+
+Eight out of ten survey participants are interested in contributing to open source software projects. Yet, confusion about where to start and lack of experience and time are often preventing them from making the leap.
+
+### Contributing to NumPy
+#### Have you ever contributed to NumPy?
+
+
+#### Are you a regular contributor to NumPy?
+
+
+12% of survey participants have contributed to NumPy and one in three of those contributes regularly.
+
+#### Have you ever contributed to any open source project before, excluding NumPy?
+
+
+#### Are you a regular contributor to these open source projects?
+
+
+When it comes to other open source projects, every second survey participant has contributed at least once, mainly to the projects within the Python ecosystem, and one in four of them contributes to these projects on a regular basis.
+
+#### Which project(s) have you contributed to? Please select all that apply.
+
+
+#### What initially led to your contributing to NumPy? Please select all that apply.
+
+
+#### How did you initially get involved with these (non-NumPy) projects? Please select all that apply.
+
+Curious what initially led these NumPy users to contributing to open source? Half of them identify as experts in NumPy. A similar pattern is found in the
+wider open source community.
+
+#### In what capacity have you contributed to NumPy? Please select all that apply.
+
+
+#### In what way(s) have you contributed to these projects. Please select all that apply.
+
+Half of NumPy contributors who participated in this survey have contributed to the project source code; nearly as many, responded to GitHub issues and contributed to the technical documentation.
+While 83% of contributors to other open source projects contributed to the codebase, 46% responded to GitHub issues, and 41% pitched in to help with documentation.
+In both instances, one in five respondents contributed to educational content and helped with community coordination.
+
+#### What are your motivations for contributing to NumPy? Please select all that apply.
+
+(images\report_images_2021\Community_Contribution_to_Open_Source\4.5_motivations_contributing_numpy.svg)
+Giving back to the scientific community, promoting open source culture, and advancing scientific research are the top three reasons for respondents to contribute to NumPy.
+
+#### Are you interested in contributing to NumPy?
+
+
+#### Do you plan to continue making contributions to NumPy?
+
+
+93% of casual NumPy contributors and 100% of those who identify themselves as regulars plan to continue making contributions to the project.
+
+#### In what ways would you be interested in contributing to NumPy? Please select all that apply.
+
+Out of those who have never contributed to NumPy, 60% like the idea of giving back to the NumPy community. Two thirds of these potential contributors expressed their interest in contributing to the source code, nearly a half would like to help with educational content, and one in three are interested in triaging issues and writing technical documentation.
+
+## Community — Demographics
+### Age Range
+#### What is your age?
+
+Millennials comprise 66% of respondents.
+
+### Gender
+#### What is your gender?
+
+85% of survey participants identify as males.
+
+### Languages
+#### What language do you prefer to use?
+
+Over 70% of NumPy users choose English as the preferred language of communication.
+
+### Location
+#### What is your country of residence?
+
+| Percentage | Country |
+|--------------|--------------------------------------------------------------------------------------------------------|
+| 26% | United States |
+| 8% | Germany |
+| 7% | France |
+| 5% | India, Japan, United Kingdom |
+| 3% | Brazil, Czech Republic, Netherlands |
+| 2% | Spain, China, Australia, Austria Canada, Poland |
+| 1% | Argentina, Peru, Russia, Singapore, Switzerland, Turkey, Belgium, Chile, Finland, Israel, Italy, Sweden |
+
+NumPy is a true global open source project. In fact, almost three quarters of survey participants come from outside of the United States. In total, we reached NumPy users living in 54 different countries.
+
+## Development, Features and Deprecation
+NumPy is committed to stability in order to provide a foundation for other packages within the scientific Python ecosystem and beyond. This commitment plays a major role in design decisions as well as the pace of introduction/deprecation of features.
+
+### Frequency
+#### How often do you use NumPy?
+
+Nearly half of respondents use it on a daily basis; 83%, at least once a week.
+
+### Primary Use
+#### How do you primarily use NumPy?
+
+Three quarters of survey participants use NumPy for work.
+
+### Sharing Code
+#### How many people do you typically share your written code with?
+
+One in three respondents shares their code with only 1-2 people; one in four, with teams of more than 10 people.
+
+### NumPy Version
+#### What version of NumPy are you currently using?
+
+27% of survey participants use NumPy version 1.18; 22% use multiple versions on a regular basis.
+
+### NumPy Components
+#### Which components of NumPy do you use regularly? Please select all that apply.
+
+Random number generator and linear algebra are the most used NumPy sub-packages.
+
+### Custom C Extensions
+#### Do you or your organization have custom C extensions making use of the NumPy C-API (aside from Cython)?
+
+
+11% have custom C extensions making use of the NumPy C-API.
+
+### New Random API
+#### NumPy version 1.17 introduced a new API to NumPy’s random package. Are you currently using the new random API?
+
+
+41% of respondents use the new random API.
+
+### Challenges and Solutions
+#### In the last year, have you experienced problems in code you’ve written stemming from a problem in NumPy?
+
+
+#### What action(s) did you take in response to this issue? Please select all that apply.
+
+In the last year, 9% of NumPy users experienced problems in code they’ve written stemming from a problem in NumPy. What did they do about it? 76% found a workaround for it, 14% opened an issue in the NumPy repository on GitHub, and 6% proposed a change to NumPy by submitting a pull request.
+
+#### In the last year, have you encountered a problem involving numerical data that you were unable to solve using NumPy?
+
+
+#### What action(s) did you take in response to this issue? Please select all that apply.
+
+During the same period of time, 18% of respondents encountered a problem involving numerical data that they were unable to solve using NumPy. Nearly 90% of respondents either found a workaround for it or switched to a different library, 3% sought help on the mailing list, and 2% opened an issue in the NumPy repository on GitHub.
+
+#### NumPy encourages users to open issues. Please describe why you decided not to open an issue for your particular problem.
+
+NumPy encourages users to open issues. Yet, last year, only 8% of respondents on average resorted to this option when they encountered a problem that might have been stemming from NumPy. One in ten admitted that they simply didn’t know how to file an issue with NumPy. 31% of respondents decided not to file an issue due to doubts that their problem was stemming from NumPy. The same percentage of users chose to use a workaround instead.
+
+### Deprecation Time Frame
+#### NumPy normally has a two release cycle (1 year) deprecation policy. Do you think this is…
+
+
+#### What do you consider as a good deprecation time frame?
+
+The NumPy maintainers were eager to know what a good deprecation time frame was according to NumPy users. About one third of survey participants believes it should depend on the feature, while another third suggests for it to be 2 release cycles. Overall, the current NumPy deprecation policy of two release cycles (1 year) is supported by over 60% of respondents.
+
+## Future Directions
+### Project Priorities
+#### If NumPy had one extra full-time developer, what would you like them to focus on? Please drag and drop the following items in order of priority with 1 being highest priority.
+
+Once again NumPy users put performance and reliability at the top of the priority list for the project.
+
+#### Performance
+The overall sentiment in the comments is that NumPy performance is already good, but further improvements are possible and should be put high on the project’s priority list.
+Parallel execution, GPU support, and improved interoperability with other packages, both via the NumPy array protocols and directly with CuPy, Dask, JAX, Numba, Cython, Pythran, PyPy, and deep learning frameworks were mentioned most often. Also, quite a few respondents suggested making better use of modern hardware capabilities like SIMD instructions and simplifying the back-end switches for, in particular, BLAS.
+
+#### Reliability
+NumPy users are almost unanimous that NumPy is very reliable already and it’s essential that it remains this way, including while introducing new features and if the speed of development were to increase in the future. Minor points identified for improvement include the clarity of error messages and dealing with inconsistencies and quirks in the parts of NumPy that are used less often.
+
+### The Big Picture
+#### A few final questions for any additional thoughts/opinions about NumPy that you’d like to share. What single immediate change to NumPy would bring the most value to you as a NumPy user?
+- Support more modern Fortran features in f2py.
+- I would like to see events where we call open source researchers to implement their papers to numpy core, like a competition who has the fastest fft.
+- I've really appreciated the renewed willingness of the numpy development community to accept larger changes. Kudos.
+- -Integration/consolidation of NumPy/Numba/CuPy/Xarray into a single framework for JIT compiling for CPU and GPU.
+- Documentation with more examples to better understand the context in which to use certain functionality. (The comment was submitted in Spanish.)
+- Translations of documentation to reach more people. (The comment was submitted in Spanish.)
+- A numpy discourse to help when asking questions.
+- Better build system, also some documentation on how to use it (especially for building FORTRAN extensions).
+
+In an open-ended question, we asked: "What single immediate change to NumPy would bring the most value to you as a NumPy user?” Just like in 2020, there was no lack of creative ideas and opinions on solving current pain points in NumPy and downstream, from introducing a unified array API to OpenBLAS optimization and better support for typing. However, documentation received by far the most attention.
+
+#### Documentation
+The majority of respondents are happy with the quality of the NumPy documentation. Still, a number of suggestions were submitted for further improvement. "Examples, more examples, and more detailed examples" is the recurrent theme in about half the comments. Translations of the documentation and educational materials into other languages are also high on NumPy users' wish lists. A common concern about the NumPy documentation is that NumPy is very large, and it's often not easy to navigate the documentation and find the functionality that best fits a user's need.
+
+### Other Significant Changes
+#### What other significant changes would you like to see in NumPy?
+When asked about what other significant changes our users would like to see in NumPy, GPU support and improved F2PY were mentioned as often as documentation.
+
+#### GPU support
+Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.
+
+#### F2PY
+A few survey participants have expressed interest in the inclusion of modern Fortran features to F2PY, as well as improved documentation including usage examples. In particular, respondents mentioned derived types support, which is currently being implemented, and some maintenance improvements that would make contributing to the codebase easier. The spontaneous mentions to F2PY in the survey show that many projects rely on it for efficient interoperability between C/Fortran and Python.
+
+#### Documentation
+The comments about the NumPy documentation in response to this question largely overlap with the responses about a single immediate change for NumPy (refer to the summary regarding the documentation in the section above).
+
+### Closing Questions
+#### Please tell us how you heard about this survey.
+
+Nearly half of respondents heard about the survey on Twitter, and one in four found out about it through the announcement on the numpy.org home page.
+
+#### Did you respond to last year's NumPy Community Survey (2020)?
+
+
+Only 13% of the 2021 respondents participated in the previous year’s survey.
+
+## Acknowledgements
+The 2021 NumPy user survey is an effort of an international group of volunteers. Each is listed alphabetically by type of contribution.
+
+### Survey Design and Analysis
+- Pantelis Antonoudiou
+- Ross Barnowski
+- Sebastian Berg
+- Ralf Gommers
+- Rohit Goswami
+- Melissa Mendonça
+- Stephanie Mendoza
+- Aerik Pawson
+- Inessa Pawson
+
+### Translations
+- Xiaoyi Deng
+- Kiko Correoso Garcia
+- Paul Ivanov
+- Guilherme Leobas
+- Mahfuza Humayra Mohona
+- Aerik Pawson
+- Zijie Poh
+- Atsushi Sakai
+- Fabrice Silva
+- Kriti Singh
+
+### Special thanks
+- Mars Lee
+- The Berkeley Institute for Data Science (UC Berkeley)
+
+## Methodology
+Over 500 NumPy users from 54 countries participated in this online survey, which was fielded from July 8th to September 6th, 2021. Respondents were recruited via the NumPy mailing list, the NumPy Slack channel, the banner display on the numpy.org homepage, the NumFOCUS newsletter, announcements at the SciPy’21 conference, and via Twitter.
+
+To engage non-English speaking stakeholders, we conducted the survey in 8 additional languages: Bangla, French, Hindi, Japanese, Mandarin, Portuguese, Russian, and Spanish.
+
+The 2021 user survey questionnaire was largely adapted from the 2020 version. Some totals don’t equal 100% due to multiple-choice answers or rounding.
diff --git a/images/report_images_2021/Community/2.13_years_of_programming.svg b/images/report_images_2021/Community/2.13_years_of_programming.svg
new file mode 100644
index 0000000..8794354
--- /dev/null
+++ b/images/report_images_2021/Community/2.13_years_of_programming.svg
@@ -0,0 +1,584 @@
+
+
+
diff --git a/images/report_images_2021/Community/2.14_years_of_numpy.svg b/images/report_images_2021/Community/2.14_years_of_numpy.svg
new file mode 100644
index 0000000..79f5495
--- /dev/null
+++ b/images/report_images_2021/Community/2.14_years_of_numpy.svg
@@ -0,0 +1,590 @@
+
+
+
diff --git a/images/report_images_2021/Community/2.18_programming_languages.svg b/images/report_images_2021/Community/2.18_programming_languages.svg
new file mode 100644
index 0000000..603a11a
--- /dev/null
+++ b/images/report_images_2021/Community/2.18_programming_languages.svg
@@ -0,0 +1,732 @@
+
+
+
diff --git a/images/report_images_2021/Community/2.7_education.svg b/images/report_images_2021/Community/2.7_education.svg
new file mode 100644
index 0000000..e9645d3
--- /dev/null
+++ b/images/report_images_2021/Community/2.7_education.svg
@@ -0,0 +1,1347 @@
+
+
+
diff --git a/images/report_images_2021/Community/2.8_current_field.svg b/images/report_images_2021/Community/2.8_current_field.svg
new file mode 100644
index 0000000..b91190a
--- /dev/null
+++ b/images/report_images_2021/Community/2.8_current_field.svg
@@ -0,0 +1,1305 @@
+
+
+
diff --git a/images/report_images_2021/Community/2.9_current_role.svg b/images/report_images_2021/Community/2.9_current_role.svg
new file mode 100644
index 0000000..4e149e2
--- /dev/null
+++ b/images/report_images_2021/Community/2.9_current_role.svg
@@ -0,0 +1,1962 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/3.2_contributed_before.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/3.2_contributed_before.svg
new file mode 100644
index 0000000..1e4ba1b
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/3.2_contributed_before.svg
@@ -0,0 +1,122 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/3.3_projects_contributed_to.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/3.3_projects_contributed_to.svg
new file mode 100644
index 0000000..1c6175a
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/3.3_projects_contributed_to.svg
@@ -0,0 +1,1456 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/3.4_ways_contribute_projects.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/3.4_ways_contribute_projects.svg
new file mode 100644
index 0000000..aa2c842
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/3.4_ways_contribute_projects.svg
@@ -0,0 +1,2053 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/3.5_regular_contributor_to_numpy.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/3.5_regular_contributor_to_numpy.svg
new file mode 100644
index 0000000..9c0bcc7
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/3.5_regular_contributor_to_numpy.svg
@@ -0,0 +1,124 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/3.6_initially_contributing_to_open_source.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/3.6_initially_contributing_to_open_source.svg
new file mode 100644
index 0000000..47f7cda
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/3.6_initially_contributing_to_open_source.svg
@@ -0,0 +1,1664 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/3.7_interest_in_contributing.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/3.7_interest_in_contributing.svg
new file mode 100644
index 0000000..b3164c1
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/3.7_interest_in_contributing.svg
@@ -0,0 +1,126 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.1_ever_contribute_to_numpy.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.1_ever_contribute_to_numpy.svg
new file mode 100644
index 0000000..973f939
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.1_ever_contribute_to_numpy.svg
@@ -0,0 +1,129 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.2_ways_contribute_to_numpy.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.2_ways_contribute_to_numpy.svg
new file mode 100644
index 0000000..85a1167
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.2_ways_contribute_to_numpy.svg
@@ -0,0 +1,2050 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.3_regular_contributor.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.3_regular_contributor.svg
new file mode 100644
index 0000000..58d5b49
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.3_regular_contributor.svg
@@ -0,0 +1,127 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.4_initially_contributing_to_numpy.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.4_initially_contributing_to_numpy.svg
new file mode 100644
index 0000000..8df89e6
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.4_initially_contributing_to_numpy.svg
@@ -0,0 +1,1667 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.5_motivations_contributing_numpy.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.5_motivations_contributing_numpy.svg
new file mode 100644
index 0000000..025d14c
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.5_motivations_contributing_numpy.svg
@@ -0,0 +1,1306 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.6_plan_to_continue.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.6_plan_to_continue.svg
new file mode 100644
index 0000000..4af0f7d
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.6_plan_to_continue.svg
@@ -0,0 +1,114 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.8_interested_in_contributing_numpy.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.8_interested_in_contributing_numpy.svg
new file mode 100644
index 0000000..a14ba8b
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.8_interested_in_contributing_numpy.svg
@@ -0,0 +1,116 @@
+
+
+
diff --git a/images/report_images_2021/Community_Contribution_to_Open_Source/4.9_ways_interested_contributing_numpy.svg b/images/report_images_2021/Community_Contribution_to_Open_Source/4.9_ways_interested_contributing_numpy.svg
new file mode 100644
index 0000000..76039d5
--- /dev/null
+++ b/images/report_images_2021/Community_Contribution_to_Open_Source/4.9_ways_interested_contributing_numpy.svg
@@ -0,0 +1,1999 @@
+
+
+
diff --git a/images/report_images_2021/Community_Demographics/2.2_age.svg b/images/report_images_2021/Community_Demographics/2.2_age.svg
new file mode 100644
index 0000000..653ec56
--- /dev/null
+++ b/images/report_images_2021/Community_Demographics/2.2_age.svg
@@ -0,0 +1,661 @@
+
+
+
diff --git a/images/report_images_2021/Community_Demographics/2.3_gender.svg b/images/report_images_2021/Community_Demographics/2.3_gender.svg
new file mode 100644
index 0000000..f44d3c2
--- /dev/null
+++ b/images/report_images_2021/Community_Demographics/2.3_gender.svg
@@ -0,0 +1,412 @@
+
+
+
diff --git a/images/report_images_2021/Community_Demographics/2.4_language.svg b/images/report_images_2021/Community_Demographics/2.4_language.svg
new file mode 100644
index 0000000..c974313
--- /dev/null
+++ b/images/report_images_2021/Community_Demographics/2.4_language.svg
@@ -0,0 +1,671 @@
+
+
+
diff --git a/images/report_images_2021/Community_Demographics/world_map.svg b/images/report_images_2021/Community_Demographics/world_map.svg
new file mode 100644
index 0000000..e47f4fa
--- /dev/null
+++ b/images/report_images_2021/Community_Demographics/world_map.svg
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/images/report_images_2021/Community_Demographics/world_map_spread.ai b/images/report_images_2021/Community_Demographics/world_map_spread.ai
new file mode 100644
index 0000000..8a17ab0
--- /dev/null
+++ b/images/report_images_2021/Community_Demographics/world_map_spread.ai
@@ -0,0 +1,6922 @@
+%PDF-1.6
%
+1 0 obj
<>/OCGs[30 0 R 31 0 R]>>/Pages 3 0 R/Type/Catalog>>
endobj
2 0 obj
<>stream
+
+;>P[! KHSP)"DѫuϠж @߷FUtba>d]ASYIS #OPGk^{8[R}1TLv4[m8o9 G&,reo"0< QVk)c [` ?Հ joh ^f.e%4@ NhYx*h[y]S k`(' Er:"ɅȯДX)U7ˮ",{
+ƍˌa gS 3` }މiTL{y! 6ѮI>SȥpD
e奱!ʠ2
+ Д
}3
+
#QEJTmfVR;UlE E2.ȃBxo;3a$k^?y%eDY E2f [vA<5OegqUت? j[\IaF_A,*sBEߗxka?rz:dsࡔsEj(2^[Kȅ!5h_)3'tj
q
g]Ij <
n 5y zT>1Tg&ԑlg>LEJ;qGJ Q =\A[хRDIs@N -Z K+S 6ZZ; ڹS0 u뇲ڀE0fx~v)sq)Bk aC^z%HPNaښͯ)TqD"C˩~OÓaR;
+eRRL6t
z'p@ia3Nf$u !rnW"*PNSqNIy5ཤQ2?32Rn "J39dV(I*ܴ쒰y8&Gay(YAutl飙8?=Z)BKa+pO.zE;x+)%#uCdbƬ-ˎA= p;Om`A5
Yxb >=.t/"Mvn7'9=k7f{𡁄g8e^^kiSK
sKΜ0Yj)Q slV b۵c $j"%\:2X!669WAQU %KL}#[>lk oED ȁʀͩ^4q^E>!D@Y11ZblXkf
q'}y˘m4 OW(S:%D
/Xצ;2D7ԉWIXƶ'"(nH63;l4DŽ 4.`}83^Z[gvDGz;ɅpN-Ą́DkyW%3}^Nf-8D}cZDe=(\9#6n M&E#?fCy 5Y^;RGɩ\fg8%:UX}"ʉNo R16麸C% e
+ ;KIQ}nI?i\=%̧OohV25̕)cqP,d #[="đ YY- Do^E$%'
Jxę9~ J :j}Vi=n
-̖]߃8PI(tKNKڏ%[Rʦc)/%`չnta?bavo (hYn<}z%J4F95%$e
iLm F"4 DɾK^CѨ3u cy $t m]{3݆@R}XlX0pƑq%|ǰ8=y!E$ 39vW1
Cd_ K{bT.ʺvqɇ,n |ɋY$|*
f"
AZbBX R*"3v 6Ș%cn=sT
ɸto+i[(>W(L*,^yE!NU8Ӊ|t% Ӹ J !1Y ue~BWߵq.^MiS2Yѥyz-`E@ ]h,ߟ JH~"^2m~ ^tåyTe#+aͫ?C1w1g5pʁz]Q\Ɉ9qGQ*O$ "< \@y6JLkf *L aF{ vJ"zNQ(f_X?&둦
+q6UB^*TsnJU.< ]lu9G92cUS@#,0Hg5v"1"®^WDjGI9J:#V}.?fj~ƩO
+tO)mRɇ~Zk
G\JעbAW\ኾuy~3N\<{Z#'NqLQt@SXg'5r 0귧Kk0=\4Ht$|*X)r/?\"EԜ2o
OR]ǿUU֕%nzSRWY8.n9!
,Y+j*#b,t˖v$[.C9Jݸc^"h}28,jyr)ga{^~b4+[|IIUn
-z^Mѳ%r<] ;GtH-~Ln_ye +eM*;rzUv\ '5< J Sp翴3%2Q4Z; Ym&ܝTm2,E尪'
+[dM\*hfm[ڳ^ZuiS\w.{^s˟_R:24^igytf~f;.nD(VTK(*5K&(4Hi} '!JAqg0Ap!✾P~$mJ^ηK b0U4$Oږ0v` e 2o4E_*h/KĒo-]8(D͟T^ij&R\)kΉXP j0LCA왼bSk}IȰ6*Q:)aD_ƅHCb]q6'^863K˂8C# ɟX %:z\Cgdz+wE|䋆,XB 5'SXb@dô )yj*^iVI=YhkH< Iㆈ
ٿ@Ӡ8JA^ YiSJqmo ASi(!ZY/],
U>dy~"LYwrwI_o$5f89 Valt`&^=Cƈa "h#},d{tծ]V4jQ
+hNi'4Ymb Kb `+MŋϮT*+ށr G)- h $v`& EK9Tl석<("_`?g$DH!N}@>.xl&ۗ7_ Wh%8wj G8+ۤsb -UD
uBְ;:ԅTN.lV;yڱWEF!#]SmjV1gz Wgj] za9YyM!cZ 1$Ƽ̏˴ :(zټDn[ӚִK>-)CO\W_<~"\,q6˭cR AauÞ+_h^E4Zs5m,y9v;ǾxxLv@~k1 M|H8=g ? ؕOT@+5y?A{>=ߞ@
+I`xUJNw:4T *b* eo- ߷i^nmh0'^b`RhRJZQsblwP̽;xqTT[Q:yf3OR7Vd!q"j*uYkXj=+@sТlݎ[ HM4E"DjDhjȓΊ$-Srs *ރ0B@Vz]~6
=UIH|ӼѼ~R̲sh2S}w0:)ʨC-_q nvHcd ~r |/7Pjo=1)7 `~~YJ
HknpO+W*
+? 3P$K_Q:.
WWSv*"7j$I=dדY&Ѹd$q/fD`U=ؾHMF<py0) 2 C!4坃OEbR"S`]u#EN)Y1N >e^&`}M+AiLS嗇kj
!tQ3;$Aytp&