Incorrect output from torch.histc when min and max are equal.

[Manas-Tiwari]

🐛 Describe the bug

t = torch.tensor([1,2,3,3,4,5,6,7,8], dtype=torch.float)
torch.histc(t, 5, 7, 7)

The output given by above code:

tensor([2., 2., 2., 1., 2.])

According to the documentation for torch.histc, it is expected that values smaller than min and greater than max should be ignored. However, the output includes values outside the specified range. This behavior contradicts the documentation. Therefore, there might be a bug in the implementation.

Versions

Collecting environment information...
PyTorch version: 2.2.1+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.3 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: 14.0.0-1ubuntu1.1
CMake version: version 3.27.9
Libc version: glibc-2.35

Python version: 3.10.12 (main, Nov 20 2023, 15:14:05) [GCC 11.4.0] (64-bit runtime)
Python platform: Linux-6.1.58+-x86_64-with-glibc2.35
Is CUDA available: False
CUDA runtime version: 12.2.140
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: Could not collect
Nvidia driver version: Could not collect
cuDNN version: Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.9.6
/usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.9.6
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 46 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 2
On-line CPU(s) list: 0,1
Vendor ID: GenuineIntel
Model name: Intel(R) Xeon(R) CPU @ 2.20GHz
CPU family: 6
Model: 79
Thread(s) per core: 2
Core(s) per socket: 1
Socket(s): 1
Stepping: 0
BogoMIPS: 4400.42
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid tsc_known_freq pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch invpcid_single ssbd ibrs ibpb stibp fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm rdseed adx smap xsaveopt arat md_clear arch_capabilities
Hypervisor vendor: KVM
Virtualization type: full
L1d cache: 32 KiB (1 instance)
L1i cache: 32 KiB (1 instance)
L2 cache: 256 KiB (1 instance)
L3 cache: 55 MiB (1 instance)
NUMA node(s): 1
NUMA node0 CPU(s): 0,1
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Mitigation; PTE Inversion
Vulnerability Mds: Vulnerable; SMT Host state unknown
Vulnerability Meltdown: Vulnerable
Vulnerability Mmio stale data: Vulnerable
Vulnerability Retbleed: Vulnerable
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Vulnerable
Vulnerability Spectre v1: Vulnerable: __user pointer sanitization and usercopy barriers only; no swapgs barriers
Vulnerability Spectre v2: Vulnerable, IBPB: disabled, STIBP: disabled, PBRSB-eIBRS: Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Vulnerable

Versions of relevant libraries:
[pip3] numpy==1.25.2
[pip3] torch==2.2.1+cu121
[pip3] torchaudio==2.2.1+cu121
[pip3] torchdata==0.7.1
[pip3] torchsummary==1.5.1
[pip3] torchtext==0.17.1
[pip3] torchvision==0.17.1+cu121
[pip3] triton==2.2.0
[conda] Could not collect

cc @albanD

[shink]

Thanks for your bug report! The result of this case on pytorch2.0 is the same as above, so what's the correct answer to this case?

Python 3.11.9 (main, Apr 19 2024, 16:48:06) [GCC 11.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> t = torch.tensor([1,2,3,3,4,5,6,7,8], dtype=torch.float)
>>> torch.histc(t, 5, 7, 7)
tensor([2., 2., 2., 1., 2.])
>>> print(torch.__version__)
2.0.0+cpu

[Manas-Tiwari]

The documentation mentions that values out of min and max range are ignored. So in this case, all values except 7 should be ignored.
I think output in this case should be

tensor([0., 0., 1., 0., 0.])

which is same as in the following case

t = torch.tensor([7], dtype=torch.float)
torch.histc(t, 5, 7, 7)

[shink]

@Manas-Tiwari Yes, you are right, I agree. However, looking at the code below, min and max will be updated when they are equal.

pytorch/aten/src/ATen/native/Histogram.cpp

Lines 207 to 221 in 9dec41b

	std::pair<double, double> histc_select_outer_bin_edges(const Tensor& input,
	const Scalar& min, const Scalar& max) {
	double leftmost_edge = min.to<double>();
	double rightmost_edge = max.to<double>();
	if (leftmost_edge == rightmost_edge && input.numel() > 0) {
	auto extrema = aminmax(input);
	leftmost_edge = std::get<0>(extrema).item<double>();
	rightmost_edge = std::get<1>(extrema).item<double>();
	}
	if (leftmost_edge == rightmost_edge) {
	leftmost_edge -= 1;
	rightmost_edge += 1;
	}

The results are the same in both cases:

>>> import torch
>>> t = torch.tensor([1,2,3,3,4,5,6,7,8], dtype=torch.float)
>>> torch.histc(t, 5, 7, 7) # min=7, max=7
tensor([2., 2., 2., 1., 2.])
>>> torch.histc(t, 5, 1, 8) # min=1, max=8
tensor([2., 2., 2., 1., 2.])

So should we update the code or the doc?

[Manas-Tiwari]

I think the code should be updated. min and max values should only be updated when they both are equal to zero.