// Exec_DataSetCmd::ChangeOutputFormat()
Exec::RetType Exec_DataSetCmd::ChangeOutputFormat(CpptrajState const& State, ArgList& argIn)
{
TextFormat::FmtType fmt;
if (argIn.hasKey("double"))
fmt = TextFormat::DOUBLE;
else if (argIn.hasKey("scientific"))
fmt = TextFormat::SCIENTIFIC;
else if (argIn.hasKey("general"))
fmt = TextFormat::GDOUBLE;
else {
mprinterr("Error: Expected either 'double', 'scientific', or 'general'\n");
return CpptrajState::ERR;
}
// Loop over all DataSet arguments
std::string ds_arg = argIn.GetStringNext();
while (!ds_arg.empty()) {
DataSetList dsl = State.DSL().GetMultipleSets( ds_arg );
for (DataSetList::const_iterator ds = dsl.begin(); ds != dsl.end(); ++ds)
if ((*ds)->SetupFormat().SetFormatType(fmt))
mprintf("\tSet '%s' output format changed to '%s'\n",
(*ds)->legend(), TextFormat::typeDescription(fmt));
ds_arg = argIn.GetStringNext();
}
return CpptrajState::OK;
}
// Array1D::AddDataSets()
int Array1D::AddDataSets(DataSetList const& SetList) {
for (DataSetList::const_iterator ds = SetList.begin(); ds != SetList.end(); ++ds)
if ( push_back( *ds ) ) {
array_.clear();
return 1;
}
return 0;
}
// Exec_SortEnsembleData::Execute()
Exec::RetType Exec_SortEnsembleData::Execute(CpptrajState& State, ArgList& argIn)
{
debug_ = State.Debug();
DataSetList setsToSort;
std::string dsarg = argIn.GetStringNext();
while (!dsarg.empty()) {
setsToSort += State.DSL().GetMultipleSets( dsarg );
dsarg = argIn.GetStringNext();
}
int err = 0;
# ifdef MPI
// For now, require ensemble mode in parallel.
if (!Parallel::EnsembleIsSetup()) {
rprinterr("Error: Data set ensemble sort requires ensemble mode in parallel.\n");
return CpptrajState::ERR;
}
// Only TrajComm masters have complete data.
if (Parallel::TrajComm().Master()) {
comm_ = Parallel::MasterComm();
# endif
DataSetList OutputSets;
err = SortData( setsToSort, OutputSets );
if (err == 0) {
// Remove unsorted sets.
for (DataSetList::const_iterator ds = setsToSort.begin(); ds != setsToSort.end(); ++ds)
State.DSL().RemoveSet( *ds );
// Add sorted sets.
for (DataSetList::const_iterator ds = OutputSets.begin(); ds != OutputSets.end(); ++ds)
State.DSL().AddSet( *ds );
// Since sorted sets have been transferred to master DSL, OutputSets now
// just has copies.
OutputSets.SetHasCopies( true );
mprintf("\tSorted sets:\n");
OutputSets.List();
}
# ifdef MPI
}
if (Parallel::World().CheckError( err ))
# else
if (err != 0)
# endif
return CpptrajState::ERR;
return CpptrajState::OK;
}
// DataSetList::GetSetsOfType()
DataSetList DataSetList::GetSetsOfType( std::string const& dsargIn, DataSet::DataType typeIn ) const
{
DataSetList dsetOut;
dsetOut.hasCopies_ = true;
DataSetList selected = SelectSets(dsargIn);
for (const_iterator ds = selected.begin(); ds != selected.end(); ++ds)
if ( (*ds)->Type() == typeIn )
dsetOut.Push_Back( *ds );
return dsetOut;
}
// DataIO_Std::WriteData3D()
int DataIO_Std::WriteData3D( CpptrajFile& file, DataSetList const& setList)
{
int err = 0;
for (DataSetList::const_iterator set = setList.begin(); set != setList.end(); ++set)
{
if (set != setList.begin()) file.Printf("\n");
err += WriteSet3D( *(*set), file );
}
return err;
}
// DataSetList::SetPrecisionOfDataSets()
void DataSetList::SetPrecisionOfDataSets(std::string const& nameIn, int widthIn,
int precisionIn)
{
if (widthIn < 1)
mprinterr("Error: Invalid data width (%i)\n", widthIn);
else {
DataSetList Sets = GetMultipleSets( nameIn );
for (DataSetList::const_iterator ds = Sets.begin(); ds != Sets.end(); ++ds)
(*ds)->SetupFormat().SetFormatWidthPrecision(widthIn, precisionIn);
}
}
// Array1D::AddTorsionSets()
int Array1D::AddTorsionSets(DataSetList const& SetList) {
// Ensure data sets are 1D and periodic
for (DataSetList::const_iterator ds = SetList.begin(); ds != SetList.end(); ++ds) {
if ( (*ds)->Meta().IsTorsionArray()) {
if ( push_back( *ds ) ) {
array_.clear();
return 1;
}
} else
mprintf("Warning: Set '%s' is not periodic, skipping.\n", (*ds)->legend());
}
return 0;
}
// DataIO_OpenDx::WriteData()
int DataIO_OpenDx::WriteData(FileName const& fname, DataSetList const& setList)
{
// Open output file
CpptrajFile outfile;
if (outfile.OpenWrite(fname)) {
mprinterr("Error: Could not open OpenDX output file.\n");
return 1;
}
// Warn about writing multiple sets
if (setList.size() > 1)
mprintf("Warning: %s: Writing multiple 3D sets in OpenDX format may result in unexpected behavior\n", fname.full());
int err = 0;
for (DataSetList::const_iterator set = setList.begin(); set != setList.end(); ++set)
err += WriteSet3D( *(*set), outfile );
return err;
}
// Exec_DataSetCmd::Concatenate()
Exec::RetType Exec_DataSetCmd::Concatenate(CpptrajState& State, ArgList& argIn) {
std::string name = argIn.GetStringKey("name");
bool use_offset = !argIn.hasKey("nooffset");
DataSet* ds3 = State.DSL().AddSet( DataSet::XYMESH, name, "CAT" );
if (ds3 == 0) return CpptrajState::ERR;
DataSet_1D& out = static_cast<DataSet_1D&>( *ds3 );
mprintf("\tConcatenating sets into '%s'\n", out.legend());
if (use_offset)
mprintf("\tX values will be offset.\n");
else
mprintf("\tX values will not be offset.\n");
std::string dsarg = argIn.GetStringNext();
double offset = 0.0;
while (!dsarg.empty()) {
DataSetList dsl = State.DSL().GetMultipleSets( dsarg );
double XY[2];
for (DataSetList::const_iterator ds = dsl.begin(); ds != dsl.end(); ++ds)
{
if ( (*ds)->Group() != DataSet::SCALAR_1D )
{
mprintf("Warning: '%s': Concatenation only supported for 1D scalar data sets.\n",
(*ds)->legend());
} else {
DataSet_1D const& set = static_cast<DataSet_1D const&>( *(*ds) );
mprintf("\t\t'%s'\n", set.legend());
for (size_t i = 0; i != set.Size(); i++) {
XY[0] = set.Xcrd( i ) + offset;
XY[1] = set.Dval( i );
out.Add( i, XY ); // NOTE: value of i does not matter for mesh
}
if (use_offset) offset = XY[0];
}
}
dsarg = argIn.GetStringNext();
}
return CpptrajState::OK;
}
// DataIO_Std::WriteCmatrix()
int DataIO_Std::WriteCmatrix(CpptrajFile& file, DataSetList const& Sets) {
for (DataSetList::const_iterator ds = Sets.begin(); ds != Sets.end(); ++ds)
{
if ( (*ds)->Group() != DataSet::CLUSTERMATRIX) {
mprinterr("Error: Write of cluster matrix and other sets to same file not supported.\n"
"Error: Skipping '%s'\n", (*ds)->legend());
continue;
}
DataSet_Cmatrix const& cm = static_cast<DataSet_Cmatrix const&>( *(*ds) );
int nrows = cm.OriginalNframes();
int col_width = std::max(3, DigitWidth( nrows ) + 1);
int dat_width = std::max(cm.Format().Width(), (int)cm.Meta().Legend().size()) + 1;
WriteNameToBuffer(file, "F1", col_width, true);
WriteNameToBuffer(file, "F2", col_width, false);
WriteNameToBuffer(file, cm.Meta().Legend(), dat_width, false);
if (cm.SieveType() != ClusterSieve::NONE)
file.Printf(" nframes %i", cm.OriginalNframes());
file.Printf("\n");
TextFormat col_fmt(TextFormat::INTEGER, col_width);
TextFormat dat_fmt = cm.Format();
dat_fmt.SetFormatAlign(TextFormat::RIGHT);
dat_fmt.SetFormatWidth( dat_width );
std::string total_fmt = col_fmt.Fmt() + col_fmt.Fmt() + dat_fmt.Fmt() + "\n";
//mprintf("DEBUG: format '%s'\n", total_fmt.c_str());
ClusterSieve::SievedFrames const& frames = cm.FramesToCluster();
int ntotal = (int)frames.size();
for (int idx1 = 0; idx1 != ntotal; idx1++) {
int row = frames[idx1];
for (int idx2 = idx1 + 1; idx2 != ntotal; idx2++) {
int col = frames[idx2];
file.Printf(total_fmt.c_str(), row+1, col+1, cm.GetFdist(col, row));
}
}
}
return 0;
}
// DataIO_Std::WriteDataInverted()
int DataIO_Std::WriteDataInverted(CpptrajFile& file, DataSetList const& Sets)
{
if (Sets.empty() || CheckAllDims(Sets, 1)) return 1;
// Determine size of largest DataSet.
size_t maxFrames = DetermineMax( Sets );
// Write each set to a line.
DataSet::SizeArray positions(1);
// Set up x column format
DataSetList::const_iterator set = Sets.begin();
TextFormat x_col_format;
if (hasXcolumn_)
x_col_format = XcolFmt();
else
x_col_format = (*set)->Format();
for (; set != Sets.end(); ++set) {
// Write dataset name as first column.
WriteNameToBuffer( file, (*set)->Meta().Legend(), x_col_format.ColumnWidth(), false);
// Write each frame to subsequent columns
for (positions[0] = 0; positions[0] < maxFrames; positions[0]++)
(*set)->WriteBuffer(file, positions);
file.Printf("\n");
}
return 0;
}
Analysis::RetType Analysis_AutoCorr::Setup(ArgList& analyzeArgs, AnalysisSetup& setup, int debugIn)
{
const char* calctype;
std::string setname = analyzeArgs.GetStringKey("name");
DataFile* outfile = setup.DFL().AddDataFile( analyzeArgs.GetStringKey("out"), analyzeArgs );
lagmax_ = analyzeArgs.getKeyInt("lagmax",-1);
calc_covar_ = !analyzeArgs.hasKey("nocovar");
usefft_ = !analyzeArgs.hasKey("direct");
// Select datasets from remaining args
dsets_.clear();
ArgList dsetArgs = analyzeArgs.RemainingArgs();
for (ArgList::const_iterator dsa = dsetArgs.begin(); dsa != dsetArgs.end(); ++dsa) {
DataSetList setsIn = setup.DSL().GetMultipleSets( *dsa );
for (DataSetList::const_iterator ds = setsIn.begin(); ds != setsIn.end(); ++ds) {
if ( (*ds)->Group() != DataSet::SCALAR_1D && (*ds)->Type() != DataSet::VECTOR )
mprintf("Warning: Set '%s' type not supported in AUTOCORR - skipping.\n",
(*ds)->legend());
else
dsets_.push_back( *ds );
}
}
if (dsets_.empty()) {
mprinterr("Error: No data sets selected.\n");
return Analysis::ERR;
}
// If setname is empty generate a default name
if (setname.empty())
setname = setup.DSL().GenerateDefaultName( "autocorr" );
// Setup output datasets
MetaData md( setname );
for (unsigned int idx = 0; idx != dsets_.size(); idx++) {
md.SetIdx( idx );
DataSet* dsout = setup.DSL().AddSet( DataSet::DOUBLE, md );
if (dsout==0) return Analysis::ERR;
dsout->SetLegend( dsets_[idx]->Meta().Legend() );
outputData_.push_back( dsout );
// Add set to output file
if (outfile != 0) outfile->AddDataSet( outputData_.back() );
}
if (calc_covar_)
calctype = "covariance";
else
calctype = "correlation";
mprintf(" AUTOCORR: Calculating auto-%s for %i data sets:\n\t", calctype, dsets_.size());
for (unsigned int idx = 0; idx != dsets_.size(); ++idx)
mprintf(" %s", dsets_[idx]->legend());
mprintf("\n");
if (lagmax_!=-1)
mprintf("\tLag max= %i\n", lagmax_);
if ( !setname.empty() )
mprintf("\tSet name: %s\n", setname.c_str() );
if ( outfile != 0 )
mprintf("\tOutfile name: %s\n", outfile->DataFilename().base());
if (usefft_)
mprintf("\tUsing FFT to calculate %s.\n", calctype);
else
mprintf("\tUsing direct method to calculate %s.\n", calctype);
return Analysis::OK;
}
// Exec_SortEnsembleData::Sort_pH_Data()
int Exec_SortEnsembleData::Sort_pH_Data(DataSetList const& setsToSort, DataSetList& OutputSets,
unsigned int maxFrames)
const
{
// Cast sets back to DataSet_PHREMD
typedef std::vector<DataSet_PHREMD*> Parray;
Parray PHsets;
for (DataSetList::const_iterator ds = setsToSort.begin(); ds != setsToSort.end(); ++ds)
PHsets.push_back( (DataSet_PHREMD*)*ds );
// Gather initial pH data values, ensure no duplicates
typedef std::vector<double> Darray;
Darray pHvalues;
# ifdef MPI
pHvalues.resize( Parallel::Ensemble_Size() );
Darray phtmp;
for (Parray::const_iterator ds = PHsets.begin(); ds != PHsets.end(); ++ds)
phtmp.push_back( (*ds)->Initial_pH() );
if (comm_.AllGather(&phtmp[0], phtmp.size(), MPI_DOUBLE, &pHvalues[0])) {
rprinterr("Error: Gathering pH values.\n");
return 1;
}
# else
for (Parray::const_iterator ds = PHsets.begin(); ds != PHsets.end(); ++ds)
pHvalues.push_back( (*ds)->Initial_pH() );
# endif
ReplicaInfo::Map<double> pH_map;
if (pH_map.CreateMap( pHvalues )) {
rprinterr("Error: Duplicate pH value detected (%.2f) in ensemble.\n", pH_map.Duplicate());
return 1;
}
Darray sortedPH;
mprintf("\tInitial pH values:");
for (ReplicaInfo::Map<double>::const_iterator ph = pH_map.begin(); ph != pH_map.end(); ++ph)
{
mprintf(" %6.2f", ph->first);
sortedPH.push_back( ph->first );
}
mprintf("\n");
// Create sets to hold sorted pH values. Create a set for each pH value
// and each residue. Final output sets will be PH0R0, PH0R1, PH1R0, ...
// TODO check that residue info all the same
DataSet_PHREMD::Rarray const& Residues = PHsets[0]->Residues();
int defaultState = 0;
# ifdef MPI
if ( PHsets[0]->Type() == DataSet::PH_IMPL)
defaultState = -1;
# endif
if (debug_ > 0)
rprintf("DEBUG: Sorting %u frames for %zu sets, %zu pH values.\n",
maxFrames, PHsets.size(), sortedPH.size());
for (unsigned int idx = 0; idx != sortedPH.size(); idx++) {
OutputSets.SetEnsembleNum( idx );
for (unsigned int res = 0; res != Residues.size(); ++res) {
MetaData md(PHsets[0]->Meta().Name(), Residues[res].Name().Truncated(), Residues[res].Num());
DataSet_pH* out = (DataSet_pH*)OutputSets.AddSet( DataSet::PH, md );
if (out==0) return 1;
//out->SetLegend( "pH " + doubleToString( sortedPH[idx] ) );
out->Set_Solvent_pH( sortedPH[idx] );
out->SetResidueInfo( Residues[res] );
out->SetTimeValues(PHsets[0]->Time());
out->Resize( maxFrames, defaultState );
}
}
// ---------------------------------------------
if ( PHsets[0]->Type() == DataSet::PH_EXPL) {
// Loop over unsorted sets
for (Parray::const_iterator ds = PHsets.begin(); ds != PHsets.end(); ++ds)
{
DataSet_PHREMD_Explicit* in = (DataSet_PHREMD_Explicit*)*ds;
unsigned int phidx = 0;
for (unsigned int n = 0; n < maxFrames; n++)
{
float phval = in->pH_Values()[n];
int setidx = pH_map.FindIndex( phval ) * Residues.size();
//rprintf("DEBUG: %6u Set %10s pH= %6.2f going to %2i\n", n+1, in->legend(), phval, idx);
//mflush();
for (unsigned int res = 0; res < in->Residues().size(); res++, setidx++, phidx++)
{
DataSet_pH* out = (DataSet_pH*)OutputSets[setidx];
//if (res == 0 && idx == 0) {
// rprintf("DEBUG: Frame %3u res %2u State %2i pH %6.2f\n",
// n, res, in->Res(res).State(n), phval);
// mflush();
//}
out->SetState(n, in->ResStates()[phidx], in->RecordType(n));
}
}
} // END loop over unsorted sets
# ifdef MPI
// Now we need to reduce down each set onto the thread where it belongs.
if (Parallel::World().Size() > 1) {
for (int idx = 0; idx != (int)OutputSets.size(); idx++) {
DataSet_pH* out = (DataSet_pH*)OutputSets[idx];
int ensembleRank = Parallel::MemberEnsCommRank( out->Meta().EnsembleNum() );
//rprintf("DEBUG: Reduce set %s to rank %i\n", out->legend(), ensembleRank);
out->Reduce( comm_, ensembleRank );
}
// Remove sets that do not belong on this rank
for (int idx = (int)OutputSets.size() - 1; idx > -1; idx--) {
DataSet* out = OutputSets[idx];
int ensembleRank = Parallel::MemberEnsCommRank( out->Meta().EnsembleNum() );
if (ensembleRank != comm_.Rank()) {
//rprintf("DEBUG: Remove set %s (%i) from rank %i\n", out->legend(),
// idx, comm_.Rank());
OutputSets.RemoveSet( out );
}
}
}
# endif
// ---------------------------------------------
} else if ( PHsets[0]->Type() == DataSet::PH_IMPL) {
# ifdef MPI
typedef std::vector<int> Iarray;
typedef std::vector<Iarray> Iarray2;
typedef std::vector<bool> Barray;
// True if I have this pH value
Barray isMyPh( sortedPH.size(), false );
// Which rank in ensemble has which pH
Iarray pHrank( sortedPH.size(), 0 );
for (int phidx = 0; phidx != (int)sortedPH.size(); phidx++)
{
int ensembleRank = Parallel::MemberEnsCommRank( phidx );
pHrank[phidx] = ensembleRank;
isMyPh[phidx] = (ensembleRank == comm_.Rank());
}
// DEBUG
for (unsigned int idx = 0; idx != pHrank.size(); idx++)
mprintf("\tpH %6.2f on rank %i\n", sortedPH[idx], pHrank[idx]);
// Hold frame-residue-state for each pH not on this rank.
Iarray2 FrmResState( sortedPH.size() );
// Loop over unsorted sets
for (Parray::const_iterator ds = PHsets.begin(); ds != PHsets.end(); ++ds)
{
DataSet_PHREMD_Implicit* in = (DataSet_PHREMD_Implicit*)*ds;
// Loop over frames
for (unsigned int n = 0; n < maxFrames; n++)
{
DataSet_PHREMD_Implicit::Record const& Rec = in->Records()[n];
float phval = Rec.pH();
int phidx = pH_map.FindIndex( phval );
if (isMyPh[phidx]) {
// This pH belongs to me. Set value.
int setidx = phidx * Residues.size();
if (Rec.RecType() == Cph::FULL_RECORD) {
// Info for all residues
for (unsigned int res = 0; res < in->Residues().size(); res++, setidx++)
((DataSet_pH*)OutputSets[setidx])->SetState(n, Rec.ResStates()[res], Rec.RecType());
} else if (Rec.RecType() > -1) {
// Info for single residue, record type for all residues
//rprintf("\tSetting my pH %6.2f frame %8i state %2i idx %6i res %6i '%s'\n", sortedPH[phidx], n, Rec.ResStates()[0], setidx, Rec.RecType(), OutputSets[setidx+Rec.RecType()]->legend());
for (int res = 0; res < (int)in->Residues().size(); res++, setidx++)
if (res == Rec.RecType())
((DataSet_pH*)OutputSets[setidx])->SetState(n, Rec.ResStates()[0], Rec.RecType());
else
((DataSet_pH*)OutputSets[setidx])->SetRecType(n, Rec.RecType());
}
} else {
// This pH belongs to another rank. Save it.
if (Rec.RecType() > -1) {
// Info for a single residue present
FrmResState[phidx].push_back( n );
FrmResState[phidx].push_back( Rec.RecType() );
FrmResState[phidx].push_back( Rec.ResStates()[0] );
} else {
// Info for all residues present
FrmResState[phidx].push_back( n );
FrmResState[phidx].push_back( Rec.RecType() );
for (unsigned int res = 0; res < in->Residues().size(); res++)
FrmResState[phidx].push_back( Rec.ResStates()[res] );
}
}
} // END loop over frames
} // END loop over sets
// DEBUG
/*
comm_.Barrier();
for (int rank = 0; rank < comm_.Size(); rank++)
{
if (rank == comm_.Rank())
{
for (unsigned int phidx = 0; phidx != sortedPH.size(); phidx++) {
rprintf("DEBUG: pH %6.2f: %8s %6s %2s\n", sortedPH[phidx], "Frm", "Res", "St");
Iarray const& FRS = FrmResState[phidx];
unsigned int idx = 0;
while (idx < FRS.size()) {
int rec = FRS[idx+1];
if (rec > -1) {
rprintf(" %8i %6i %2i\n", FRS[idx], rec, FRS[idx+2]);
idx += 3;
} else {
rprintf(" %8i %6i All Residues\n", FRS[idx], rec);
idx += (2 + Residues.size());
}
}
}
}
comm_.Barrier();
}
*/
// Communicate states to other ranks
typedef std::vector<unsigned int> Uarray;
Uarray sizeOnRank( comm_.Size() );
for (unsigned int phidx = 0; phidx != sortedPH.size(); phidx++)
{
// Each rank says how many frames of this pH they have collected and
// send to rank the pH belongs to.
unsigned int nph = FrmResState[phidx].size();
comm_.Gather(&nph, 1, MPI_UNSIGNED, &sizeOnRank[0], pHrank[phidx]);
if (pHrank[phidx] == comm_.Rank())
{
// This pH belongs to me. I should have no frames at this pH.
if (sizeOnRank[comm_.Rank()] > 0) {
rprinterr("Internal Error: Rank has frames to communicate at its pH\n");
Parallel::Abort(1);
}
unsigned int totalSize = 0;
for (unsigned int idx = 0; idx != sizeOnRank.size(); idx++) {
totalSize += sizeOnRank[idx];
//rprintf("DEBUG: Rank %4u has %8u frames of pH %6.2f\n",
// idx, sizeOnRank[idx], sortedPH[phidx]);
}
//rprintf("DEBUG: Total incoming size: %u\n", totalSize);
FrmResState[phidx].resize( totalSize );
// Receive state info for this pH from other ranks
int* frsArray = &(FrmResState[phidx][0]);
for (int rank = 0; rank != comm_.Size(); rank++) {
if (rank != comm_.Rank()) {
comm_.Recv(frsArray, sizeOnRank[rank], MPI_INT, rank, 1600+rank);
frsArray += sizeOnRank[rank];
}
}
} else {
// This pH belongs to another rank. Send my info there.
int* frsArray = &(FrmResState[phidx][0]);
comm_.Send(frsArray, nph, MPI_INT, pHrank[phidx], 1600+comm_.Rank());
}
comm_.Barrier();
}
// Fill in state info
std::vector<DataSet*> ToRemove;
for (unsigned int phidx = 0; phidx != sortedPH.size(); phidx++)
{
int setidx = phidx * Residues.size();
if (pHrank[phidx] == comm_.Rank())
{
Iarray const& FRS = FrmResState[phidx];
// This pH belongs to me. Fill in the information received from
// other ranks.
unsigned int idx = 0;
while (idx < FRS.size()) {
int rec = FRS[idx+1];
if (rec > -1) {
// Info for single residue, record type for all residues
//rprintf("\tSetting pH %6.2f frame %8i state %2i idx %6i res %6i '%s'\n", sortedPH[phidx], FRS[idx], FRS[idx+2], setidx, rec, OutputSets[setidx+rec]->legend());
for (int res = 0; res != (int)Residues.size(); res++) {
DataSet_pH* out = (DataSet_pH*)OutputSets[setidx + res];
if (rec == res)
out->SetState( FRS[idx], FRS[idx+2], rec );
else
out->SetRecType( FRS[idx], rec );
}
idx += 3;
} else {
//rprintf(" %8i %6i All Residues\n", FRS[idx], rec);
int frm = FRS[idx];
idx += 2;
for (unsigned int res = 0; res != Residues.size(); res++, idx++) {
DataSet_pH* out = (DataSet_pH*)OutputSets[setidx + res];
out->SetState( frm, FRS[idx], rec );
}
}
}
// Fill in any remaining data. FIXME safe to assume first frame is set?
for (unsigned int res = 0; res != Residues.size(); res++) {
DataSet_pH* out = (DataSet_pH*)OutputSets[setidx + res];
for (unsigned int n = 1; n < maxFrames; n++)
if (out->State(n) == -1)
out->SetState(n, out->State(n-1), out->RecordType(n));
}
} else {
// This pH does not belong to me. Mark associated data sets to be removed.
for (unsigned int res = 0; res != Residues.size(); res++)
ToRemove.push_back( OutputSets[setidx + res] );
}
}
// Remove data sets that do not belong to me.
for (std::vector<DataSet*>::reverse_iterator it = ToRemove.rbegin();
it != ToRemove.rend(); ++it)
{
//rprintf("DEBUG: '%s' does not belong to me.\n", (*it)->legend());
OutputSets.RemoveSet( *it );
}
# else /* if not MPI */
// Loop over frames
for (unsigned int n = 0; n < maxFrames; n++)
{
// Loop over unsorted sets
for (Parray::const_iterator ds = PHsets.begin(); ds != PHsets.end(); ++ds)
{
DataSet_PHREMD_Implicit* in = (DataSet_PHREMD_Implicit*)*ds;
DataSet_PHREMD_Implicit::Record const& Rec = in->Records()[n];
float phval = Rec.pH();
int setidx = pH_map.FindIndex( phval ) * Residues.size();
if (Rec.RecType() == Cph::FULL_RECORD) {
for (unsigned int res = 0; res < in->Residues().size(); res++, setidx++)
{
DataSet_pH* out = (DataSet_pH*)OutputSets[setidx];
//if (res == 0 && idx == 0) {
// rprintf("DEBUG: Frame %3u res %2u State %2i pH %6.2f\n",
// n, res, in->Res(res).State(n), phval);
// mflush();
//}
out->SetState(n, Rec.ResStates()[res], Rec.RecType());
}
} else {
for (int res = 0; res < (int)in->Residues().size(); res++, setidx++)
{
DataSet_pH* out = (DataSet_pH*)OutputSets[setidx];
if (res == Rec.RecType())
out->SetState(n, Rec.ResStates()[0], Rec.RecType());
else
// State for this residue not recorded - use previous state.
// Should be fine since first state always has all residues.
out->SetState(n, out->State(n-1), Rec.RecType());
}
}
} // END loop over unsorted sets
} // END loop over frames
# endif /* MPI */
// ---------------------------------------------
} else {
return 1; // Sanity check
}
return 0;
}
// DataIO_Std::WriteDataNormal()
int DataIO_Std::WriteDataNormal(CpptrajFile& file, DataSetList const& Sets) {
// Assume all 1D data sets.
if (Sets.empty() || CheckAllDims(Sets, 1)) return 1;
// For this output to work the X-dimension of all sets needs to match.
// The most important things for output are min and step so just check that.
// Use X dimension of set 0 for all set dimensions.
CheckXDimension( Sets );
// TODO: Check for empty dim.
DataSet* Xdata = Sets[0];
Dimension const& Xdim = static_cast<Dimension const&>( Xdata->Dim(0) );
int xcol_width = Xdim.Label().size() + 1; // Only used if hasXcolumn_
if (xcol_width < 8) xcol_width = 8;
int xcol_precision = 3;
// Determine size of largest DataSet.
size_t maxFrames = DetermineMax( Sets );
// Set up X column.
TextFormat x_col_format(XcolFmt());
if (hasXcolumn_) {
if (XcolPrecSet()) {
xcol_width = XcolWidth();
x_col_format = TextFormat(XcolFmt(), XcolWidth(), XcolPrec());
} else {
// Create format string for X column based on dimension in first data set.
// Adjust X col precision as follows: if the step is set and has a
// fractional component set the X col width/precision to either the data
// width/precision or the current width/precision, whichever is larger. If
// the set is XYMESH but step has not been set (so we do not know spacing
// between X values) use default precision. Otherwise the step has no
// fractional component so make the precision zero.
double step_i;
double step_f = modf( Xdim.Step(), &step_i );
double min_f = modf( Xdim.Min(), &step_i );
if (Xdim.Step() > 0.0 && (step_f > 0.0 || min_f > 0.0)) {
xcol_precision = std::max(xcol_precision, Xdata->Format().Precision());
xcol_width = std::max(xcol_width, Xdata->Format().Width());
} else if (Xdata->Type() != DataSet::XYMESH)
xcol_precision = 0;
x_col_format.SetCoordFormat( maxFrames, Xdim.Min(), Xdim.Step(), xcol_width, xcol_precision );
}
} else {
// If not writing an X-column, no leading space for the first dataset.
Xdata->SetupFormat().SetFormatAlign( TextFormat::RIGHT );
}
// Write header to buffer
std::vector<int> Original_Column_Widths;
if (writeHeader_) {
// If x-column present, write x-label
if (hasXcolumn_)
WriteNameToBuffer( file, Xdim.Label(), xcol_width, true );
// To prevent truncation of DataSet legends, adjust the width of each
// DataSet if necessary.
for (DataSetList::const_iterator ds = Sets.begin(); ds != Sets.end(); ++ds) {
// Record original column widths in case they are changed.
Original_Column_Widths.push_back( (*ds)->Format().Width() );
int colLabelSize;
if (ds == Sets.begin() && !hasXcolumn_)
colLabelSize = (int)(*ds)->Meta().Legend().size() + 1;
else
colLabelSize = (int)(*ds)->Meta().Legend().size();
//mprintf("DEBUG: Set '%s', fmt width= %i, colWidth= %i, colLabelSize= %i\n",
// (*ds)->legend(), (*ds)->Format().Width(), (*ds)->Format().ColumnWidth(),
// colLabelSize);
if (colLabelSize >= (*ds)->Format().ColumnWidth())
(*ds)->SetupFormat().SetFormatWidth( colLabelSize );
}
// Write dataset names to header, left-aligning first set if no X-column
DataSetList::const_iterator set = Sets.begin();
if (!hasXcolumn_)
WriteNameToBuffer( file, (*set)->Meta().Legend(), (*set)->Format().ColumnWidth(), true );
else
WriteNameToBuffer( file, (*set)->Meta().Legend(), (*set)->Format().ColumnWidth(), false );
++set;
for (; set != Sets.end(); ++set)
WriteNameToBuffer( file, (*set)->Meta().Legend(), (*set)->Format().ColumnWidth(), false );
file.Printf("\n");
}
// Write Data
DataSet::SizeArray positions(1);
for (positions[0] = 0; positions[0] < maxFrames; positions[0]++) {
// Output Frame for each set
if (hasXcolumn_)
file.Printf( x_col_format.fmt(), Xdata->Coord(0, positions[0]) );
for (DataSetList::const_iterator set = Sets.begin(); set != Sets.end(); ++set)
(*set)->WriteBuffer(file, positions);
file.Printf("\n");
}
// Restore original column widths if necessary
if (!Original_Column_Widths.empty())
for (unsigned int i = 0; i != Original_Column_Widths.size(); i++)
Sets[i]->SetupFormat().SetFormatWidth( Original_Column_Widths[i] );
return 0;
}
// Exec_SortEnsembleData::SortData()
int Exec_SortEnsembleData::SortData(DataSetList const& setsToSort, DataSetList& OutputSets)
const
{
int err = 0;
if (setsToSort.empty()) {
rprinterr("Error: No sets selected.\n");
err = 1;
}
if (CheckError(err)) return 1;
mprintf("\tSorting the following sets:\n");
setsToSort.List();
# ifdef MPI
// Number of sets to sort should be equal to # members I am responsible for.
if (Parallel::N_Ens_Members() != (int)setsToSort.size()) {
rprinterr("Internal Error: Number of ensemble members (%i) != # sets to sort (%zu)\n",
Parallel::N_Ens_Members(), setsToSort.size());
return 1;
}
# endif
DataSet::DataType dtype = setsToSort[0]->Type();
unsigned int maxSize = 0;
for (DataSetList::const_iterator ds = setsToSort.begin(); ds != setsToSort.end(); ++ds) {
if ((*ds)->Size() < 1) { //TODO check sizes match
rprinterr("Error: Set '%s' is empty.\n", (*ds)->legend());
err = 1;
break;
}
if (ds == setsToSort.begin())
maxSize = (*ds)->Size();
else if ((*ds)->Size() < maxSize) {
rprintf("Warning: Set '%s' has fewer frames (%zu) than previous set(s) (%u)\n"
"Warning: Only using the first %zu frames of all sets.\n",
(*ds)->legend(), (*ds)->Size(), maxSize, (*ds)->Size());
maxSize = (unsigned int)(*ds)->Size();
} else if ((*ds)->Size() > maxSize) {
rprintf("Warning: Set '%s' has more frames (%zu) than previous set(s) (%u)\n"
"Warning: Only using the first %u frames of all sets.\n",
(*ds)->legend(), (*ds)->Size(), maxSize, maxSize);
}
if (dtype != (*ds)->Type()) {
rprinterr("Error: Set '%s' has different type than first set.\n", (*ds)->legend());
err = 1;
break;
}
}
if (CheckError(err)) return 1;
# ifdef MPI
unsigned int threadSize = maxSize;
comm_.AllReduce( &maxSize, &threadSize, 1, MPI_UNSIGNED, MPI_MIN );
typedef std::vector<int> Iarray;
Iarray Dtypes( comm_.Size(), -1 );
if ( comm_.AllGather( &dtype, 1, MPI_INT, &Dtypes[0] ) ) return 1;
for (int rank = 1; rank < comm_.Size(); rank++)
if (Dtypes[0] != Dtypes[rank]) {
rprinterr("Error: Set types on rank %i do not match types on rank 0.\n", rank);
err = 1;
break;
}
if (comm_.CheckError( err )) return 1;
# endif
// Only work for pH data for now.
if (dtype != DataSet::PH_EXPL && dtype != DataSet::PH_IMPL) {
rprinterr("Error: Only works for pH REMD data for now.\n");
return 1;
}
err = Sort_pH_Data( setsToSort, OutputSets, maxSize );
return err;
}
// Exec_DataSetCmd::ModifyPoints()
Exec::RetType Exec_DataSetCmd::ModifyPoints(CpptrajState& State, ArgList& argIn, bool drop) {
const char* mode;
if (drop)
mode = "Drop";
else
mode = "Kee";
// Keywords
std::string name = argIn.GetStringKey("name");
int start = argIn.getKeyInt("start", 0) - 1;
int stop = argIn.getKeyInt("stop", -1);
int offset = argIn.getKeyInt("offset", -1);
Range points;
if (start < 0 && stop < 0 && offset < 0) {
std::string rangearg = argIn.GetStringKey("range");
if (rangearg.empty()) {
mprinterr("Error: Must specify range or start/stop/offset.\n");
return CpptrajState::ERR;
}
points.SetRange( rangearg );
if (points.Empty()) {
mprinterr("Error: Range '%s' is empty.\n", rangearg.c_str());
return CpptrajState::ERR;
}
mprintf("\t%sping points in range %s\n", mode, rangearg.c_str());
// User args start from 1
points.ShiftBy(-1);
}
// Get data set to drop/keep points from
// Loop over all DataSet arguments
std::string ds_arg = argIn.GetStringNext();
while (!ds_arg.empty()) {
DataSetList dsl = State.DSL().GetMultipleSets( ds_arg );
for (DataSetList::const_iterator it = dsl.begin(); it != dsl.end(); ++it)
{
DataSet* DS = *it;
if (DS->Size() < 1) {
mprinterr("Error: Set '%s' is empty.\n", DS->legend());
return CpptrajState::ERR;
}
// Restrict to 1D sets for now TODO more types
if (DS->Group() != DataSet::SCALAR_1D) {
mprinterr("Error: Currently only works for 1D scalar data sets.\n");
return CpptrajState::ERR;
}
DataSet_1D* ds1 = (DataSet_1D*)DS;
// Output data set
DataSet* out = 0;
if (name.empty()) {
// Modifying this set. Create new temporary set.
out = State.DSL().Allocate( ds1->Type() );
if (out == 0) return CpptrajState::ERR;
*out = *ds1;
mprintf("\tOverwriting set '%s'\n", ds1->legend());
} else {
// Write to new set
MetaData md = ds1->Meta();
md.SetName( name );
out = State.DSL().AddSet(ds1->Type(), md);
if (out == 0) return CpptrajState::ERR;
mprintf("\tNew set is '%s'\n", out->Meta().PrintName().c_str());
}
out->Allocate(DataSet::SizeArray(1, ds1->Size()));
if (points.Empty()) {
// Drop by start/stop/offset. Set defaults if needed
if (start < 0) start = 0;
if (stop < 0) stop = ds1->Size();
if (offset < 0) offset = 1;
mprintf("\t%sping points from %i to %i, step %i\n", mode, start+1, stop, offset);
for (int idx = start; idx < stop; idx += offset)
points.AddToRange( idx );
} // TODO check that range values are valid?
if (State.Debug() > 0) mprintf("DEBUG: Keeping points:");
Range::const_iterator pt = points.begin();
int idx = 0;
int odx = 0;
if (drop) {
// Drop points
for (; idx < (int)ds1->Size(); idx++) {
if (pt == points.end()) break;
if (*pt != idx) {
if (State.Debug() > 0) mprintf(" %i", idx + 1);
KeepPoint(ds1, out, idx, odx);
} else
++pt;
}
// Keep all remaining points
for (; idx < (int)ds1->Size(); idx++) {
if (State.Debug() > 0) mprintf(" %i", idx + 1);
KeepPoint(ds1, out, idx, odx);
}
} else {
// Keep points
for (; pt != points.end(); pt++) {
if (*pt >= (int)ds1->Size()) break;
if (State.Debug() > 0) mprintf(" %i", *pt + 1);
KeepPoint(ds1, out, *pt, odx);
}
}
if (State.Debug() > 0) mprintf("\n");
if (name.empty()) {
// Replace old set with new set
State.DSL().RemoveSet( ds1 );
State.DSL().AddSet( out );
}
} // END loop over sets
ds_arg = argIn.GetStringNext();
} // END loop over set args
return CpptrajState::OK;
}
// Analysis_HausdorffDistance::Setup()
Analysis::RetType Analysis_HausdorffDistance::Setup(ArgList& analyzeArgs, AnalysisSetup& setup, int debugIn)
{
// Keywords
int nrows = -1;
int ncols = -1;
std::string outtypearg = analyzeArgs.GetStringKey("outtype");
if (!outtypearg.empty()) {
if (outtypearg == "basic")
outType_ = BASIC;
else if (outtypearg == "trimatrix") {
outType_ = UPPER_TRI_MATRIX;
nrows = analyzeArgs.getKeyInt("nrows", -1);
if (nrows < 1) {
mprinterr("Error: 'nrows' must be specified and > 0 for 'trimatrix'\n");
return Analysis::ERR;
}
} else if (outtypearg == "fullmatrix") {
outType_ = FULL_MATRIX;
nrows = analyzeArgs.getKeyInt("nrows", -1);
if (nrows < 1) {
mprinterr("Error: 'nrows' must be specified and > 0 for 'fullmatrix'\n");
return Analysis::ERR;
}
ncols = analyzeArgs.getKeyInt("ncols", nrows);
if (ncols < 1) {
mprinterr("Error: 'ncols' must be > 0 for 'fullmatrix'\n");
return Analysis::ERR;
}
} else {
mprinterr("Error: Unrecognized keyword for 'outtype': %s\n", outtypearg.c_str());
return Analysis::ERR;
}
} else
outType_ = BASIC;
std::string dsname = analyzeArgs.GetStringKey("name");
DataFile* df = setup.DFL().AddDataFile( analyzeArgs.GetStringKey("out"), analyzeArgs );
DataFile* dfab = setup.DFL().AddDataFile( analyzeArgs.GetStringKey("outab"), analyzeArgs );
DataFile* dfba = setup.DFL().AddDataFile( analyzeArgs.GetStringKey("outba"), analyzeArgs );
// Get input data sets
std::string dsarg = analyzeArgs.GetStringNext();
while (!dsarg.empty()) {
DataSetList selected = setup.DSL().GetMultipleSets( dsarg );
for (DataSetList::const_iterator set = selected.begin(); set != selected.end(); ++set)
{
if ((*set)->Group() == DataSet::MATRIX_2D)
inputSets_.AddCopyOfSet( *set );
else
mprintf("Warning: Currently only 2D matrices supported; skipping set '%s'\n",
(*set)->legend());
}
//inputSets_ += setup.DSL().GetMultipleSets( dsarg );
dsarg = analyzeArgs.GetStringNext();
}
if (inputSets_.empty()) {
mprinterr("Error: No data sets specified.\n");
return Analysis::ERR;
}
// Output data set
out_ = 0;
if (outType_ == BASIC) {
out_ = setup.DSL().AddSet(DataSet::FLOAT, dsname, "HAUSDORFF");
if (out_ == 0) return Analysis::ERR;
// Directed sets
ab_out_ = setup.DSL().AddSet(DataSet::FLOAT, MetaData(out_->Meta().Name(),"AB"));
if (ab_out_ == 0) return Analysis::ERR;
ba_out_ = setup.DSL().AddSet(DataSet::FLOAT, MetaData(out_->Meta().Name(),"BA"));
if (ba_out_ == 0) return Analysis::ERR;
} else if (outType_ == UPPER_TRI_MATRIX || outType_ == FULL_MATRIX) {
out_ = setup.DSL().AddSet(DataSet::MATRIX_FLT, dsname, "HAUSDORFF");
ab_out_ = setup.DSL().AddSet(DataSet::MATRIX_FLT, MetaData(out_->Meta().Name(),"AB"));
ba_out_ = setup.DSL().AddSet(DataSet::MATRIX_FLT, MetaData(out_->Meta().Name(),"BA"));
if (out_ == 0 || ab_out_ == 0 || ba_out_ == 0) return Analysis::ERR;
if (outType_ == UPPER_TRI_MATRIX) {
if (((DataSet_2D*)out_)->AllocateTriangle( nrows )) return Analysis::ERR;
if (((DataSet_2D*)ab_out_)->AllocateTriangle( nrows )) return Analysis::ERR;
if (((DataSet_2D*)ba_out_)->AllocateTriangle( nrows )) return Analysis::ERR;
} else if (outType_ == FULL_MATRIX) {
if (((DataSet_2D*)out_)->Allocate2D( nrows,ncols )) return Analysis::ERR;
if (((DataSet_2D*)ab_out_)->Allocate2D( nrows,ncols )) return Analysis::ERR;
if (((DataSet_2D*)ba_out_)->Allocate2D( nrows,ncols )) return Analysis::ERR;
}
if (out_->Size() != inputSets_.size()) {
mprinterr("Warning: Number of input data sets (%zu) != number of expected"
" sets in matrix (%zu)\n", inputSets_.size(), out_->Size());
return Analysis::ERR;
}
// Directed sets
}
if (df != 0)
df->AddDataSet( out_ );
if (dfab != 0)
df->AddDataSet( ab_out_ );
if (dfba != 0)
df->AddDataSet( ba_out_ );
mprintf(" HAUSDORFF:\n");
mprintf("\tCalculating Hausdorff distances from the following 2D distance matrices:\n\t ");
for (DataSetList::const_iterator it = inputSets_.begin(); it != inputSets_.end(); ++it)
mprintf(" %s", (*it)->legend());
mprintf("\n");
if (outType_ == BASIC)
mprintf("\tOutput will be stored in 1D array set '%s'\n", out_->legend());
else if (outType_ == UPPER_TRI_MATRIX)
mprintf("\tOutput will be stored in upper-triangular matrix set '%s' with %i rows.\n",
out_->legend(), nrows);
else if (outType_ == FULL_MATRIX)
mprintf("\tOutput will be stored in matrix set '%s' with %i rows, %i columns.\n",
out_->legend(), nrows, ncols);
mprintf("\tDirected A->B distance output set: %s\n", ab_out_->legend());
mprintf("\tDirected B->A distance output set: %s\n", ba_out_->legend());
if (df != 0) mprintf("\tOutput set written to '%s'\n", df->DataFilename().full());
if (dfab != 0) mprintf("\tA->B output set written to '%s'\n", dfab->DataFilename().full());
if (dfba != 0) mprintf("\tB->A output set written to '%s'\n", dfba->DataFilename().full());
return Analysis::OK;
}
// Exec_DataSetCmd::Remove()
Exec::RetType Exec_DataSetCmd::Remove(CpptrajState& State, ArgList& argIn) {
std::string status;
// Get criterion type
CriterionType criterion = UNKNOWN_C;
for (int i = 1; i < (int)N_C; i++)
if (argIn.hasKey( CriterionKeys[i] )) {
criterion = (CriterionType)i;
status.assign( CriterionKeys[i] );
break;
}
if (criterion == UNKNOWN_C) {
mprinterr("Error: No criterion specified for 'remove'.\n");
return CpptrajState::ERR;
}
// Get select type
SelectType select = UNKNOWN_S;
std::string val1, val2;
for (const SelectPairType* ptr = SelectKeys; ptr->key_ != 0; ptr++)
if (argIn.Contains( ptr->key_ )) {
select = ptr->type_;
val1 = argIn.GetStringKey( ptr->key_ );
status.append( " " + std::string(ptr->key_) + " " + val1 );
// Get 'and' value for between/outside. TODO put nargs in SelectPairType?
if (select == BETWEEN || select == OUTSIDE) {
val2 = argIn.GetStringKey("and");
if (val2.empty()) {
mprinterr("Error: Missing 'and' value for selection '%s'\n", ptr->key_);
return CpptrajState::ERR;
}
status.append(" and " + val2);
}
break;
}
if (select == UNKNOWN_S || val1.empty()) {
mprinterr("Error: No selection specified for 'remove'.\n");
return CpptrajState::ERR;
}
if ( (criterion == SMODE || criterion == STYPE) &&
(select != EQUAL && select != NOT_EQUAL) )
{
mprinterr("Error: Specified select not valid for criterion '%s'\n", CriterionKeys[criterion]);
return CpptrajState::ERR;
}
mprintf("\tRemoving data sets");
std::string setSelectArg = argIn.GetStringNext();
if (setSelectArg.empty())
setSelectArg.assign("*");
else
mprintf(" within selection '%s'", setSelectArg.c_str());
mprintf(" %s\n", status.c_str());
DataSetList tempDSL = State.DSL().GetMultipleSets( setSelectArg );
if (tempDSL.empty()) {
mprinterr("Error: No data sets selected.\n");
return CpptrajState::ERR;
}
// Remove sets
unsigned int Nremoved = 0;
if ( criterion == AVERAGE ) {
if (!validDouble( val1 )) {
mprinterr("Error: '%s' is not a valid number\n", val1.c_str());
return CpptrajState::ERR;
}
double d_val1 = convertToDouble( val1 );
double d_val2 = d_val1;
if (!val2.empty()) {
if (!validDouble( val2 )) {
mprinterr("Error: '%s' is not a valid number\n", val2.c_str());
return CpptrajState::ERR;
}
d_val2 = convertToDouble( val2 );
}
for (DataSetList::const_iterator ds = tempDSL.begin(); ds != tempDSL.end(); ++ds)
{
if ( (*ds)->Group() != DataSet::SCALAR_1D )
mprintf("Warning: '%s' is not a valid data set for 'average' criterion.\n", (*ds)->legend());
else {
DataSet_1D const& ds1 = static_cast<DataSet_1D const&>( *(*ds) );
double avg = ds1.Avg();
bool remove = false;
switch (select) {
case EQUAL : remove = (avg == d_val1); break;
case NOT_EQUAL : remove = (avg != d_val1); break;
case LESS_THAN : remove = (avg < d_val1); break;
case GREATER_THAN : remove = (avg > d_val1); break;
case BETWEEN : remove = (avg > d_val1 && avg < d_val2); break;
case OUTSIDE : remove = (avg < d_val1 || avg > d_val2); break;
case UNKNOWN_S:
case N_S : return CpptrajState::ERR; // Sanity check
}
if (remove) {
mprintf("\t Removing set '%s' (avg is %g)\n", (*ds)->legend(), avg);
State.RemoveDataSet( *ds );
++Nremoved;
}
}
}
} else if ( criterion == SIZE ) {
if (!validInteger( val1 )) {
mprinterr("Error: '%s' is not a valid number\n", val1.c_str());
return CpptrajState::ERR;
}
unsigned int i_val1 = (unsigned int)convertToInteger( val1 );
unsigned int i_val2 = i_val1;
if (!val2.empty()) {
if (!validInteger( val2 )) {
mprinterr("Error: '%s' is not a valid number\n", val2.c_str());
return CpptrajState::ERR;
}
i_val2 = convertToInteger( val2 );
}
for (DataSetList::const_iterator ds = tempDSL.begin(); ds != tempDSL.end(); ++ds)
{
unsigned int size = (*ds)->Size();
bool remove = false;
switch ( select ) {
case EQUAL : remove = (size == i_val1); break;
case NOT_EQUAL : remove = (size != i_val1); break;
case LESS_THAN : remove = (size < i_val1); break;
case GREATER_THAN : remove = (size > i_val1); break;
case BETWEEN : remove = (size > i_val1 && size < i_val2); break;
case OUTSIDE : remove = (size < i_val1 || size > i_val2); break;
case UNKNOWN_S:
case N_S : return CpptrajState::ERR; // Sanity check
}
if (remove) {
mprintf("\t Removing set '%s' (size is %u)\n", (*ds)->legend(), size);
State.RemoveDataSet( *ds );
++Nremoved;
}
}
} else if ( criterion == SMODE ) {
MetaData::scalarMode mode_val = MetaData::ModeFromKeyword( val1 );
if (mode_val == MetaData::UNKNOWN_MODE) {
mprinterr("Error: '%s' is not a valid mode.\n", val1.c_str());
return CpptrajState::ERR;
}
for (DataSetList::const_iterator ds = tempDSL.begin(); ds != tempDSL.end(); ++ds)
{
bool remove = false;
MetaData::scalarMode mode = (*ds)->Meta().ScalarMode();
if (select == EQUAL ) remove = ( mode == mode_val );
else if (select == NOT_EQUAL) remove = ( mode != mode_val );
else return CpptrajState::ERR; // Sanity check
if (remove) {
mprintf("\t Removing set '%s' (mode is '%s')\n", (*ds)->legend(), MetaData::ModeString(mode));
State.RemoveDataSet( *ds );
++Nremoved;
}
}
} else if ( criterion == STYPE ) {
MetaData::scalarType type_val = MetaData::TypeFromKeyword( val1, MetaData::UNKNOWN_MODE );
if (type_val == MetaData::UNDEFINED) {
mprinterr("Error: '%s' is not a valid type.\n", val1.c_str());
return CpptrajState::ERR;
}
for (DataSetList::const_iterator ds = tempDSL.begin(); ds != tempDSL.end(); ++ds)
{
bool remove = false;
MetaData::scalarType type = (*ds)->Meta().ScalarType();
if (select == EQUAL ) remove = ( type == type_val );
else if (select == NOT_EQUAL) remove = ( type != type_val );
else return CpptrajState::ERR; // Sanity check
if (remove) {
mprintf("\t Removing set '%s' (typeis '%s')\n", (*ds)->legend(), MetaData::TypeString(type));
State.RemoveDataSet( *ds );
++Nremoved;
}
}
} else {
mprinterr("Internal Error: Criterion not yet implemented.\n");
return CpptrajState::ERR;
}
mprintf("\tRemoved %u of %zu sets.\n", Nremoved, tempDSL.size());
return CpptrajState::OK;
}
/** Syntax: dataset invert <set arg0> ... name <new name> */
Exec::RetType Exec_DataSetCmd::InvertSets(CpptrajState& State, ArgList& argIn) {
DataSetList& DSL = State.DSL();
// Get keywords
DataSet* inputNames = 0;
std::string dsname = argIn.GetStringKey("legendset");
if (!dsname.empty()) {
inputNames = DSL.GetDataSet( dsname );
if (inputNames == 0) {
mprinterr("Error: Name set '%s' not found.\n", dsname.c_str());
return CpptrajState::ERR;
}
if (inputNames->Type() != DataSet::STRING) {
mprinterr("Error: Set '%s' does not contain strings.\n", inputNames->legend());
return CpptrajState::ERR;
}
mprintf("\tUsing names from set '%s' as legends for inverted sets.\n", inputNames->legend());
}
dsname = argIn.GetStringKey("name");
if (dsname.empty()) {
mprinterr("Error: 'invert' requires that 'name <new set name>' be specified.\n");
return CpptrajState::ERR;
}
mprintf("\tNew sets will be named '%s'\n", dsname.c_str());
DataFile* outfile = State.DFL().AddDataFile( argIn.GetStringKey("out"), argIn );
if (outfile != 0)
mprintf("\tNew sets will be output to '%s'\n", outfile->DataFilename().full());
// TODO determine type some other way
DataSet::DataType outtype = DataSet::DOUBLE;
// Get input DataSets
std::vector<DataSet_1D*> input_sets;
std::string dsarg = argIn.GetStringNext();
while (!dsarg.empty()) {
DataSetList sets = DSL.GetMultipleSets( dsarg );
for (DataSetList::const_iterator ds = sets.begin(); ds != sets.end(); ++ds)
{
if ( (*ds)->Group() != DataSet::SCALAR_1D ) {
mprintf("Warning: '%s': Inversion only supported for 1D scalar data sets.\n",
(*ds)->legend());
} else {
if (!input_sets.empty()) {
if ( (*ds)->Size() != input_sets.back()->Size() ) {
mprinterr("Error: Set '%s' has different size (%zu) than previous set (%zu)\n",
(*ds)->legend(), (*ds)->Size(), input_sets.back()->Size());
return CpptrajState::ERR;
}
}
input_sets.push_back( (DataSet_1D*)*ds );
}
}
dsarg = argIn.GetStringNext();
}
if (input_sets.empty()) {
mprinterr("Error: No sets selected.\n");
return CpptrajState::ERR;
}
if (inputNames != 0 && inputNames->Size() != input_sets.front()->Size()) {
mprinterr("Error: Name set '%s' size (%zu) differs from # data points (%zu).\n",
inputNames->legend(), inputNames->Size(), input_sets.front()->Size());
return CpptrajState::ERR;
}
mprintf("\t%zu input sets; creating %zu output sets.\n",
input_sets.size(), input_sets.front()->Size());
// Need an output data set for each point in input sets
std::vector<DataSet*> output_sets;
int column = 1;
for (int idx = 0; idx != (int)input_sets[0]->Size(); idx++, column++) {
DataSet* ds = 0;
ds = DSL.AddSet(outtype, MetaData(dsname, column));
if (ds == 0) return CpptrajState::ERR;
if (inputNames != 0)
ds->SetLegend( (*((DataSet_string*)inputNames))[idx] );
output_sets.push_back( ds );
if (outfile != 0) outfile->AddDataSet( ds );
}
// Create a data set containing names of each input data set
DataSet* nameset = DSL.AddSet(DataSet::STRING, MetaData(dsname, column));
if (nameset == 0) return CpptrajState::ERR;
if (inputNames != 0)
nameset->SetLegend("Names");
if (outfile != 0) outfile->AddDataSet( nameset );
// Populate output data sets
for (int jdx = 0; jdx != (int)input_sets.size(); jdx++)
{
DataSet_1D const& INP = static_cast<DataSet_1D const&>( *(input_sets[jdx]) );
nameset->Add( jdx, INP.legend() );
for (unsigned int idx = 0; idx != INP.Size(); idx++)
{
double dval = INP.Dval( idx );
output_sets[idx]->Add( jdx, &dval );
}
}
return CpptrajState::OK;
}
// Exec_DataSetCmd::ChangeModeType()
Exec::RetType Exec_DataSetCmd::ChangeModeType(CpptrajState const& State, ArgList& argIn) {
std::string modeKey = argIn.GetStringKey("mode");
std::string typeKey = argIn.GetStringKey("type");
if (modeKey.empty() && typeKey.empty()) {
mprinterr("Error: No valid keywords specified.\n");
return CpptrajState::ERR;
}
// First determine mode if specified.
MetaData::scalarMode dmode = MetaData::UNKNOWN_MODE;
if (!modeKey.empty()) {
dmode = MetaData::ModeFromKeyword( modeKey );
if (dmode == MetaData::UNKNOWN_MODE) {
mprinterr("Error: Invalid mode keyword '%s'\n", modeKey.c_str());
return CpptrajState::ERR;
}
}
// Next determine type if specified.
MetaData::scalarType dtype = MetaData::UNDEFINED;
if (!typeKey.empty()) {
dtype = MetaData::TypeFromKeyword( typeKey, dmode );
if (dtype == MetaData::UNDEFINED) {
mprinterr("Error: Invalid type keyword '%s'\n", typeKey.c_str());
return CpptrajState::ERR;
}
}
// Additional options for type 'noe'
AssociatedData_NOE noeData;
if (dtype == MetaData::NOE) {
if (noeData.NOE_Args(argIn))
return CpptrajState::ERR;
}
if (dmode != MetaData::UNKNOWN_MODE)
mprintf("\tDataSet mode = %s\n", MetaData::ModeString(dmode));
if (dtype != MetaData::UNDEFINED)
mprintf("\tDataSet type = %s\n", MetaData::TypeString(dtype));
// Loop over all DataSet arguments
std::string ds_arg = argIn.GetStringNext();
while (!ds_arg.empty()) {
DataSetList dsl = State.DSL().GetMultipleSets( ds_arg );
for (DataSetList::const_iterator ds = dsl.begin(); ds != dsl.end(); ++ds)
{
if (dmode != MetaData::UNKNOWN_MODE) {
// Warn if mode does not seem appropriate for the data set type.
if ( dmode >= MetaData::M_DISTANCE &&
dmode <= MetaData::M_RMS &&
(*ds)->Group() != DataSet::SCALAR_1D )
mprintf("Warning: '%s': Expected scalar 1D data set type for mode '%s'\n",
(*ds)->legend(), MetaData::ModeString(dmode));
else if ( dmode == MetaData::M_VECTOR &&
(*ds)->Type() != DataSet::VECTOR )
mprintf("Warning: '%s': Expected vector data set type for mode '%s'\n",
(*ds)->legend(), MetaData::ModeString(dmode));
else if ( dmode == MetaData::M_MATRIX &&
(*ds)->Group() != DataSet::MATRIX_2D )
mprintf("Warning: '%s': Expected 2D matrix data set type for mode '%s'\n",
(*ds)->legend(), MetaData::ModeString(dmode));
}
if ( dtype == MetaData::NOE ) (*ds)->AssociateData( &noeData );
mprintf("\t\t'%s'\n", (*ds)->legend());
MetaData md = (*ds)->Meta();
md.SetScalarMode( dmode );
md.SetScalarType( dtype );
(*ds)->SetMeta( md );
}
ds_arg = argIn.GetStringNext();
}
return CpptrajState::OK;
}
// Exec_DataSetCmd::ChangeDim()
Exec::RetType Exec_DataSetCmd::ChangeDim(CpptrajState const& State, ArgList& argIn) {
int ndim = -1;
if (argIn.hasKey("xdim"))
ndim = 0;
else if (argIn.hasKey("ydim"))
ndim = 1;
else if (argIn.hasKey("zdim"))
ndim = 2;
else
ndim = argIn.getKeyInt("ndim", -1);
if (ndim < 0) {
mprinterr("Error: Specify xdim/ydim/zdim or dimension number with ndim.\n");
return CpptrajState::ERR;
}
if (ndim < 3) {
static const char DIMSTR[3] = { 'X', 'Y', 'Z' };
mprintf("\tChanging the following in the %c dimension:\n", DIMSTR[ndim]);
} else
mprintf("\tChanging the following in dimension %i\n", ndim);
bool changeLabel, changeMin, changeStep;
std::string label;
double min = 0.0;
double step = 0.0;
if (argIn.Contains("label")) {
label = argIn.GetStringKey("label");
changeLabel = true;
mprintf("\tNew Label: %s\n", label.c_str());
} else
changeLabel = false;
if (argIn.Contains("step")) {
step = argIn.getKeyDouble("step", 0.0);
changeStep = true;
mprintf("\tNew step: %g\n", step);
} else
changeStep = false;
if (argIn.Contains("min")) {
min = argIn.getKeyDouble("min", 0.0);
changeMin = true;
mprintf("\tNew min: %g\n", min);
} else
changeMin = false;
// Loop over all DataSet arguments
std::string ds_arg = argIn.GetStringNext();
while (!ds_arg.empty()) {
DataSetList dsl = State.DSL().GetMultipleSets( ds_arg );
for (DataSetList::const_iterator ds = dsl.begin(); ds != dsl.end(); ++ds)
{
if (ndim < (int)(*ds)->Ndim()) {
mprintf("\t%s\n", (*ds)->legend());
Dimension dim = (*ds)->Dim(ndim);
if (changeLabel) dim.SetLabel( label );
if (changeMin) dim.ChangeMin( min );
if (changeStep) dim.ChangeStep( step );
(*ds)->SetDim(ndim, dim);
} else
mprintf("Warning: Set '%s' has fewer then %i dimensions - skipping.\n",
(*ds)->legend(), ndim);
}
ds_arg = argIn.GetStringNext();
}
return CpptrajState::OK;
}
