Add sum_to_zero constraint, free, and check based on ILR transform by WardBrian · Pull Request #3101 · stan-dev/math · GitHub

Summary

This is an alternative to #3099 based on the discussion in the comments there. Rather than x[1:N-1] = y; x[N] = -sum(y) this uses a modification of the isometric log ratio (ILR) transformation for simplexes. The implementation is based on a single for-loop, but was adapted from Sean's code that used several temporary vectors, which was adapted from a version which used a full matrix

Tests

Added tests in the err and constraint folders, based on the existing tests for simplex.

In addition, I iterated on the implementation of this algorithm in Python, so I have some out-of-tree tests here: https://gist.github.com/WardBrian/aae35c65fb841f923d89cc30a3d70808

Side Effects

Release notes

Added sum_to_zero_constrain, sum_to_zero_free, and check_sum_to_zero.

Checklist

Copyright holder: Simons Foundation

The copyright holder is typically you or your assignee, such as a university or company. By submitting this pull request, the copyright holder is agreeing to the license the submitted work under the following licenses:
- Code: BSD 3-clause (https://opensource.org/licenses/BSD-3-Clause)
- Documentation: CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/)
the basic tests are passing
- unit tests pass (to run, use: ./runTests.py test/unit)
- header checks pass, (make test-headers)
- dependencies checks pass, (make test-math-dependencies)
- docs build, (make doxygen)
- code passes the built in C++ standards checks (make cpplint)
the code is written in idiomatic C++ and changes are documented in the doxygen
the new changes are tested

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.38	0.3	1.26	20.66% faster
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	1.24	19.56% faster
gp_regr/gen_gp_data.stan	0.03	0.03	1.05	4.41% faster
gp_regr/gp_regr.stan	0.1	0.09	1.08	7.63% faster
sir/sir.stan	69.23	68.04	1.02	1.73% faster
irt_2pl/irt_2pl.stan	4.17	4.05	1.03	2.93% faster
eight_schools/eight_schools.stan	0.06	0.06	1.06	5.79% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.25	0.25	1.01	1.21% faster
pkpd/one_comp_mm_elim_abs.stan	19.58	19.48	1.01	0.52% faster
garch/garch.stan	0.43	0.41	1.06	5.59% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.75	2.63	1.04	4.27% faster
arK/arK.stan	1.83	1.74	1.05	4.78% faster
gp_pois_regr/gp_pois_regr.stan	2.92	2.73	1.07	6.36% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.84	8.41	1.05	4.85% faster
performance.compilation	180.36	183.06	0.99	-1.5% slower
Mean result: 1.0678008347228487

Jenkins Console Log
Blue Ocean
Commit hash: 4f8838bd96b1d1590209d845f484e5867c6c54a0

Machine information No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 20.04.3 LTS Release: 20.04 Codename: focal

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 46 bits physical, 48 bits virtual
CPU(s): 80
On-line CPU(s) list: 0-79
Thread(s) per core: 2
Core(s) per socket: 20
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 85
Model name: Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz
Stepping: 4
CPU MHz: 2400.000
CPU max MHz: 3700.0000
CPU min MHz: 1000.0000
BogoMIPS: 4800.00
Virtualization: VT-x
L1d cache: 1.3 MiB
L1i cache: 1.3 MiB
L2 cache: 40 MiB
L3 cache: 55 MiB
NUMA node0 CPU(s): 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70,72,74,76,78
NUMA node1 CPU(s): 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71,73,75,77,79
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable
Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Retbleed: Mitigation; IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; IBRS, IBPB conditional, STIBP conditional, RSB filling, PBRSB-eIBRS Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti intel_ppin ssbd mba ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req pku ospke md_clear flush_l1d arch_capabilities

G++:
g++ (Ubuntu 9.4.0-1ubuntu1~20.04) 9.4.0
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Clang:
clang version 10.0.0-4ubuntu1
Target: x86_64-pc-linux-gnu
Thread model: posix
InstalledDir: /usr/bin

the math and tests look good to me! @SteveBronder I'll let you comment on the cpp style

Few small things but overall good!

Suggested change

      
              typename plain_type_t<Vec>::Scalar sum_w(0);
          
              value_type_t<Vec> sum_w(0);

Suggested change

      
              auto arena_y = to_arena(y);
          
              auto arena_y = to_arena(std::forward<T>(y));

Suggested change

      
            inline auto sum_to_zero_constrain(const T& y) {
          
            inline auto sum_to_zero_constrain(T&& y) {

Suggested change

      
            inline auto sum_to_zero_constrain(const T& y, scalar_type_t<T>& lp) {
          
              return sum_to_zero_constrain(y);
          
            inline auto sum_to_zero_constrain(T&& y, scalar_type_t<T>& lp) {
          
              return sum_to_zero_constrain(std::forward<T>(y));

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.35	0.32	1.12	10.56% faster
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	1.0	-0.06% slower
gp_regr/gen_gp_data.stan	0.04	0.03	1.4	28.79% faster
gp_regr/gp_regr.stan	0.1	0.09	1.06	5.65% faster
sir/sir.stan	69.13	68.98	1.0	0.22% faster
irt_2pl/irt_2pl.stan	4.15	4.05	1.03	2.6% faster
eight_schools/eight_schools.stan	0.06	0.06	1.04	3.94% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.26	0.24	1.08	7.3% faster
pkpd/one_comp_mm_elim_abs.stan	20.1	19.83	1.01	1.33% faster
garch/garch.stan	0.44	0.41	1.07	6.6% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.81	2.6	1.08	7.48% faster
arK/arK.stan	1.85	1.71	1.08	7.5% faster
gp_pois_regr/gp_pois_regr.stan	2.94	2.73	1.08	7.12% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.8	8.41	1.05	4.44% faster
performance.compilation	184.88	181.73	1.02	1.71% faster
Mean result: 1.0744503895418258

Jenkins Console Log
Blue Ocean
Commit hash: 4f8838bd96b1d1590209d845f484e5867c6c54a0

Machine information No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 20.04.3 LTS Release: 20.04 Codename: focal

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 46 bits physical, 48 bits virtual
CPU(s): 80
On-line CPU(s) list: 0-79
Thread(s) per core: 2
Core(s) per socket: 20
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 85
Model name: Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz
Stepping: 4
CPU MHz: 3531.326
CPU max MHz: 3700.0000
CPU min MHz: 1000.0000
BogoMIPS: 4800.00
Virtualization: VT-x
L1d cache: 1.3 MiB
L1i cache: 1.3 MiB
L2 cache: 40 MiB
L3 cache: 55 MiB
NUMA node0 CPU(s): 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70,72,74,76,78
NUMA node1 CPU(s): 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71,73,75,77,79
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable
Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Retbleed: Mitigation; IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; IBRS, IBPB conditional, STIBP conditional, RSB filling, PBRSB-eIBRS Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti intel_ppin ssbd mba ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req pku ospke md_clear flush_l1d arch_capabilities

G++:
g++ (Ubuntu 9.4.0-1ubuntu1~20.04) 9.4.0
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Clang:
clang version 10.0.0-4ubuntu1
Target: x86_64-pc-linux-gnu
Thread model: posix
InstalledDir: /usr/bin

WardBrian and others added 18 commits July 30, 2024 12:15

Add sum_to_zero constraint, free, and check

8330f93

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

4564935

Header check fix

c75a793

Try rev overload

69ffa68

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

fb97a24

fix bug in reverse pass for sum to zero constraint

3fb6ff4

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

40d0e4b

Changes per Steve's comments

cf9b012

inv ilr constrain prim impl

ed4a64d

Single loop

aa41da4

off-by-1 fixes

ba179cb

Names

4197bfd

Free impl

004c7fe

Basic rev impl

e132643

Loop version

9cf06f1

Clarifying

86fed93

Docstrings

25460e9

Comment fixes

edd826e

WardBrian requested review from SteveBronder, bob-carpenter and spinkney August 5, 2024 13:36

Header fixes

8be67c6

spinkney reviewed Aug 5, 2024

View reviewed changes

Comment thread stan/math/prim/constraint/sum_to_zero_constrain.hpp

WardBrian mentioned this pull request Aug 5, 2024

Add sum_to_zero transform stan-dev/stanc3#1443

Merged

2 tasks

WardBrian and others added 2 commits August 5, 2024 14:25

Second citation from @spinkney

c64bace

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

51ac4c6

SteveBronder requested changes Aug 6, 2024

View reviewed changes

Changes per @SteveBronder

5ae902d

SteveBronder approved these changes Aug 7, 2024

View reviewed changes

WardBrian merged commit 9b1f08b into develop Aug 7, 2024

WardBrian deleted the sum-to-zero-constraint-ilr branch August 7, 2024 13:22

+               * @return Zero-sum vector of dimensionality K.
+               */
+              template <typename T, require_rev_col_vector_t<T>* = nullptr>
+              inline auto sum_to_zero_constrain(const T& y) {

Add sum_to_zero constraint, free, and check based on ILR transform by WardBrian · Pull Request #3101 · stan-dev/math · GitHub

Uh oh!

Uh oh!

Add sum_to_zero constraint, free, and check based on ILR transform - #3101

Add sum_to_zero constraint, free, and check based on ILR transform#3101
WardBrian merged 22 commits into
developfrom
sum-to-zero-constraint-ilr

WardBrian commented Aug 5, 2024

Uh oh!

Uh oh!

stan-buildbot commented Aug 6, 2024

Uh oh!

spinkney commented Aug 6, 2024 •
edited
Loading

Uh oh!

Uh oh!

SteveBronder left a comment

Uh oh!

Uh oh!

SteveBronder Aug 6, 2024

Uh oh!

SteveBronder Aug 6, 2024

Uh oh!

SteveBronder Aug 6, 2024

Uh oh!

SteveBronder Aug 6, 2024

Uh oh!

stan-buildbot commented Aug 7, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants


	typename plain_type_t<Vec>::Scalar sum_w(0);
	value_type_t<Vec> sum_w(0);

+                if (unlikely(y.size() == 0)) {
+                  return arena_t<ret_type>(Eigen::VectorXd{{0}});
+                }
+                auto arena_y = to_arena(y);

	auto arena_y = to_arena(y);
	auto arena_y = to_arena(std::forward<T>(y));

	inline auto sum_to_zero_constrain(const T& y) {
	inline auto sum_to_zero_constrain(T&& y) {

		inline auto sum_to_zero_constrain(const T& y, scalar_type_t<T>& lp) {
		return sum_to_zero_constrain(y);

Add sum_to_zero constraint, free, and check based on ILR transform by WardBrian · Pull Request #3101 · stan-dev/math · GitHub

Uh oh!

Uh oh!

Add sum_to_zero constraint, free, and check based on ILR transform - #3101

Add sum_to_zero constraint, free, and check based on ILR transform#3101WardBrian merged 22 commits intodevelopstan-dev/math:developfrom sum-to-zero-constraint-ilrstan-dev/math:sum-to-zero-constraint-ilrCopy head branch name to clipboard

Conversation

WardBrian commented Aug 5, 2024

Summary

Tests

Side Effects

Release notes

Checklist

Uh oh!

Uh oh!

stan-buildbot commented Aug 6, 2024

Uh oh!

spinkney commented Aug 6, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

SteveBronder left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

SteveBronder Aug 6, 2024

Choose a reason for hiding this comment

Uh oh!

SteveBronder Aug 6, 2024

Choose a reason for hiding this comment

Uh oh!

SteveBronder Aug 6, 2024

Choose a reason for hiding this comment

Uh oh!

SteveBronder Aug 6, 2024

Choose a reason for hiding this comment

Uh oh!

stan-buildbot commented Aug 7, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Add sum_to_zero constraint, free, and check based on ILR transform#3101
WardBrian merged 22 commits into
developfrom
sum-to-zero-constraint-ilr

spinkney commented Aug 6, 2024 •
edited
Loading