CMBLensed input PS shifts multipoles by 2 when using new CAMB-produced power spectra

[JamesAmato]

The CMBLensed code appears to be misaligning power spectra. When it calculates the intermediate cl_tebp_arr from a power spectrum produced using CAMB, it has 0's for $\ell \in {0, 1, 2, 3}$. I believe it should only have zero values for $\ell \in {0,1}$.

I looked into this further and believe I've found the cause of the issue and propose a solution. I have another issue involving apply_delens that is tangled with this; I'll need more feedback on that one Issue 213.

Cause:
I think the code was written with the assumption that rows for $\ell=0$ and $\ell=1$ will not occur in CAMB's output text file. In the input file used with c1, the first multipole is $\ell=2$. CAMB's documentation states that output power spectra will have $\ell=0$ and $\ell=1$ rows. Values in cl_tebp_arr are assigned beginning at index 2, pulling values from the input data (cmb spectra) at index 0. Clearly this fails.

Solution:
I've attached a snippet ("snippet.txt") that can replace lines 290-338 of pysm/src/pysm3/models/cmb.py. I can submit a pull request, but need to resolve some confusion on a related issue. I've also attached a demo of the snippet as an alternative ("CMBLensed_l0l1_min_repro_demo.txt"); it will need two power spectra - one with rows for $\ell=0$ and $\ell=1$ and one without.

snippet.txt
CMBLensed_l0l1_min_repro_demo.txt

[zonca]

Please provide fully executable example, possibly In a notebook which contains plots

…

On Tue, Feb 25, 2025, 16:40 JamesAmato ***@***.***> wrote: The CMBLensed code appears to be misaligning power spectra. When it calculates the intermediate cl_tebp_arr from a power spectrum produced using CAMB, it has 0's for $\ell \in {0, 1, 2, 3}$. I believe it should only have zero values for $\ell \in {0,1}$. I looked into this further and believe I've found the cause of the issue and propose a solution. I have another issue involving apply_delens that is tangled with this; I'll need more feedback on that one. Cause: I think the code was written with the assumption that rows for $\ell=0$ and $\ell=1$ will not occur in CAMB's output text file. In the input file used with c1, the first multipole is $\ell=2$. CAMB's documentation <https://camb.readthedocs.io/en/latest/results.html#camb.results.CAMBdata.get_cmb_power_spectra> states that output power spectra will have $\ell=0$ and $\ell=1$ rows. Values in cl_tebp_arr are assigned beginning at index 2, pulling values from the input data (cmb spectra) at index 0. Clearly this fails. Solution: I've attached a snippet ("snippet.txt") that can replace lines 290-338 of pysm/src/pysm3/models/cmb.py. I can submit a pull request, but need to resolve some confusion on a related issue. I've also attached a demo of the snippet as an alternative ("CMBLensed_l0l1_min_repro_demo.txt"); it will need two power spectra - one with rows for $\ell=0$ and $\ell=1$ and one without. snippet.txt <https://github.com/user-attachments/files/18974729/snippet.txt> CMBLensed_l0l1_min_repro_demo.txt <https://github.com/user-attachments/files/18974724/CMBLensed_l0l1_min_repro_demo.txt> — Reply to this email directly, view it on GitHub <#212>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAC5Q4VYTZDRUQUBMNMC2FL2RUERXAVCNFSM6AAAAABX365ZSOVHI2DSMVQWIX3LMV43ASLTON2WKOZSHA3TSOBUGY3DCMQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***> [image: JamesAmato]*JamesAmato* created an issue (galsci/pysm#212) <#212> The CMBLensed code appears to be misaligning power spectra. When it calculates the intermediate cl_tebp_arr from a power spectrum produced using CAMB, it has 0's for $\ell \in {0, 1, 2, 3}$. I believe it should only have zero values for $\ell \in {0,1}$. I looked into this further and believe I've found the cause of the issue and propose a solution. I have another issue involving apply_delens that is tangled with this; I'll need more feedback on that one. Cause: I think the code was written with the assumption that rows for $\ell=0$ and $\ell=1$ will not occur in CAMB's output text file. In the input file used with c1, the first multipole is $\ell=2$. CAMB's documentation <https://camb.readthedocs.io/en/latest/results.html#camb.results.CAMBdata.get_cmb_power_spectra> states that output power spectra will have $\ell=0$ and $\ell=1$ rows. Values in cl_tebp_arr are assigned beginning at index 2, pulling values from the input data (cmb spectra) at index 0. Clearly this fails. Solution: I've attached a snippet ("snippet.txt") that can replace lines 290-338 of pysm/src/pysm3/models/cmb.py. I can submit a pull request, but need to resolve some confusion on a related issue. I've also attached a demo of the snippet as an alternative ("CMBLensed_l0l1_min_repro_demo.txt"); it will need two power spectra - one with rows for $\ell=0$ and $\ell=1$ and one without. snippet.txt <https://github.com/user-attachments/files/18974729/snippet.txt> CMBLensed_l0l1_min_repro_demo.txt <https://github.com/user-attachments/files/18974724/CMBLensed_l0l1_min_repro_demo.txt> — Reply to this email directly, view it on GitHub <#212>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAC5Q4VYTZDRUQUBMNMC2FL2RUERXAVCNFSM6AAAAABX365ZSOVHI2DSMVQWIX3LMV43ASLTON2WKOZSHA3TSOBUGY3DCMQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[JamesAmato]

Here's a notebook, showing the misalignment, the impact on maps produced from CAMB output power spectra, and a proposed solution.

https://github.com/JamesAmato/pysm/blob/demo_issues/issue_212_demo.ipynb

[JamesAmato]

I found a far simpler approach, with two lines added in the __init__() method. There's no need to touch the taylens code. I apologize for muddying the water.

Here's an example.

When combined with Issue 213, a total of six lines need to be changed: two for 212, two for 213, and two for the intersection of them: Example.

[zonca]

ok, thanks, can you please make a pull request with those changes? Actually I prefer 2 separate pull requests each with 1 notebook that tests just that issue.

Also if you add a unit test in tests/test_cmb.py it is useful.