Member mx::AO::analysis::fourierTemporalPSD< _realT, aosysT >::analyzePSDGrid (const std::string &subDir, const std::string &psdDir, int mnMax, int mnCon, realT gfixed, int lpNc, realT lpRegPrecision, std::vector< realT > &mags, int lifetimeTrials=0, bool uncontrolledLifetimes=false, bool writePSDs=false, bool writeXfer=false)

need upstream NCP and CP NCP and NCP NCP

Member mx::AO::analysis::fourierTemporalPSD< _realT, aosysT >::intensityPSD (const std::string &subDir, const std::string &psdDir, const std::string &CvdPath, int mnMax, int mnCon, const std::string &si_or_lp, std::vector< realT > &mags, int lifetimeTrials, bool writePSDs)

should be able to scale the covar by r0, and possibly D

Member mx::AO::analysis::fourierTemporalPSD< _realT, aosysT >::multiLayerPSD (std::vector< realT > &PSD, std::vector< realT > &freq, realT m, realT n, int p, realT fmax=0)

implement error checking

handle reports of convergence failures form singleLayerPSD when implemented.

Member mx::AO::analysis::fourierTemporalPSD< _realT, aosysT >::singleLayerPSD (std::vector< realT > &PSD, std::vector< realT > &freq, realT m, realT n, int layer_i, int p, realT fmax=0)

implement error checking.

need a way to track convergence failures in integral without throwing an error.

need better handling of averaging for the -17/3 extension.

Member mx::AO::analysis::speckleAmpPSD (std::vector< realT > &spFreq, std::vector< realT > &spPSD, const std::vector< realT > &freq, const std::vector< realT > &fmPSD, const std::vector< std::complex< realT > > &fmXferFxn, const std::vector< realT > &nPSD, const std::vector< std::complex< realT > > &nXferFxn, int N, std::vector< realT > *vars=nullptr, std::vector< realT > *bins=nullptr, bool noPSD=false)

Figure out what's going on with psdFilter normalization!

probably need to not do overlapped averaging in periodogram. Could drop mean-sub then.

Member mx::AO::analysis::vonKarmanSpectrum< realT >::dumpPSD (iosT &ios)

generatlize for different alpha and beta

Member mx::AO::analysis::vonKarmanSpectrum< realT >::operator() (psdParamsT &par, size_t layer_i, realT k, realT sec_zeta)

this needs to handle layers with different L_0

Member mx::AO::analysis::zernikeCovariance< realT, aosysT >::getCovariance (realT &error)

document me

handle gsl errors

Struct mx::AO::analysis::zernikeTemporalPSD< _realT, aosysT >

Split off the integration parameters in a separate structure.

once integration parameters are in a separate structure, make this a class with protected members.

GSL error handling

Member mx::AO::analysis::zernikeTemporalPSD< _realT, aosysT >::multiLayerPSD (std::vector< realT > &PSD, std::vector< realT > &freq, int zern_j, realT fmax=0)

implement error checking

handle reports of convergence failures form singleLayerPSD when implemented.

Member mx::AO::analysis::zernikeTemporalPSD< _realT, aosysT >::singleLayerPSD (std::vector< realT > &PSD, std::vector< realT > &freq, int zern_j, int layer_i, realT fmax=0)

implement error checking.

need a way to track convergence failures in integral without throwing an error.

need better handling of averaging for the -17/3 extension.

Class mx::AO::sim::generalIntegrator< _realT >

document the math for the G.I.

document the filter coefficients, etc.

Struct mx::AO::sim::turbAtmosphere< _aoSystemT >

add facility for switching interpolators

need to include wrap detection when using subharmonics (e.g. frames needs to be right)

Member mx::AO::sim::turbAtmosphere< _aoSystemT >::m_forceGen

need check for changes in parameters to decide if saved frames are valid per layer

Struct mx::AO::sim::turbLayer< _aoSystemT >

document this

add facility for changing interpolator

Struct mx::app::appConfigurator

add handling of += in subsequent files.

should just swith to strict TOML

Member mx::app::appConfigurator::readConfig (const std::string &fname, bool reportFileNotFound=true)

update error handling to include >0 (line numer of parse error) and -2 memory allocation error.

handle += here, by appending to the last value as if a vector.

Member mx::app::iniFile::insert (const char *section, const char *name, const char *value)

need error checking.

Member mx::astro::baseSpectrum< realT >::charFlux (realT &flambda0, realT &fnu0, realT &fphot0, const realT &lambda_0, const std::vector< realT > &lambda, const filterT &trans)

use unit conversions to make it work for everything.

check on integration method, should it be trap?

Member mx::astro::baseSpectrum< realT >::charFlux (realT &flambda0, realT &fnu0, realT &fphot0, const realT &lambda_0, const std::vector< realT > &lambda, const std::vector< realT > &trans)

use unit conversions to make it work for everything.

check on integration method, should it be trap?

Member mx::astro::getGMST (realT &GMST, int Yr, int Mo, int Dy, int hr, int min, realT sec)

handle SOFA error.

Struct mx::cuda::cublasHandle

throw exceptions in cuda::cublasHandle

Struct mx::cuda::cusolverDnHandle

throw exceptions in cuda::cusolverDNHandle

Struct mx::cuda::cusolverDnParams

throw exceptions in cuda::cusolverDnParams

Member mx::fits::fitsFile< dataT >::read (arrT &data)

this can probably be made part of one read function (or call read(data *)) with a call to resize with SFINAE

Member mx::fits::fitsFile< dataT >::read (dataT *data)

test if there is a speed difference for full reads for fits_read_pix/subset

Member mx::improc::ADIobservation< _realT, _derotFunctObj >::injectFake (eigenCube< realT > &ims, std::vector< std::string > &fileList, derotFunctObj &derotF, realT RDIfluxScale, realT RDISepScale)

should pad the fake before calling the single image version

throw exceptions for all errors, and switch to void

Member mx::improc::ADIobservation< _realT, _derotFunctObj >::injectFake (eigenImageT &fakePSF, eigenCube< realT > &ims, int image_i, realT derotAngle, realT PA, realT sep, realT contrast, realT scale, realT RDIfluxScale, realT RDISepScale)

should pad the fake before this point

throw exceptions for all errors, and switch to void

Member mx::improc::cubicConvolTransform< _arithT >::operator() (arrT &kern, arithT x, arithT y)

why is this arithT not just the class's arithT?

Member mx::improc::HCIobservation< _realT >::coaddImages (int coaddCombineMethod, int coaddMaxImno, int coaddMaxTime, std::vector< std::string > &coaddKeywords, std::vector< double > &imageMJD, std::vector< fits::fitsHeader > &heads, eigenCube< realT > &ims)

this should really include end of exposure too.

this needs to handle averaging of angles

Member mx::improc::HCIobservation< _realT >::m_comboWeights

check how comboWeights are handled in coadding

Member mx::improc::HCIobservation< _realT >::outputPreProcessed ()

Should add a HISTORY card here

Member mx::improc::HCIobservation< _realT >::readMask ()

here re-size mask if needed to match imSize

Member mx::improc::imageDownSample (imageOutT &imout, const imageInT &imin)

Allow selection of interpolator, providing a default version.

Member mx::improc::imCenterCircleSym< realT >::center (realT &x, realT &y, eigenImage< realT > &im, realT x0, realT y0)

This needs to be able to handle highly non-symmetric peaks much better.

Member mx::improc::imCenterRadon< transformT >::fitGrid (realT &x, realT &y, eigenImage< realT > grid)

This needs to be able to handle highly non-symmetric peaks much better.

Member mx::improc::milkImage< _dataT >::post ()

need to set wtime, have a version with atime

Member mx::improc::milkImage< _dataT >::valid ()

check inode!

Member mx::ioutils::getSequentialFilename (const std::string &basename, const std::string &extension="", const int startat=0, int ndigit=4)

switch to using a regex or something so we can detect the missing file.

Member mx::ioutils::readColumns (const std::string &fname, arrTs &...arrays)

lineSize should be configurable

Struct mx::ioutils::textTable

have ability to add whole row, with a variadic template to handle different types

allow non-wrapping (i.e. unlimited width) columns

headers and footers

add centering (h and v) and right and bottom justification

have a latex() setup function

add sort capability, with vector of column choices to sort first by column N, then M, etc.

Class mx::math::ft::fftT< _inputT, _outputT, _rank, 0 >

add execute interface with fftw like signature

add plan interface where user passes in pointers (to avoid allocations)

Member mx::math::func::root_epsilon ()

this isn't, but should be constexpr – need constr sqrt.

Class mx::math::gnuPlot

Use mxError for error reporting.

Member mx::math::gnuPlot::checkResponse (std::string &response, double timeout=0)

investigate having a second thread monitor for responses.

Member mx::math::gsl_interpolate (const gsl_interp_type *interpT, realT *xin, realT *yin, size_t Nin, realT *xout, realT *yout, size_t Nout)

report errors iaw mxlib standard in gsl_interpolate

Member mx::math::radprofIntegral (vectorScT r, vectorT p, bool inczero=false)

the trapezoid rule may not be strictly valid as implemented. Is there a weighting to apply based on r?

Struct mx::math::syevrMem< floatT >

this should have the working memory for the first exploratory call to ?syevr as well.

Member mx::sigproc::autocorrelation (std::vector< T > &ac, std::vector< T > &sig)

this should probably re-allocate if ac.size() != sig.size(), or at least <

Struct mx::sigproc::linearPredictor< _realT >

document linearPredictor

Member mx::sigproc::makeFourierModeFreqs_Circ (std::vector< fourierModeDef > &spf, int N)

use push_back instead of resize

Member mx::sigproc::makeFourierModeFreqs_PandV (std::vector< fourierModeDef > &spf, int N)

use push_back instead of resize

Class mx::sigproc::psdFilter< _realT, 1 >

once fftT has a plan interface with pointers for working memory, use it.

Class mx::sigproc::psdFilter< _realT, _rank, 0 >

once fftT has a plan interface with pointers for working memory, use it.

Struct mx::sigproc::psdVarMean< paramsT >

document the paramsT interface

Member mx::sigproc::psdVarMean< paramsT >::init ()

We'll get the occasional failure to reach tolerance error, just ignore them all for now.

Member mx::sigproc::psdVarMeanParams< _realT >::psdVal (realT k)

implement better extrapolation past max-freq.

Member mx::sigproc::zernikeQ (realT k, realT phi, int n, int m)

need a more robust jinc_n function for n > 1

Member mx::sigproc::zernikeQNorm (realT k, realT phi, int n, int m)

need a more robust jinc_n function for n > 1

Member mx::wfp::drawLine (arrayT &im, typename arrayT::Scalar x0, typename arrayT::Scalar y0, typename arrayT::Scalar x1, typename arrayT::Scalar y1, typename arrayT::Scalar halfWidth)

should handle width much more intelligently, this only works for ~45 degree lines.

Class mx::wfp::fraunhoferPropagator< _wavefrontT >

check if we should reverse the FFT orders to be proper

File zernike.hpp

the basic zernike polys should be in math::func.