sequence & seq) {
vector<int> endlabels;
char end = is_cap ? 'E' : 'e';
string END = is_cap ? "E" : "e";
char inside = is_cap ? 'I' : 'i';
string str1, str2, estr, newlabel;
int i, len = seq.size();
for (i = 0; i < len; i++) {
int col1 = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
str1 = seq[i][col1];
if (str1[0] != inside) {
continue;
}
if (i == len - 1) {
endlabels.push_back(i);
} else {
int col2 = is_last ? seq[i + 1].size() - 1 : seq[i + 1].size() - 2;
str2 = seq[i + 1][col2];
estr = END;
estr += strtail(str1);
if (str2 != str1 && str2 != estr) {
endlabels.push_back(i);
}
}
}
for (i = 0; i < endlabels.size(); i++) {
int col = is_last ? seq[endlabels[i]].size() - 1 : seq[endlabels[i]].size() - 2;
newlabel = END;
newlabel += strtail(seq[endlabels[i]][col]);
seq[endlabels[i]][col] = newlabel;
}
}
// **************************************************************************
// bool test_basic(const sequence& test, size_t s, bool has_cursor)
// Postcondition: A return value of true indicates:
// a. test.size() is s, and
// b. test.is_item() is has_cursor.
// Otherwise the return value is false.
// In either case, a description of the test result is printed to cout.
// **************************************************************************
bool test_basic(const sequence& test, size_t s, bool has_cursor)
{
bool answer;
cout << "Testing that size() returns " << s << " ... ";
cout.flush( );
answer = (test.size( ) == s);
cout << (answer ? "Passed." : "Failed.") << endl;
if (answer)
{
cout << "Testing that is_item() returns ";
cout << (has_cursor ? "true" : "false") << " ... ";
cout.flush( );
answer = (test.is_item( ) == has_cursor);
cout << (answer ? "Passed." : "Failed.") << endl;
}
return answer;
}
void convert_IOB1_IOB2(int is_cap, int is_last, sequence & seq) {
vector<int> firstlabels;
char begin = is_cap ? 'B' : 'b';
string BEGIN = is_cap ? "B" : "b";
char inside = is_cap ? 'I' : 'i';
string str1, str2, bstr, newlabel;
int i, len = seq.size();
for (i = 0; i < len; i++) {
int col2 = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
str2 = seq[i][col2];
if (str2[0] != inside) {
continue;
}
if (i == 0) {
firstlabels.push_back(i);
} else {
int col1 = is_last ? seq[i - 1].size() - 1 : seq[i - 1].size() - 2;
str1 = seq[i - 1][col1];
bstr = BEGIN;
bstr += strtail(str2);
if (str2 != str1 && str1 != bstr) {
firstlabels.push_back(i);
}
}
}
for (i = 0; i < firstlabels.size(); i++) {
int col = is_last ? seq[firstlabels[i]].size() - 1 : seq[firstlabels[i]].size() - 2;
newlabel = BEGIN;
newlabel += strtail(seq[firstlabels[i]][col]);
seq[firstlabels[i]][col] = newlabel;
}
}
void Print_Cpg(sequence& hidden_sequence, uint shift, string& chromosome_name)
{
uint start_cpg = 0;
bool is_open_cpg = false;
ull i;
for (i = 0; i < hidden_sequence.size(); ++i)
{
if (hidden_sequence[i] < 4)
{
if (!is_open_cpg)
{
start_cpg = i + shift + 1;
is_open_cpg = true;
}
}
else
{
if (is_open_cpg)
{
uint end_cpg = i + shift + 1;
if (Check_Sequence(hidden_sequence, start_cpg - shift - 1, end_cpg - shift - 1))
{
cout << chromosome_name << "\t" << start_cpg << "\t" << end_cpg << endl;
}
is_open_cpg = false;
}
}
}
if (is_open_cpg)
{
uint end_cpg = i + shift + 1;
if (Check_Sequence(hidden_sequence, start_cpg, end_cpg))
{
cout << chromosome_name << "\t" << start_cpg << "\t" << end_cpg << endl;
}
}
}
sequence<R> operator+( sequence<T1> X, sequence<T2> Y )
{
if( X.size() == 0 )
return Y;
if( Y.size() == 0 )
return X;
// Intervals
int t1 = min(X.t1(),Y.t1());
int ta = max(X.t1(),Y.t1());
int tb = min(X.t2(),Y.t2());
int t2 = max(X.t2(),Y.t2());
int sx = X.size();
int sy = Y.size();
// First interval
vec<R> v1(0);
if( t1 == X.t1() && t1 != Y.t1() )
v1 = X.buffer()( range( 0, min(ta-X.t1(),sx) ) );
else if( t1 != X.t1() && t1 == Y.t1() )
v1 = Y.buffer()( range( 0, min(ta-Y.t1(),sy) ) );
// Second interval
vec<R> v2;
if( ta <= tb )
v2 = X.buffer()( range( ta-X.t1(), tb-X.t1()+1 ) ) + Y.buffer()( range( ta-Y.t1(), tb-Y.t1()+1 ) );
else {
int I = ta-tb-1;
v2.resize(I);
for( int i = 0; i < I; i++ )
v2(i) = 0*X.buffer()(0);
}
// Third interval
vec<R> v3(0);
if( t2 == X.t2() && t2 != Y.t2() )
v3 = X.buffer()( range( max(tb-X.t1()+1,0), X.size() ) );
else if( t2 != X.t2() && t2 == Y.t2() )
v3 = Y.buffer()( range( max(tb-Y.t1()+1,0), Y.size() ) );
// Sum
return sequence<R>( vec<R>{v1,v2,v3}, t1 );
}
double f2(double x, sequence a) {
double sum = 0;
for (int i = 0; i < a.size(); i++) {
sum += a[i] * cos(i * x);
}
}
double f1(double x, sequence a) {
double sum = 0;
for (int i = 1; i < a.size(); i++) {
sum += a[i] * sin(i * x);
}
}
int main(int argc, char **argv)
{
using namespace std;
unsigned j;
/*
cout << "Input number of tests (for each pattern size): " << flush;
cin >> Number_Of_Tests;
cout << "Input number of pattern sizes: " << flush;
cin >> Number_Of_Pattern_Sizes;
cout << "Input pattern sizes: " << flush;
*/
if (argc < 4)
return 1;
Number_Of_Tests = strtoul(argv[1], NULL, 10);
Number_Of_Pattern_Sizes = strtoul(argv[2], NULL, 10);
vector<unsigned> Pattern_Size(Number_Of_Pattern_Sizes);
for (j = 0; j < Number_Of_Pattern_Sizes; ++j)
Pattern_Size[j] = strtoul(argv[j + 3], NULL, 10);
cout << "\nNumber of tests: " << Number_Of_Tests << endl;
cout << "Pattern sizes: ";
for (j = 0; j < Number_Of_Pattern_Sizes; ++j)
cout << Pattern_Size[j] << " ";
cout << endl;
ifstream ifs(textFileName);
char C;
while (ifs.get(C))
S1.push_back(C);
cout << S1.size() << " characters read." << endl;
ifstream dictfile(wordFileName);
typedef istream_iterator<string> string_input;
typedef map<int, vector<sequence>, less<int> > map_type;
map_type dictionary;
sequence S;
string S0;
string_input si(dictfile);
while (si != string_input()) {
S0 = *si++;
S.erase(S.begin(), S.end());
copy(S0.begin(), S0.end() - 1, back_inserter(S));
dictionary[S.size()].push_back(S);
}
for (j = 0; j < Number_Of_Pattern_Sizes; ++j) {
vector<sequence>& diction = dictionary[Pattern_Size[j]];
if (diction.size() > Number_Of_Tests) {
vector<sequence> temp;
unsigned Skip_Amount = diction.size() / Number_Of_Tests;
for (unsigned T = 0; T < Number_Of_Tests; ++T) {
temp.push_back(diction[T * Skip_Amount]);
}
diction = temp;
}
Increment = (S1.size() - Pattern_Size[j]) / Number_Of_Tests;
cout << "\n\n-----------------------------------------------------------\n"
<< "Searching for patterns of size " << Pattern_Size[j]
<< "..." << endl;
cout << "(" << Number_Of_Tests << " patterns from the text, "
<< dictionary[Pattern_Size[j]].size() << " from the dictionary)" << endl;
cerr << Pattern_Size[j] << " " << flush;
Base_Time = 0.0;
for (int k = 0; k < number_of_algorithms; ++k) {
if (k != 0)
cout << "Timing " << algorithm_names[k] << ":" << endl;
Run(k, S1, dictionary[Pattern_Size[j]], Pattern_Size[j]);
}
cout << endl;
}
cerr << endl;
}
void sequence_analyzer::analyze(sequence& sequence)
{
for (uint64_t t = 0; t < sequence.size(); ++t)
analyze(sequence, t);
}
void convert_IOE2_IOB2(int is_cap, int is_last, sequence & seq) {
map<int, int> beginlabels;
map<int, int>::iterator blbit;
char begin = is_cap ? 'B' : 'b';
string BEGIN = is_cap ? "B" : "b";
char inside = is_cap ? 'I' : 'i';
string INSIDE = is_cap ? "I" : "i";
char end = is_cap ? 'E' : 'e';
string str1, str2, istr, str, newlabel;
int i, len = seq.size();
for (i = 0; i < len; i++) {
int col2 = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
str2 = seq[i][col2];
if (str2[0] != end && str2[0] != inside) {
continue;
}
if (i == 0) {
beginlabels.insert(pair<int, int>(i, i));
} else {
int col1 = is_last ? seq[i - 1].size() - 1 : seq[i - 1].size() - 2;
str1 = seq[i - 1][col1];
istr = INSIDE;
istr += strtail(str2);
if (str2[0] == end && str1 != istr) {
beginlabels.insert(pair<int, int>(i, i));
} else if (str2[0] == inside && str1 != str2) {
beginlabels.insert(pair<int, int>(i, i));
}
}
}
for (i = 0; i < len; i++) {
int col = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
str = seq[i][col];
if (str[0] != end && str[0] != inside) {
continue;
}
blbit = beginlabels.find(i);
if (blbit != beginlabels.end()) {
newlabel = BEGIN;
newlabel += strtail(str);
seq[i][col] = newlabel;
} else {
newlabel = INSIDE;
newlabel += strtail(str);
seq[i][col] = newlabel;
}
}
}
static void copy_sequence_to_list(list &dst, const sequence &src) {
for(size_t i = 0; i < src.size(); ++i) {
dst.append(src[i]);
}
}
void convert_IOB2_IOE2(int is_cap, int is_last, sequence & seq) {
map<int, int> endlabels;
map<int, int>::iterator elbit;
char begin = is_cap ? 'B' : 'b';
char inside = is_cap ? 'I' : 'i';
string INSIDE = is_cap ? "I" : "i";
char end = is_cap ? 'E' : 'e';
string END = is_cap ? "E" : "e";
string str1, str2, istr, str, newlabel;
int i, len = seq.size();
for (i = 0; i < len; i++) {
int col1 = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
str1 = seq[i][col1];
if (str1[0] != begin && str1[0] != inside) {
continue;
}
if (i == len - 1) {
endlabels.insert(pair<int, int>(i, i));
} else {
int col2 = is_last ? seq[i + 1].size() - 1 : seq[i + 1].size() - 2;
str2 = seq[i + 1][col2];
istr = INSIDE;
istr += strtail(str1);
if (str1[0] == begin && str2 != istr) {
endlabels.insert(pair<int, int>(i, i));
} else if (str1[0] == inside && str2 != str1) {
endlabels.insert(pair<int, int>(i, i));
}
}
}
for (i = 0; i < len; i++) {
int col = is_last ? seq[i].size() - 1 : seq[i].size() - 2;
str = seq[i][col];
if (str[0] != begin && str[0] != inside) {
continue;
}
elbit = endlabels.find(i);
if (elbit != endlabels.end()) {
newlabel = END;
newlabel += strtail(str);
seq[i][col] = newlabel;
} else {
newlabel = INSIDE;
newlabel += strtail(str);
seq[i][col] = newlabel;
}
}
}
