math

math copied to clipboard

Errors when right-censoring: bug in normal_lccdf?

6

I've been running into issues fitting models with a right-censored normal distribution, using `brms`. I've submitted an issue with `brms` here: https://github.com/paul-buerkner/brms/issues/1423 However, it appears that the error does not...

simoncolumbus

## Description Calculation of the cholesky factor is very expensive, which is why updating a given Cholesky factor can save substantial amount of effort. It would this be great to...

wds15

ODE Methods not `fvar<T>` compatible

4

## Description The ODE functors aren't compatible with `fvar` inputs and fail to compile #### Current Version: v4.4.0

andrjohns

## Description Functions to calculate convolution in 1D and 2D (circular and linear) using the FFT function in Stan. For instance, `conv1(a,b)` and `conv2(a,b)`. A general description of the algorithm...

nikunj410

weibull_lcdf might not have been implemented in the best way

2

## Description I was using `weibull_lcdf` in a very complex model and had problems starting: ``` Chain 1 Stan can't start sampling from this initial value. Chain 1 Rejecting initial...

bnicenboim

Use finite-differences where forward-mode not implementable

2

## Description As discussed over in #2839, the ODE methods are not `fvar` compatible, and it would be a significant amount of work to do so. This means that any...

andrjohns

Cleanup accumulator?

5

## Description It's an extra cost, but when we use [`accumulator.add(Vector)`](https://github.com/stan-dev/math/blob/develop/stan/math/prim/fun/accumulator.hpp#L52) method in a Stan model should the sum we do there use `log_sum_exp()` since we are summing a bunch...

SteveBronder

log1p_exp() for logistic_lcdf() and logistic_lccdf()

2

## Description The logistic_lcdf() and logistic_lccdf() function break down at standardised values beyond 36 or so, due to numerical underflow (e.g., [here](https://github.com/stan-dev/math/blob/92075708b1d1796eb82e3b284cd11e544433518e/stan/math/prim/prob/logistic_lcdf.hpp#L76) and [here](https://github.com/stan-dev/math/blob/92075708b1d1796eb82e3b284cd11e544433518e/stan/math/prim/prob/logistic_lccdf.hpp#L76)). This underflow can be solved by...

jaburgoyne

Add a signed log_sum_exp

3

## Description When using `log_sum_exp`, it would be useful to be able specify a 'sign' for the variables being aggregated - such that some inputs are added and some subtracted....

andrjohns

Allow Hessian functors to return Hessian as compressed sparse matrix

6

## Summary The Hessian functors currently return a full symmetric matrix for the Hessian, using double the memory needed. This PR updates the Hessian functors to return only the lower-triangular...

andrjohns