Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Binary file added Publications_and_Presentations/IR_Note.pdf
Binary file not shown.
Binary file modified Publications_and_Presentations/MPArray.pdf
Binary file not shown.
6 changes: 6 additions & 0 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -335,6 +335,10 @@ A new paper exploits most of the package. I use the examples in that paper for C
- (KEL23a) C. T. Kelley, [__Newton's method in three precisions__](https://arxiv.org/abs/2307.16051) To appear in Pacific Journal of Optimization.
- This paper has a [repo](https://github.com/ctkelley/N3Presults) for reproducing the results with an early version of this package.

I just submitted a paper about interprecison transfers.

- (Kel24a) C. T. Kelley, [__Interprecision transfers in iterative refinement__](https://github.com/ctkelley/MultiPrecisionArrays.jl/blob/main/Publications_and_Presentations/IR_Note.pdf)


## __Interprecision transfers__:

Expand All @@ -343,6 +347,8 @@ Most users of this package can ignore this issue.

- [Interprecision Transfers in Iterative Refinement: Making Half Precision on Desktops Less Painful](Publications_and_Presentations/MPArrays_XSDK-MULTIPRECISION_June_15.pdf).

I have a new paper on this. See ref (Kel 24a) [__Interprecision transfers in iterative refinement__](https://github.com/ctkelley/MultiPrecisionArrays.jl/blob/main/Publications_and_Presentations/IR_Note.pdf)

## Other IR in Julia

See [the docs](https://ctkelley.github.io/MultiPrecisionArrays.jl/dev/#Other-IR-software-in-Julia) for a couple citations.
Expand Down
2 changes: 1 addition & 1 deletion docs/src/Details/Extended.md
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@ residual in a precision ```TR```
higher than the working precision. If you do this,
you should store both the solution and the residual in precision
TR and to interprecision transfers on the fly. In that case you are really
solving a promoted problem
solving a promoted problem [ctk:irnote](@cite)
```math
(I_W^R A) x = I_W^R b
```
Expand Down
11 changes: 11 additions & 0 deletions docs/src/MPArray.bib
Original file line number Diff line number Diff line change
Expand Up @@ -223,3 +223,14 @@ @misc{accelerate
year=2023,
howpublished="https://github.com/JuliaLinearAlgebra/AppleAccelerate.jl"
}

@misc{ctk:irnote,
author="C. T. Kelley",
title="Interprecision transfers in iterative refinement",
year = 2024,
archivePrefix={arXiv},
primaryClass={math.NA},
eprint={submit/5701252},
note="submitted for publication"
}