lijun99
commented
2 years ago This error happens when AlTar could not produce a proper proposal distribution from the statistics of current samples. The cause of poor sampling or low acceptance ratio varies, from too few samples, too big jump distance, choices of prior, too small data covariance, to ill-imposed inverse problems for AlTar's MCMC sampler. You may need to examine the cause by yourself. Something you could try is to adjust the scaling parameters in samplers or to use another sampler, https://altar.readthedocs.io/en/cuda/cuda/AlTarFramework.html#metropolis-cuda-version.
Dear author, $Chelosky factorization error: the leading minor of order 117 is not positive definite I try to use the slipmodel to get the paraments value. My Green function array is 3841040 and observations data array is 3841. It will exit after running 6 iteration, and print the below error.