-
Notifications
You must be signed in to change notification settings - Fork 607
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat(osmomath): Exp2 function #3708
Conversation
0cc8250
to
a548068
Compare
I think as long as the error is within the an acceptable range (from the chart |
1cd4928
to
79beaae
Compare
There are some cases where the error is outside of the chart - Line 112 in 79beaae
I have not been able to explain why that is the case. One guess is that approximated results in Go are being compared against WolframAlpha. The chart is constructed from a Python library where both the approximations and the actual results are using the same framework. I'm wondering if different computational platforms introduce additional errors. Please let me know your thoughts. |
|
||
func TestExp2ChebyshevRationalApprox(t *testing.T) { | ||
smallValue := osmomath.MustNewDecFromStr("0.00001") | ||
smallerValue := osmomath.MustNewDecFromStr("0.00000000000000000001") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What is this smaller than & what is relation to minDecTolerance (sorry not sure if theres some important relation)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I used these to implement test cases closer to the boundaries.
The reason is that the errors tend to be large at the boundaries for non-Chebyshev approximations. This is just to make sure that, as expected, Chebyshev does not allow for larger errors close to boundaries. I will add a comment relaying this
Python and actual go code both LGTM, nice job! Still reviewing go tests |
The plot difference LGTM Were actually seeing what we'd expect to see between the absolute error vs relative error The additive error on RHS side was larger than LHS. The relative error 'dampens' the RHS relative to the LHS, as exp only increasing on the range |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice job!
That makes sense. Thanks for confirming |
All comments are now addressed. This should be ready for merge once CI passes. |
* feat(osmomath): exp2 function * export exp2 * changelog * refactor ErrTolerance to use Dec instead of Int for additive tolerance * Update osmomath/exp2.go * Update osmomath/exp2.go * Update osmomath/exp2.go * Update osmomath/exp2_test.go * Update osmomath/exp2_test.go * do bit shift instead of multiplication * godoc about error bounds * comment about bit shift equivalency * merge conflict * improve godoc * typo * remove TODOs - confirmed obsolete * Runge's phenomenon comment (cherry picked from commit 89bf606) # Conflicts: # CHANGELOG.md
* feat(osmomath): exp2 function * export exp2 * changelog * refactor ErrTolerance to use Dec instead of Int for additive tolerance * Update osmomath/exp2.go * Update osmomath/exp2.go * Update osmomath/exp2.go * Update osmomath/exp2_test.go * Update osmomath/exp2_test.go * do bit shift instead of multiplication * godoc about error bounds * comment about bit shift equivalency * merge conflict * improve godoc * typo * remove TODOs - confirmed obsolete * Runge's phenomenon comment
…14 (#3925) * Upgrade IBC to v4.2.0 (#3838) * initial changes to migrate to ibc v4 * added checksum to proposal * begin and end block are now being called inside nextBlock * added changelog * linked pr on changelog * remove local replace * using error acks from osmoutils * osmoutils tagged * go sum * added checksum * feat(x/twap): modify cli to add geometric option (#3812) * feat(x/twap): geometric twap code gen query boilerplate * revert cli change * query gen * wire up API * test * fix * fixes * cli * lint * refactor via flag * refactor * refactor * fixes * lint * add arithmetic twap alias * Make wasm hooks importable (#3850) * moved ibc-hooks to its own go.mod * updated ibc hooks version * go sum * add ics23 patch into x/ibc-hooks * Fix wasm import version conflict * Update x/ibc-hooks to osmoutils v0.0.2 * Update versions Co-authored-by: Dev Ojha <[email protected]> * refactor(x/twap): handle spot price error case in the context of geometric twap (#3845) * refactor(x/twap): handle spot price error case * supporting test cases * table-driven log tests * test(x/twap): add randomized geometric twap test on a balancer pool (#3844) * test(x/twap): add randomized test with a balancer pool * comments * multiplicative tolerance, fewer retries and larger initial supply range * Basic geometric twap e2e test (#3835) * feat(x/twap): geometric twap code gen query boilerplate * revert cli change * query gen * wire up API * test * fix * fixes * add geometric queries * create pool.json * add test * resolve conflict * fix: swap uosmo in * fix problem with wallet creation * updates * simplify and add comments * Update tests/e2e/e2e_test.go * Update tests/e2e/e2e_test.go * Update tests/e2e/configurer/chain/queries.go Co-authored-by: Adam Tucker <[email protected]> * Update tests/e2e/configurer/chain/queries.go Co-authored-by: Adam Tucker <[email protected]> * Update tests/e2e/e2e_test.go Co-authored-by: Adam Tucker <[email protected]> * Update tests/e2e/e2e_test.go Co-authored-by: Adam Tucker <[email protected]> * Update tests/e2e/e2e_test.go Co-authored-by: Adam Tucker <[email protected]> * Update tests/e2e/e2e_test.go Co-authored-by: Adam Tucker <[email protected]> Co-authored-by: Roman <[email protected]> Co-authored-by: Roman <[email protected]> Co-authored-by: Adam Tucker <[email protected]> * feat(x/twap): whitelist GeometricTwap and GeometricTwapToNow (#3852) * feat(x/twap): GeometricTwap and GeometricTwapToNow queries added to Stargate whitelist * update docs * fix(scripts): proto gen for osmoutils (#3854) * fix(scripts): proto gen for osmoutils * Update scripts/protocgen.sh * fix(scripts): proto gen osmoutils path (#3859) * added packet timeouts to wasm hooks (#3862) * add negative to cli (#3888) * Making osmoutils compile on chains that don't use our SDK fork (#3899) * making osmoutils compile on chains that don't use osmosis' fork of the cosmos sdk * updated imports for e2e tests * go fumpt * updated version everywhere * added changelog entry * remove deprecation from arithmetic & geometric twap to now query (#3917) * Add types & boilerplate for the Downtime detector module (#3609) Sub-component of #3603 Adds types for the thin module intended for downtime detection - Add downtime detection module types No tests added - Does this pull request introduce a new feature or user-facing behavior changes? somewhat - Is a relevant changelog entry added to the `Unreleased` section in `CHANGELOG.md`? yes - How is the feature or change documented? In its spec * Add downtime detector module (#3688) * WIP * Switch to enum * Remove params query * Add query * Add wiring, add import/export test * Start begin block test * Finish keeper tests * Add CLI * Wire downtime detector CLI + queries * more module wiring * add types test * Fix state alteration test * Fix superfluid test * Add stargate whitelist support * Apply code comment * Rename folder * Add requested godoc * Update x/downtime-detector/genesis.go Co-authored-by: Adam Tucker <[email protected]> * Apply adam suggestion for having a `-` * move genesis test to own file Co-authored-by: Adam Tucker <[email protected]> * Initial by hand fixes * feat(osmomath): Exp2 function (#3708) * feat(osmomath): exp2 function * export exp2 * changelog * refactor ErrTolerance to use Dec instead of Int for additive tolerance * Update osmomath/exp2.go * Update osmomath/exp2.go * Update osmomath/exp2.go * Update osmomath/exp2_test.go * Update osmomath/exp2_test.go * do bit shift instead of multiplication * godoc about error bounds * comment about bit shift equivalency * merge conflict * improve godoc * typo * remove TODOs - confirmed obsolete * Runge's phenomenon comment * [x/TWAP] Expose a geometric TWAP API (#3529) * refactored twap api.go for geometric TWAP * added barebon docs * romans feedback * new * fix * nichola feedback * final roman comments * fix twap by hand * change to gamm * fix balancer test * bump to v14 upgrade * e2e fix * add remaining diff from main to ibc-rate-limit * update contracts test * osmomath: `AddMut` and `QuoMut` (#3779) * mut add * test add mut * quo mut * test quo mut/ remove want from test struct * refactor exp * change mutatives code * change * not allocaing * exp change to quomut * remove file * refactor quo * refactor ad * refactor tests * Modify CHANGELOG * Whitelist EstimateSwapExactAmountOut (#3693) * whitelist EstimateSwapExactAmountOut * doc: changelog * updated rate limit contract * Fix rust checks (#3576) * added cargo.lock * added Cargo.lock as an artifact * added new bytecode with lock file Co-authored-by: Nicolas Lara <[email protected]> Co-authored-by: Roman <[email protected]> Co-authored-by: Dev Ojha <[email protected]> Co-authored-by: Hieu Vu <[email protected]> Co-authored-by: Roman <[email protected]> Co-authored-by: Supanat <[email protected]> Co-authored-by: Dev Ojha <[email protected]> Co-authored-by: Sishir Giri <[email protected]> Co-authored-by: Ruslan Akhtariev <[email protected]> Co-authored-by: mattverse <[email protected]> Co-authored-by: ByeongSu Hong <[email protected]>
Closes: #XXX
What is the purpose of the change
This PR adds an$$2^{decimal}$$
Exp2
function that computesThe following property is used:
$$2^{decimal} = 2^{integer} * 2^{fractional}$$
The integer exponent result is computed using the existing
PowerInteger
function.The fractional exponent result is computed using the 13-parameter Chebyshev rational approximation. The numerator coefficients are rounded down and the denominator coefficients are rounded up at the precision end. However, this does not guarantee that the approximated result is always smaller than the actual results (as determined by WolframAlpha).
The Python approximation script is modified to utilize the production coefficients for estimating relative errors. Please see the results:
Additionally,
ErrTolerance
was refactored to usesdk.Dec
forAdditiveTolerance
for more accuracy in tests.In a future PR,
Power
function with a decimal exponent can be added. This function will useexp2
by utilizing the following property:x^y = exp2(y * log_2{x})
Testing and Verifying
Tests with links to WolframAlpha were added. The absolute and the relative errors are bounded.
Open Questions
Documentation and Release Note
Unreleased
section inCHANGELOG.md
? yes